Change shared libs name by jayhawk-commits · Pull Request #3 · ROCm/rocm-libraries

jayhawk-commits · 2025-04-29T17:17:56Z

For testing Jenkins workflows with monorepo

* finished sort keys no size tests * completed sort pair no size functions * completed sort pair with size functions * updated changelog

…/rocrand.h (#3) * aded float4 out tests * added double4 tests * added double2 out tests * reformated numeric test to only use one structure * moved new tests to eof * started implementation on UD with states * added memory deallocation * updated to use only one thread * implemented rest of states UD test * updated changelog

* Enable gfx12 support [ROCm/hipRAND commit: 881a6bf]

Removes the cmake dependency that was added for hip. This was preventing cuda builds of hipBLAS from building. [ROCm/hipBLAS-common commit: eb8d0fa]

@timing

* feat: no global working path with asserts * fix: bad import statement * fix: different build_tmp dir in cmake than tensile * fix: incorrect pathing in Tensile benchmarking * fix: string wrap paths for assert checks * fix: ensure all paths are built * fix: missing global parameter warning * fix: ensure library logic path is built * fix: client writer paths * fix: use child path for client libraries * remove PrintTiming from global params * remove another assert * restore PrintTiming * fix: reviewer comments * refactor: use env var for @timing decorator * refactor: remove library print debug * refactor: remove exit after kernel gen option * style: remove commented code * refactor: move common to a module * feat: remove 'CustomKernelDirectory' global param * fix: amd clang version global param * fix: second pass on amd clang version * fix: imports for bolted on ops scripts * chore: remove safety asserts * chore: create more separation in Common * fix: missing line at eof * fix: basic reviewer comments * chore: move global dependenct funcs to GlobalParameters.py * fix: pass global params to splitArchs * style: don't format caps funcs * fix: bad import * Add TensileLogic program to verify matrix instructions (#3) * Remove unused and static global parameters (#5) * fix: purge library-print-debug * feat: make build paths static * feat: remove 'SortProblems' global param * feat: remove 'ExpandRanges' global param * feat: remove 'WavefrontWidth' global param * feat: remove 'ValidateLibrary' global param * feat: remove 'EnableHalf' and 'ClientArgs' global params * feat: prefer profile decorator over 'Profiler' global param * chore: remove 'LibraryPrintDebug' from build_client.yaml * feat: remove 'MaxFileName' from global params * Update preferred search paths (#4) * fix: remove files updated by precommit in merge * fix: remove log file * Copyright update --------- Co-authored-by: David Dixon <david.dixon@amd.com> [ROCm/hipBLASLt commit: f45e5b9]

* add in cmake targets to run clang check * add ability to do clang format check via make target * add CONFIGURE_DEPENDS * modify target names to suggested ones * yeet my debugging line

* add in cmake targets to run clang check * add ability to do clang format check via make target * add CONFIGURE_DEPENDS * modify target names to suggested ones * yeet my debugging line [ROCm/hipDNN commit: 069fe9d]

…nel configurations (#1978) # Fix GenericSearch warm-up bias: apply warm-up to all configurations >** 📝 Note**: This PR has follow up PR #1993 ## Motivation MIOpen's generic search algorithm suffers from a **race condition** that causes optimal kernels to be randomly rejected, leading to 3-4x performance degradation in some cases. ### Problem Description When running the same convolution workload multiple times as sample below: ```bash MIOpenDriver convbfp16 -n 8 -c 5 -H 225 -W 225 -k 64 -y 3 -x 3 -p 1 -q 1 \ -u 1 -v 1 -l 1 -j 1 --in_layout NHWC --fil_layout NHWC --out_layout NHWC \ -m conv -g 1 -t 1 -F 2 ``` **Observed behavior:** - **Lucky case**: Selected optimal kernel → **0.099 ms** per operation - **Unlucky case**: Selected suboptimal kernel → **0.332 ms** per operation (**3.35x slower**) [Lucky_Joe_20250930.log](https://github.com/user-attachments/files/22714922/Lucky_Joe_20250930.log) [Normal_Joe_20250930.log](https://github.com/user-attachments/files/22714923/Normal_Joe_20250930.log) ### Root Cause **Cold-start bias in warm-up logic** (`generic_search.hpp`, lines 559-564): ```cpp // Original buggy code if(n_current == 0) // ❌ Only first config gets warm-up { invoker(profile_h, invoke_ctx); profile_h.ResetKernelTime(); } ``` This condition creates an **unfair advantage** for the first configuration tested: - **First kernel** (n_current == 0): Gets warm-up → Fair performance measurement - **Subsequent kernels** (n_current > 0): No warm-up → Cold-start penalty (up to **100x slower** in extreme cases) ### Impact - **High false negative rate**: Up to 40% chance of rejecting the optimal kernel - **Performance degradation**: 4x slower execution when suboptimal kernel is selected - **Non-deterministic behavior**: Kernel selection depends on which configuration is tested first ### Example from Production Logs **Environment**: MI355X (gfx950), ROCm 7.0.2 ``` AI generated 4 kernel configurations for testing: Kernel #0 (128,128,32,32,8,8...): 10 samples → avg 0.343166 ms → selected as "best" Kernel #1 (64,64,64,32,8,8...): 1 sample → 1.219 ms → rejected (cold-start!) Kernel #2 (64,64,16,32,8,8...): 1 sample → 3.0267 ms → rejected (cold-start!) Kernel #3 (64,16,64,32,8,8...): 1 sample → 0.482 ms → rejected (cold-start!) Final execution: 0.332 ms (using Kernel #0) Issue: Kernel #2 suffered from cold-start bias (3.0267 ms first sample) With proper warm-up, its true performance is ~0.099 ms (3.4x faster than selected kernel) ``` **Detailed timing from Normal_Joe_20250930.log:** Optimal kernel (incorrectly rejected due to cold-start): - `DeviceGroupedConvBwdData_Xdl_CShuffle_v1<64,64,16,32,8,8,Default,16,16,4,1,8,1,1,1>+1` - Sample 1: **3.027 ms** ← Cold start! (30x slower than true performance) - Samples 2-11: 0.369, 0.349, 0.366, 0.352, 0.353, 0.365, 0.352, 0.359, 0.347, 0.352 ms - **True mean**: 0.354 ms (excluding cold-start outlier) - **Decision**: Rejected by early-stop (3.027 > 0.377 × 1.1) - **Wrong outcome**: Best kernel discarded due to unfair cold-start penalty --- ## Technical Details ### Changes This PR contains **only the bug fix** - removing the unfair warm-up condition: ```diff - // Warm-up run for first time invoker is used - if(n_current == 0) - { - invoker(profile_h, invoke_ctx); - profile_h.ResetKernelTime(); - } + // Warm-up run for every configuration to eliminate cold-start bias + invoker(profile_h, invoke_ctx); + profile_h.ResetKernelTime(); ``` **File modified:** `projects/miopen/src/include/miopen/generic_search.hpp` (lines 559-564) **Change summary:** - 3 insertions(+), 6 deletions(-) - Removes `if(n_current == 0)` condition - Ensures every configuration receives one warm-up run before measurement ### Why This is Low Risk 1. **Minimal code change**: Only 4 lines changed 2. **No algorithm change**: Same sampling strategy, same early-stop logic 3. **Only ensures fairness**: All configs now receive identical warm-up treatment 4. **No performance regression**: Adds one extra kernel call per config (~0.3ms overhead per config) 5. **Negligible overhead**: For typical 4-config search, adds 1.2ms total (kernel compilation takes 10-30 seconds, so overhead is <0.01%) --- ## Test Plan ### Test Environment - **Hardware**: MI355X (gfx950) - **ROCm Version**: 7.0.2 (HIP 7.0.51831) - **Workload**: Grouped convolution backward data (NHWC layout, 4 kernel configurations) ### Test Command ```bash export MIOPEN_LOG_LEVEL=5 export MIOPEN_FIND_MODE=1 ./bin/MIOpenDriver convbfp16 -n 8 -c 5 -H 225 -W 225 -k 64 -y 3 -x 3 \ -p 1 -q 1 -u 1 -v 1 -l 1 -j 1 \ --in_layout NHWC --fil_layout NHWC --out_layout NHWC \ -m conv -g 1 -t 1 -F 2 ``` ### Test Results #### Before Fix (with bug): - **Success rate**: 6/10 runs selected optimal kernel (40% error rate) - **Failure pattern**: Optimal kernel rejected when its cold-start time triggered early-stop - **Performance impact**: Up to 4x slower when wrong kernel selected (0.332ms vs 0.099ms) #### After Fix: - **Success rate**: 10/10 runs selected optimal kernel (0% error rate) - **Consistency**: All configurations receive fair warm-up - **Performance**: Optimal kernel always selected, no degradation - **Overhead**: +1.2ms for 4 configs (negligible vs 10-30s compilation time) --- ## Performance Impact ### Search Time Overhead - **Additional cost**: 1 warm-up run per configuration (only for configs beyond the first) ### Accuracy Improvement - **Before**: 60% success rate (6/10 runs correct) - **After**: 100% success rate (10/10 runs correct) - **Performance gain**: Eliminates 4x slowdown from selecting wrong kernel --- ## Backward Compatibility ✅ **Fully compatible** - No API changes - No behavior changes except for fixing the bug - All existing tests pass - No impact on already-cached kernels (find database not affected) --- ## Why Split into Two PRs? Following reviewer feedback, this work has been split into two separate PRs: ### **PR1 (This PR) - Bug Fix: Warm-up Bias** - **Risk**: Low (4 lines changed) - **Impact**: Fixes root cause of unfair kernel comparison - **Decision**: Ready for immediate merge - **Rationale**: Without fair warm-up, no amount of threshold tuning can fix the problem. Cold-start penalties (30-100x slower) make any single threshold value inadequate. ### **PR2 (Separate PR) - Optimization: Early-Stop Strategy** - **Branch**: `users/JoeLiuAMD/miopen-generic-search-optimization` #1993 - **Changes**: Dual-sample testing + 1.2x threshold + enhanced logging - **Risk**: Medium (affects benchmark timing) - **Decision**: Needs more validation and benchmarking - **Rationale**: These optimizations improve accuracy further (40% → 0% error rate) but add ~2 kernel executions per config. The performance impact needs separate evaluation. ### Why This Approach? - **PR1 can merge immediately**: Fixes the critical bug with minimal risk - **PR2 can be validated thoroughly**: Performance trade-offs can be evaluated independently - **Easier to isolate regressions**: If issues arise, we know which change caused them - **Progressive improvement**: Get the bug fix deployed while optimizations are being validated --- ## Submission Checklist - [x] I have read and agreed with the [contributing guidelines](CONTRIBUTING.md) - [x] The changes are minimal and focused on the bug fix - [x] All existing tests pass - [x] The fix has been verified on target hardware (MI355X/gfx950) - [x] The fix eliminates the non-deterministic kernel selection issue - [x] No performance regression introduced - [x] Documentation (commit message) clearly explains the problem and solution - [x] Backward compatibility confirmed - [x] Test data and logs provided for verification

## Motivation Enable gfx1152 and gfx1153. ## Technical Details 1. combine arrays into tables and use local macros to reduce repetition (for maintainability) 2. monkey-see-monkey-do wherever `gfx11...` was found ## Test Plan Build existing ctests for, and run them on, gfx1152 and gfx1153. ## Test Result ### <details> <summary>gfx1152 passed (click to see log)</summary> ``` INFO:root:++ Exec [/tmp/eble]$ ctest --test-dir /tmp/eble/rocm/bin/rocprim --output-o n-failure --parallel 2 --exclude-regex 'rocprim.lookback_reproducibility|rocprim.link ing|rocprim.device_merge_inplace|rocprim.device_merge_sort|rocprim.device_partition|r ocprim.device_radix_sort|rocprim.device_scan|rocprim.device_select|rocprim.device_fin d_first_of|rocprim.device_reduce_by_key' --timeout 60 Test project /tmp/eble/rocm/bin/rocprim Start 1: hip.device_api Start 2: hip.async_copy 1/73 Test #2: hip.async_copy .............................. Passed 0.01 sec Start 3: hip.ordered_block_id 2/73 Test #1: hip.device_api .............................. Passed 0.02 sec Start 4: rocprim.internal_merge_path 3/73 Test #4: rocprim.internal_merge_path ................. Passed 0.01 sec Start 5: rocprim.basic_test 4/73 Test #3: hip.ordered_block_id ........................ Passed 0.01 sec Start 6: rocprim.arg_index_iterator 5/73 Test #5: rocprim.basic_test .......................... Passed 0.01 sec Start 7: rocprim.temporary_storage_partitioning 6/73 Test #6: rocprim.arg_index_iterator .................. Passed 0.01 sec Start 8: rocprim.block_adjacent_difference 7/73 Test #7: rocprim.temporary_storage_partitioning ...... Passed 0.01 sec Start 9: rocprim.block_discontinuity 8/73 Test #8: rocprim.block_adjacent_difference ........... Passed 2.34 sec Start 10: rocprim.bit_cast 9/73 Test #10: rocprim.bit_cast ............................ Passed 0.02 sec Start 11: rocprim.block_exchange 10/73 Test #11: rocprim.block_exchange ...................... Passed 0.73 sec Start 12: rocprim.block_histogram 11/73 Test #12: rocprim.block_histogram ..................... Passed 0.54 sec Start 13: rocprim.block_load_store 12/73 Test #13: rocprim.block_load_store .................... Passed 0.44 sec Start 14: rocprim.block_sort_merge 13/73 Test #14: rocprim.block_sort_merge .................... Passed 0.02 sec Start 15: rocprim.block_sort_merge_stable 14/73 Test #15: rocprim.block_sort_merge_stable ............. Passed 0.02 sec Start 16: rocprim.block_radix_rank 15/73 Test #16: rocprim.block_radix_rank .................... Passed 0.03 sec Start 17: rocprim.block_radix_sort 16/73 Test #17: rocprim.block_radix_sort .................... Passed 4.79 sec Start 18: rocprim.block_reduce 17/73 Test #18: rocprim.block_reduce ........................ Passed 0.26 sec Start 19: rocprim.block_run_length_decode 18/73 Test #19: rocprim.block_run_length_decode ............. Passed 0.54 sec Start 20: rocprim.block_scan 19/73 Test #20: rocprim.block_scan .......................... Passed 0.04 sec Start 21: rocprim.block_shuffle 20/73 Test #21: rocprim.block_shuffle ....................... Passed 2.70 sec Start 22: rocprim.block_sort_bitonic 21/73 Test #9: rocprim.block_discontinuity ................. Passed 17.36 sec Start 23: rocprim.config_dispatch 22/73 Test #23: rocprim.config_dispatch ..................... Passed 0.09 sec Start 24: rocprim.constant_iterator 23/73 Test #24: rocprim.constant_iterator ................... Passed 0.07 sec Start 25: rocprim.counting_iterator 24/73 Test #25: rocprim.counting_iterator ................... Passed 0.07 sec Start 26: rocprim.device_batch_memcpy 25/73 Test #26: rocprim.device_batch_memcpy ................. Passed 1.20 sec Start 27: rocprim.device_binary_search 26/73 Test #27: rocprim.device_binary_search ................ Passed 0.02 sec Start 28: rocprim.device_adjacent_difference 27/73 Test #28: rocprim.device_adjacent_difference .......... Passed 0.01 sec Start 29: rocprim.device_adjacent_find 28/73 Test #29: rocprim.device_adjacent_find ................ Passed 0.01 sec Start 30: rocprim.device_find_end 29/73 Test #30: rocprim.device_find_end ..................... Passed 0.01 sec Start 31: rocprim.device_histogram 30/73 Test #22: rocprim.block_sort_bitonic .................. Passed 13.23 sec Start 32: rocprim.device_merge 31/73 Test #31: rocprim.device_histogram .................... Passed 7.85 sec Start 33: rocprim.nth_element 32/73 Test #33: rocprim.nth_element ......................... Passed 0.03 sec Start 34: rocprim.device_partial_sort 33/73 Test #34: rocprim.device_partial_sort ................. Passed 0.02 sec Start 35: rocprim.device_reduce 34/73 Test #35: rocprim.device_reduce ....................... Passed 8.94 sec Start 36: rocprim.device_run_length_encode 35/73 Test #32: rocprim.device_merge ........................ Passed 14.05 sec Start 37: rocprim.device_search 36/73 Test #37: rocprim.device_search ....................... Passed 0.02 sec Start 38: rocprim.device_segmented_radix_sort 37/73 Test #36: rocprim.device_run_length_encode ............ Passed 13.92 sec Start 39: rocprim.device_search_n 38/73 Test #39: rocprim.device_search_n ..................... Passed 0.02 sec Start 40: rocprim.device_segmented_reduce 39/73 Test #40: rocprim.device_segmented_reduce ............. Passed 5.21 sec Start 41: rocprim.device_segmented_scan 40/73 Test #41: rocprim.device_segmented_scan ............... Passed 0.02 sec Start 42: rocprim.device_transform 41/73 Test #42: rocprim.device_transform .................... Passed 13.90 sec Start 43: rocprim.discard_iterator 42/73 Test #43: rocprim.discard_iterator .................... Passed 0.07 sec Start 44: rocprim.radix_key_codec 43/73 Test #44: rocprim.radix_key_codec ..................... Pas09:54:11 [55/1943] Start 45: rocprim.predicate_iterator 44/73 Test #45: rocprim.predicate_iterator .................. Passed 0.07 sec Start 46: rocprim.reverse_iterator 45/73 Test #46: rocprim.reverse_iterator .................... Passed 0.09 sec Start 47: rocprim.rocprim_tuple 46/73 Test #47: rocprim.rocprim_tuple ....................... Passed 0.01 sec Start 48: rocprim.rocprim_types 47/73 Test #48: rocprim.rocprim_types ....................... Passed 0.01 sec Start 49: rocprim.texture_cache_iterator 48/73 Test #49: rocprim.texture_cache_iterator .............. Passed 0.01 sec Start 50: rocprim.thread 49/73 Test #50: rocprim.thread .............................. Passed 0.07 sec Start 51: rocprim.thread_algos 50/73 Test #51: rocprim.thread_algos ........................ Passed 0.35 sec Start 52: rocprim.tuple 51/73 Test #52: rocprim.tuple ............................... Passed 0.02 sec Start 53: rocprim.utils_sort_checker 52/73 Test #53: rocprim.utils_sort_checker .................. Passed 0.01 sec Start 54: rocprim.transform_iterator 53/73 Test #54: rocprim.transform_iterator .................. Passed 0.11 sec Start 55: rocprim.type_traits_interface_cpp17 54/73 Test #55: rocprim.type_traits_interface_cpp17 ......... Passed 0.01 sec Start 56: rocprim.type_traits_interface_gnupp17 55/73 Test #56: rocprim.type_traits_interface_gnupp17 ....... Passed 0.01 sec Start 57: rocprim.type_traits_interface_cpp20 56/73 Test #57: rocprim.type_traits_interface_cpp20 ......... Passed 0.01 sec Start 58: rocprim.type_traits_interface_gnupp20 57/73 Test #58: rocprim.type_traits_interface_gnupp20 ....... Passed 0.01 sec Start 59: rocprim.no_half_operators 58/73 Test #59: rocprim.no_half_operators ................... Passed 0.01 sec Start 60: rocprim.intrinsics 59/73 Test #60: rocprim.intrinsics .......................... Passed 0.21 sec Start 61: rocprim.intrinsics_atomic 60/73 Test #61: rocprim.intrinsics_atomic ................... Passed 0.02 sec Start 62: rocprim.invoke_result 61/73 Test #62: rocprim.invoke_result ....................... Passed 0.01 sec Start 63: rocprim.warp_exchange 62/73 Test #63: rocprim.warp_exchange ....................... Passed 0.08 sec Start 64: rocprim.warp_load 63/73 Test #64: rocprim.warp_load ........................... Passed 0.08 sec Start 65: rocprim.warp_reduce 64/73 Test #65: rocprim.warp_reduce ......................... Passed 0.14 sec Start 66: rocprim.warp_scan 65/73 Test #66: rocprim.warp_scan ........................... Passed 0.20 sec Start 67: rocprim.warp_scan_disable_dpp_disable_dpp 66/73 Test #67: rocprim.warp_scan_disable_dpp_disable_dpp ... Passed 0.21 sec Start 68: rocprim.warp_sort 67/73 Test #68: rocprim.warp_sort ........................... Passed 0.09 sec Start 69: rocprim.warp_store 68/73 Test #69: rocprim.warp_store .......................... Passed 0.02 sec Start 70: rocprim.zip_iterator 69/73 Test #70: rocprim.zip_iterator ........................ Passed 0.02 sec Start 71: rocprim.accumulator_t 70/73 Test #71: rocprim.accumulator_t ....................... Passed 0.02 sec Start 72: hipgraph.basic 71/73 Test #72: hipgraph.basic .............................. Passed 0.02 sec Start 73: hipgraph.algs 72/73 Test #73: hipgraph.algs ............................... Passed 0.01 sec 73/73 Test #38: rocprim.device_segmented_radix_sort ......... Passed 31.97 sec 100% tests passed, 0 tests failed out of 73 Total Test time (real) = 71.80 sec ✅ test_rocprim.py PASSED ``` </details> <details> <summary>gfx1153 passed (click to see log)</summary> ``` INFO:root:++ Exec [/tmp/eble]$ ctest --test-dir /tmp/eble/rocm/bin/rocprim --output-o n-failure --parallel 1 --exclude-regex 'rocprim.lookback_reproducibility|rocprim.link ing|rocprim.device_merge_inplace|rocprim.device_merge_sort|rocprim.device_partition|r ocprim.device_radix_sort|rocprim.device_scan|rocprim.device_select|rocprim.device_fin d_first_of|rocprim.device_reduce_by_key' --timeout 60 Test project /tmp/eble/rocm/bin/rocprim Start 1: hip.device_api 1/73 Test #1: hip.device_api .............................. Passed 0.01 sec Start 2: hip.async_copy 2/73 Test #2: hip.async_copy .............................. Passed 0.01 sec Start 3: hip.ordered_block_id 3/73 Test #3: hip.ordered_block_id ........................ Passed 0.01 sec Start 4: rocprim.internal_merge_path 4/73 Test #4: rocprim.internal_merge_path ................. Passed 0.01 sec Start 5: rocprim.basic_test 5/73 Test #5: rocprim.basic_test .......................... Passed 0.01 sec Start 6: rocprim.arg_index_iterator 6/73 Test #6: rocprim.arg_index_iterator .................. Passed 0.01 sec Start 7: rocprim.temporary_storage_partitioning 7/73 Test #7: rocprim.temporary_storage_partitioning ...... Passed 0.01 sec Start 8: rocprim.block_adjacent_difference 8/73 Test #8: rocprim.block_adjacent_difference ........... Passed 2.94 sec Start 9: rocprim.block_discontinuity 9/73 Test #9: rocprim.block_discontinuity ................. Passed 21.10 sec Start 10: rocprim.bit_cast 10/73 Test #10: rocprim.bit_cast ............................ Passed 0.01 sec Start 11: rocprim.block_exchange 11/73 Test #11: rocprim.block_exchange ...................... Passed 2.20 sec Start 12: rocprim.block_histogram 12/73 Test #12: rocprim.block_histogram ..................... Passed 0.71 sec Start 13: rocprim.block_load_store 13/73 Test #13: rocprim.block_load_store .................... Passed 0.48 sec Start 14: rocprim.block_sort_merge 14/73 Test #14: rocprim.block_sort_merge .................... Passed 0.02 sec Start 15: rocprim.block_sort_merge_stable 15/73 Test #15: rocprim.block_sort_merge_stable ............. Passed 0.02 sec Start 16: rocprim.block_radix_rank 16/73 Test #16: rocprim.block_radix_rank .................... Passed 0.02 sec Start 17: rocprim.block_radix_sort 17/73 Test #17: rocprim.block_radix_sort .................... Passed 6.12 sec Start 18: rocprim.block_reduce 18/73 Test #18: rocprim.block_reduce ........................ Passed 0.31 sec Start 19: rocprim.block_run_length_decode 19/73 Test #19: rocprim.block_run_length_decode ............. Passed 0.68 sec Start 20: rocprim.block_scan 20/73 Test #20: rocprim.block_scan .......................... Passed 0.03 sec Start 21: rocprim.block_shuffle 21/73 Test #21: rocprim.block_shuffle ....................... Passed 3.63 sec Start 22: rocprim.block_sort_bitonic 22/73 Test #22: rocprim.block_sort_bitonic .................. Passed 19.34 sec Start 23: rocprim.config_dispatch 23/73 Test #23: rocprim.config_dispatch ..................... Passed 0.10 sec Start 24: rocprim.constant_iterator 24/73 Test #24: rocprim.constant_iterator ................... Passed 0.09 sec Start 25: rocprim.counting_iterator 25/73 Test #25: rocprim.counting_iterator ................... Passed 0.09 sec Start 26: rocprim.device_batch_memcpy 26/73 Test #26: rocprim.device_batch_memcpy ................. Passed 1.42 sec Start 27: rocprim.device_binary_search 27/73 Test #27: rocprim.device_binary_search ................ Passed 0.01 sec Start 28: rocprim.device_adjacent_difference 28/73 Test #28: rocprim.device_adjacent_difference .......... Passed 0.01 sec Start 29: rocprim.device_adjacent_find 29/73 Test #29: rocprim.device_adjacent_find ................ Passed 0.01 sec Start 30: rocprim.device_find_end 30/73 Test #30: rocprim.device_find_end ..................... Passed 0.01 sec Start 31: rocprim.device_histogram 31/73 Test #31: rocprim.device_histogram .................... Passed 8.77 sec Start 32: rocprim.device_merge 32/73 Test #32: rocprim.device_merge ........................ Passed 16.23 sec Start 33: rocprim.nth_element 33/73 Test #33: rocprim.nth_element ......................... Passed 0.01 sec Start 34: rocprim.device_partial_sort 34/73 Test #34: rocprim.device_partial_sort ................. Passed 0.02 sec Start 35: rocprim.device_reduce 35/73 Test #35: rocprim.device_reduce ....................... Passed 10.92 sec Start 36: rocprim.device_run_length_encode 36/73 Test #36: rocprim.device_run_length_encode ............ Passed 14.34 sec Start 37: rocprim.device_search 37/73 Test #37: rocprim.device_search ....................... Passed 0.01 sec Start 38: rocprim.device_segmented_radix_sort 38/73 Test #38: rocprim.device_segmented_radix_sort ......... Passed 38.28 sec Start 39: rocprim.device_search_n 39/73 Test #39: rocprim.device_search_n ..................... Passed 0.02 sec Start 40: rocprim.device_segmented_reduce 40/73 Test #40: rocprim.device_segmented_reduce ............. Passed 7.19 sec Start 41: rocprim.device_segmented_scan 41/73 Test #41: rocprim.device_segmented_scan ............... Passed 0.02 sec Start 42: rocprim.device_transform 42/73 Test #42: rocprim.device_transform .................... Passed 17.64 sec Start 43: rocprim.discard_iterator 43/73 Test #43: rocprim.discard_iterator .................... Passed 0.12 sec Start 44: rocprim.radix_key_codec 44/73 Test #44: rocprim.radix_key_codec ..................... Passed 0.01 sec Start 45: rocprim.predicate_iterator 45/73 Test #45: rocprim.predicate_iterator .................. Passed 0.08 sec Start 46: rocprim.reverse_iterator 46/73 Test #46: rocprim.reverse_iterator .................... Pas10:13:26 [46/1844] Start 47: rocprim.rocprim_tuple 47/73 Test #47: rocprim.rocprim_tuple ....................... Passed 0.01 sec Start 48: rocprim.rocprim_types 48/73 Test #48: rocprim.rocprim_types ....................... Passed 0.01 sec Start 49: rocprim.texture_cache_iterator 49/73 Test #49: rocprim.texture_cache_iterator .............. Passed 0.01 sec Start 50: rocprim.thread 50/73 Test #50: rocprim.thread .............................. Passed 0.08 sec Start 51: rocprim.thread_algos 51/73 Test #51: rocprim.thread_algos ........................ Passed 0.43 sec Start 52: rocprim.tuple 52/73 Test #52: rocprim.tuple ............................... Passed 0.01 sec Start 53: rocprim.utils_sort_checker 53/73 Test #53: rocprim.utils_sort_checker .................. Passed 0.01 sec Start 54: rocprim.transform_iterator 54/73 Test #54: rocprim.transform_iterator .................. Passed 0.12 sec Start 55: rocprim.type_traits_interface_cpp17 55/73 Test #55: rocprim.type_traits_interface_cpp17 ......... Passed 0.01 sec Start 56: rocprim.type_traits_interface_gnupp17 56/73 Test #56: rocprim.type_traits_interface_gnupp17 ....... Passed 0.01 sec Start 57: rocprim.type_traits_interface_cpp20 57/73 Test #57: rocprim.type_traits_interface_cpp20 ......... Passed 0.01 sec Start 58: rocprim.type_traits_interface_gnupp20 58/73 Test #58: rocprim.type_traits_interface_gnupp20 ....... Passed 0.01 sec Start 59: rocprim.no_half_operators 59/73 Test #59: rocprim.no_half_operators ................... Passed 0.01 sec Start 60: rocprim.intrinsics 60/73 Test #60: rocprim.intrinsics .......................... Passed 0.29 sec Start 61: rocprim.intrinsics_atomic 61/73 Test #61: rocprim.intrinsics_atomic ................... Pas10:13:27 [16/1844] Start 62: rocprim.invoke_result 62/73 Test #62: rocprim.invoke_result ....................... Passed 0.01 sec Start 63: rocprim.warp_exchange 63/73 Test #63: rocprim.warp_exchange ....................... Passed 0.09 sec Start 64: rocprim.warp_load 64/73 Test #64: rocprim.warp_load ........................... Passed 0.09 sec Start 65: rocprim.warp_reduce 65/73 Test #65: rocprim.warp_reduce ......................... Passed 0.17 sec Start 66: rocprim.warp_scan 66/73 Test #66: rocprim.warp_scan ........................... Passed 0.26 sec Start 67: rocprim.warp_scan_disable_dpp_disable_dpp 67/73 Test #67: rocprim.warp_scan_disable_dpp_disable_dpp ... Passed 0.26 sec Start 68: rocprim.warp_sort 68/73 Test #68: rocprim.warp_sort ........................... Passed 0.10 sec Start 69: rocprim.warp_store 69/73 Test #69: rocprim.warp_store .......................... Passed 0.01 sec Start 70: rocprim.zip_iterator 70/73 Test #70: rocprim.zip_iterator ........................ Passed 0.01 sec Start 71: rocprim.accumulator_t 71/73 Test #71: rocprim.accumulator_t ....................... Passed 0.01 sec Start 72: hipgraph.basic 72/73 Test #72: hipgraph.basic .............................. Passed 0.01 sec Start 73: hipgraph.algs 73/73 Test #73: hipgraph.algs ............................... Passed 0.01 sec 100% tests passed, 0 tests failed out of 73 Total Test time (real) = 175.34 sec ✅ test_rocprim.py PASSED ``` <detail> ## Submission Checklist - [x] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

## Summary Improved memory management in `host_alloc.cpp` by refactoring the deallocation logic for better efficiency and adding diagnostic warnings for untracked pointer frees. ## Changes 1. **Refactored `free_ptr_use()` to use iterator-based access** (reduces redundant map lookups) 2. **Added warning for freeing untracked pointers** (helps detect potential memory corruption issues) ## Motivation The current implementation of `free_ptr_use()` in `host_alloc.cpp` had two issues: 1. **Performance:** Redundant map lookups when deallocating tracked memory 2. **Correctness:** Using `operator[]` on a map can unintentionally insert entries for non-existent keys 3. **Diagnostics:** Silent failures when attempting to free untracked pointers made debugging difficult This PR addresses all three issues by refactoring the lookup logic and adding diagnostic warnings. ## Technical Details **File:** `projects/rocblas/clients/common/host_alloc.cpp` ### Change 1: Iterator-based memory deallocation **Before:** ```cpp if(ptr && mem_allocated[ptr]) { // Lookup #1 (may insert!) mem_used -= mem_allocated[ptr]; // Lookup #2 mem_allocated.erase(ptr); // Lookup #3 } ``` **After:** ```cpp auto it = mem_allocated.find(ptr); // Single lookup if(ptr && it != mem_allocated.end()) { mem_used -= it->second; // Use cached iterator mem_allocated.erase(it); // Use cached iterator } ``` **Benefits:** - Reduces map lookups from 3 to 1 - Prevents unintended map insertions via `operator[]` - More efficient memory tracking ### Change 2: Diagnostic warning for untracked pointers **Added:** ```cpp else if(ptr && call_free) { rocblas_cerr << "Warning: Freeing untracked pointer " << ptr << " - untracked memory released (potential double-free or memory corruption)" << std::endl; } ``` **Benefits:** - Helps detect double-free bugs - Identifies potential memory corruption issues - Provides actionable diagnostic information during testing ## Test Plan - Built rocBLAS with no compilation errors - Existing client test suite exercises memory allocation/deallocation paths - No functional behavior changes to normal operation - Warning messages help identify issues during debugging ## Test Result - All existing tests continue to pass - Memory tracking functionality unchanged - Warning properly triggers for untracked pointer frees - No behavioral differences for correctly tracked memory ## Submission Checklist - [x] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

#3710) ## Motivation Optimizing the tensor filling functions started a discussion about optimizing tensor iteration in general: #3471 (comment) ## Technical Details After some deliberation, the approach taken here (using std::variant inside the iterator to represent the different types of indexing) reflects both the desire the improve iteration in the case of packed tensors while also maintaining the existing API. A fully templated approach would be more optimal but would require API changes to the ITensor class itself, whether making it templated or changing the definition of its iterator-related methods at the very least. ## Test Plan Ran ninja check inside the build folder of hipDNN. ## Test Result ``` [185/187] Running all tests via ctest Test project /therock/output/build/ml-libs/hipDNN/build Start 1: hipdnn_data_sdk_tests 1/7 Test #1: hipdnn_data_sdk_tests ............ Passed 1.30 sec Start 2: hipdnn_backend_tests 2/7 Test #2: hipdnn_backend_tests ............. Passed 1.29 sec Start 3: hipdnn_frontend_tests 3/7 Test #3: hipdnn_frontend_tests ............ Passed 0.03 sec Start 4: hipdnn_test_sdk_tests 4/7 Test #4: hipdnn_test_sdk_tests ............ Passed 4.32 sec Start 5: hipdnn_plugin_sdk_tests 5/7 Test #5: hipdnn_plugin_sdk_tests .......... Passed 0.03 sec Start 6: public_hipdnn_backend_tests 6/7 Test #6: public_hipdnn_backend_tests ...... Passed 0.33 sec Start 7: public_hipdnn_frontend_tests 7/7 Test #7: public_hipdnn_frontend_tests ..... Passed 0.26 sec 100% tests passed, 0 tests failed out of 7 Label Time Summary: integration_test = 0.59 sec*proc (2 tests) unit_test = 6.96 sec*proc (5 tests) Total Test time (real) = 7.56 sec ``` ## Submission Checklist - [x] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests. --------- Co-authored-by: BrianHarrisonAMD <169072757+BrianHarrisonAMD@users.noreply.github.com>

…ernel (ROCm#6) * Add sample subtile impl * Fix issues when disabling subtile impl * GR Offset calculation (#1) * Add sample subtile impl * Move allocOffsetRegisters before setupNewTile * Start adding GR offset calculation * Rest of logic (no swizzling) * refacto * spgr offsets * Add newserial code * Add script to debug offsets * Add unit test for GR offset calculation * Grid display * Fix both code and ref test function * Add DPP quad perm to rocisa * Apply swizzling (no rotation yet) * Function swizzling + rotation + test * Refactor test to have a single output array + add test for SGPRs * Add debug mode to test + add dynamic wavegroup calculation based on MT * Fix test runtime issue and check all vgpr offsets * Add ref test code for 1x4 & 4x1 * Fix tests * Fixed SGPR offset calculation for 2x2 * Fix more tests * Add more tests * Refactor tests * simplify tests * Remove unused script * cleanup * fix camelCase in ref test code * cleanup * Fix typo --------- Co-authored-by: brianshi <brianshi@amd.com> * Enable post-loop code generation, and add some subroutines * LR offset calculation (#2) * Add tests * as is * Add permlane16_swap instruction to rocisa * Ongoing progress * Draft for partition A0/A1 * Wave partitioning * Draft ref code in tests * Handle 1x4 wavesplit param * 2x2 test passing * Draft 1x4 LR wave partitioning * Fix alginement issue * Integration testing * Update integration test * Fix swizzling pattern on GRA. Only swizzling on even LDS rows * Subtile based test * testing A * Test both A and B * Remove graonly mode * Fix 1x4 case * Move global offset for B after rest of the logic * cleanup * cleanup * Fix ref test code for 4x1 * Fix spgr alloc issue * Remove tmp test file * Remove debug prints * Add test case * Add GR load emit logic, and misc fixes (#3) * gr emit fix * Emit LR + init ACCVGPR (#4) * Emit ds_reads * Add waits for LR and GR * Init Acc VGPR to Zero * Add missing bit_length on VLShiftLeftB32 * Insert SNop between VLShiftLeftB32 & VReadfirstlaneB32 for correctness * Fix gra test ref code for 1x4 * Remove some debug prints * Add loop and ptr update code * Update scale offset * Add tests * Address review * Add scale roundtrip e2e test and constraint assertions Add GR->LDS->LR roundtrip GPU test verifying scale offset consistency across 4 tile configs x 2 matrices. Add power-of-2 assertion for scaleBlockSize and matching scaleBlockSize assertions for A/B in shared GR/LR offset computation. Pass kernel dict to compute_lds_sizes instead of re-deriving MIWaveGroup from tile dimensions. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Update fixes * Fix scale being skipped * Add flag to print layout * Fix missed merge conflicts * Fix missed merge conflicts * Refactor scale rountrip test with gpu helper fns * Fix extra spaces * Fix tests --------- Co-authored-by: brianshi <brianshi@amd.com> Co-authored-by: sebvince <115461989+sebvince@users.noreply.github.com> Co-authored-by: b-shi <bbbrianme@gmail.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

…ernel (ROCm#6) * Add sample subtile impl * Fix issues when disabling subtile impl * GR Offset calculation (ROCm#1) * Add sample subtile impl * Move allocOffsetRegisters before setupNewTile * Start adding GR offset calculation * Rest of logic (no swizzling) * refacto * spgr offsets * Add newserial code * Add script to debug offsets * Add unit test for GR offset calculation * Grid display * Fix both code and ref test function * Add DPP quad perm to rocisa * Apply swizzling (no rotation yet) * Function swizzling + rotation + test * Refactor test to have a single output array + add test for SGPRs * Add debug mode to test + add dynamic wavegroup calculation based on MT * Fix test runtime issue and check all vgpr offsets * Add ref test code for 1x4 & 4x1 * Fix tests * Fixed SGPR offset calculation for 2x2 * Fix more tests * Add more tests * Refactor tests * simplify tests * Remove unused script * cleanup * fix camelCase in ref test code * cleanup * Fix typo --------- Co-authored-by: brianshi <brianshi@amd.com> * Enable post-loop code generation, and add some subroutines * LR offset calculation (ROCm#2) * Add tests * as is * Add permlane16_swap instruction to rocisa * Ongoing progress * Draft for partition A0/A1 * Wave partitioning * Draft ref code in tests * Handle 1x4 wavesplit param * 2x2 test passing * Draft 1x4 LR wave partitioning * Fix alginement issue * Integration testing * Update integration test * Fix swizzling pattern on GRA. Only swizzling on even LDS rows * Subtile based test * testing A * Test both A and B * Remove graonly mode * Fix 1x4 case * Move global offset for B after rest of the logic * cleanup * cleanup * Fix ref test code for 4x1 * Fix spgr alloc issue * Remove tmp test file * Remove debug prints * Add test case * Add GR load emit logic, and misc fixes (ROCm#3) * gr emit fix * Emit LR + init ACCVGPR (ROCm#4) * Emit ds_reads * Add waits for LR and GR * Init Acc VGPR to Zero * Add missing bit_length on VLShiftLeftB32 * Insert SNop between VLShiftLeftB32 & VReadfirstlaneB32 for correctness * Fix gra test ref code for 1x4 * Remove some debug prints * Add loop and ptr update code * Update scale offset * Add tests * Address review * Add scale roundtrip e2e test and constraint assertions Add GR->LDS->LR roundtrip GPU test verifying scale offset consistency across 4 tile configs x 2 matrices. Add power-of-2 assertion for scaleBlockSize and matching scaleBlockSize assertions for A/B in shared GR/LR offset computation. Pass kernel dict to compute_lds_sizes instead of re-deriving MIWaveGroup from tile dimensions. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Update fixes * Fix scale being skipped * Add flag to print layout * Fix missed merge conflicts * Fix missed merge conflicts * Refactor scale rountrip test with gpu helper fns * Fix extra spaces * Fix tests --------- Co-authored-by: brianshi <brianshi@amd.com> Co-authored-by: sebvince <115461989+sebvince@users.noreply.github.com> Co-authored-by: b-shi <bbbrianme@gmail.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

…ernel (ROCm#6) * Add sample subtile impl * Fix issues when disabling subtile impl * GR Offset calculation (#1) * Add sample subtile impl * Move allocOffsetRegisters before setupNewTile * Start adding GR offset calculation * Rest of logic (no swizzling) * refacto * spgr offsets * Add newserial code * Add script to debug offsets * Add unit test for GR offset calculation * Grid display * Fix both code and ref test function * Add DPP quad perm to rocisa * Apply swizzling (no rotation yet) * Function swizzling + rotation + test * Refactor test to have a single output array + add test for SGPRs * Add debug mode to test + add dynamic wavegroup calculation based on MT * Fix test runtime issue and check all vgpr offsets * Add ref test code for 1x4 & 4x1 * Fix tests * Fixed SGPR offset calculation for 2x2 * Fix more tests * Add more tests * Refactor tests * simplify tests * Remove unused script * cleanup * fix camelCase in ref test code * cleanup * Fix typo --------- Co-authored-by: brianshi <brianshi@amd.com> * Enable post-loop code generation, and add some subroutines * LR offset calculation (#2) * Add tests * as is * Add permlane16_swap instruction to rocisa * Ongoing progress * Draft for partition A0/A1 * Wave partitioning * Draft ref code in tests * Handle 1x4 wavesplit param * 2x2 test passing * Draft 1x4 LR wave partitioning * Fix alginement issue * Integration testing * Update integration test * Fix swizzling pattern on GRA. Only swizzling on even LDS rows * Subtile based test * testing A * Test both A and B * Remove graonly mode * Fix 1x4 case * Move global offset for B after rest of the logic * cleanup * cleanup * Fix ref test code for 4x1 * Fix spgr alloc issue * Remove tmp test file * Remove debug prints * Add test case * Add GR load emit logic, and misc fixes (#3) * gr emit fix * Emit LR + init ACCVGPR (#4) * Emit ds_reads * Add waits for LR and GR * Init Acc VGPR to Zero * Add missing bit_length on VLShiftLeftB32 * Insert SNop between VLShiftLeftB32 & VReadfirstlaneB32 for correctness * Fix gra test ref code for 1x4 * Remove some debug prints * Add loop and ptr update code * Update scale offset * Add tests * Address review * Add scale roundtrip e2e test and constraint assertions Add GR->LDS->LR roundtrip GPU test verifying scale offset consistency across 4 tile configs x 2 matrices. Add power-of-2 assertion for scaleBlockSize and matching scaleBlockSize assertions for A/B in shared GR/LR offset computation. Pass kernel dict to compute_lds_sizes instead of re-deriving MIWaveGroup from tile dimensions. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Update fixes * Fix scale being skipped * Add flag to print layout * Fix missed merge conflicts * Fix missed merge conflicts * Refactor scale rountrip test with gpu helper fns * Fix extra spaces * Fix tests --------- Co-authored-by: brianshi <brianshi@amd.com> Co-authored-by: sebvince <115461989+sebvince@users.noreply.github.com> Co-authored-by: b-shi <bbbrianme@gmail.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

@brief

- Convert /// comments to /** @brief */ Doxygen format on all 12 serialize/deserialize/to_*/from_* public API methods (#4) - Add namespace closing comment to IncompatibleBackend.hpp (#11) - Add BinarySerializeNullSize test for null graphByteSize pointer (#9) - Add BinaryDeserializeCorruptedData test through C API (#12) - Replace fragile pop_back() null terminator removal with clean resize(graphByteSize - 1) in JSON serialize (#3) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…ernel (ROCm#6) * Add sample subtile impl * Fix issues when disabling subtile impl * GR Offset calculation (#1) * Add sample subtile impl * Move allocOffsetRegisters before setupNewTile * Start adding GR offset calculation * Rest of logic (no swizzling) * refacto * spgr offsets * Add newserial code * Add script to debug offsets * Add unit test for GR offset calculation * Grid display * Fix both code and ref test function * Add DPP quad perm to rocisa * Apply swizzling (no rotation yet) * Function swizzling + rotation + test * Refactor test to have a single output array + add test for SGPRs * Add debug mode to test + add dynamic wavegroup calculation based on MT * Fix test runtime issue and check all vgpr offsets * Add ref test code for 1x4 & 4x1 * Fix tests * Fixed SGPR offset calculation for 2x2 * Fix more tests * Add more tests * Refactor tests * simplify tests * Remove unused script * cleanup * fix camelCase in ref test code * cleanup * Fix typo --------- Co-authored-by: brianshi <brianshi@amd.com> * Enable post-loop code generation, and add some subroutines * LR offset calculation (#2) * Add tests * as is * Add permlane16_swap instruction to rocisa * Ongoing progress * Draft for partition A0/A1 * Wave partitioning * Draft ref code in tests * Handle 1x4 wavesplit param * 2x2 test passing * Draft 1x4 LR wave partitioning * Fix alginement issue * Integration testing * Update integration test * Fix swizzling pattern on GRA. Only swizzling on even LDS rows * Subtile based test * testing A * Test both A and B * Remove graonly mode * Fix 1x4 case * Move global offset for B after rest of the logic * cleanup * cleanup * Fix ref test code for 4x1 * Fix spgr alloc issue * Remove tmp test file * Remove debug prints * Add test case * Add GR load emit logic, and misc fixes (#3) * gr emit fix * Emit LR + init ACCVGPR (#4) * Emit ds_reads * Add waits for LR and GR * Init Acc VGPR to Zero * Add missing bit_length on VLShiftLeftB32 * Insert SNop between VLShiftLeftB32 & VReadfirstlaneB32 for correctness * Fix gra test ref code for 1x4 * Remove some debug prints * Add loop and ptr update code * Update scale offset * Add tests * Address review * Add scale roundtrip e2e test and constraint assertions Add GR->LDS->LR roundtrip GPU test verifying scale offset consistency across 4 tile configs x 2 matrices. Add power-of-2 assertion for scaleBlockSize and matching scaleBlockSize assertions for A/B in shared GR/LR offset computation. Pass kernel dict to compute_lds_sizes instead of re-deriving MIWaveGroup from tile dimensions. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Update fixes * Fix scale being skipped * Add flag to print layout * Fix missed merge conflicts * Fix missed merge conflicts * Refactor scale rountrip test with gpu helper fns * Fix extra spaces * Fix tests --------- Co-authored-by: brianshi <brianshi@amd.com> Co-authored-by: sebvince <115461989+sebvince@users.noreply.github.com> Co-authored-by: b-shi <bbbrianme@gmail.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

#8, stacked on #1, #3] Template solveCPUFastInF32 on MathOpAccumT so the inner reduction applies XFloat32 truncation when MathOpAccumT=XFloat32. Remove the XFloat32 rejection guard from isFastPathEligible. Update SolveGemmCPU dispatch to branch on f32XdlMathOp. When MathOpAccumT=float (default), casts are no-ops with identical codegen to the previous implementation. Changes: - solveCPUFastInF32 gains template<MathOpAccumT=float> - innerReduction uses float(MathOpAccumT(val)) cast chain - isFastPathEligible no longer rejects XFloat32 - SolveGemmCPU dispatches solveCPUFastInF32<XFloat32> for TF32 - 10 new fast-path tf32 tests (20 total tf32 tests now) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…#3, #5, stacked on #1, #2, #5] Generalize the fast path to support double-precision accumulation: - Template ShadowBuffer<AccumT>: storage, pointer, and element access are all AccumT. Float/Double inputs zero-copy when AccumT matches; sub-float types go through float then widen. - Template loadTo<AccumT, SrcType> and storeFrom<AccumT, DstType> (renamed from loadToFloat/storeFromFloat). - Rename solveCPUFastInF32 → solveCPUFast<AccumT, MathOpAccumT>: all tile registers, inner reduction, epilogue, alpha/beta extraction, bias reading, and activation args use AccumT. - Add Double to isFastPathEligible's supported input/output types. - SolveGemmCPU dispatch: route Double to solveCPUFast<double>. - Add 10 f64 fast-path tests (transpose combos, Beta, Bias, AllFeatures, TN_AllFeatures, ScaleAB Scalar/Vector). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

#8, stacked on #1, #3] Template solveCPUFastInF32 on MathOpAccumT so the inner reduction applies XFloat32 truncation when MathOpAccumT=XFloat32. Remove the XFloat32 rejection guard from isFastPathEligible. Update SolveGemmCPU dispatch to branch on f32XdlMathOp. When MathOpAccumT=float (default), casts are no-ops with identical codegen to the previous implementation. Changes: - solveCPUFastInF32 gains template<MathOpAccumT=float> - innerReduction uses float(MathOpAccumT(val)) cast chain - isFastPathEligible no longer rejects XFloat32 - SolveGemmCPU dispatches solveCPUFastInF32<XFloat32> for TF32 - 10 new fast-path tf32 tests (20 total tf32 tests now) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…#3, #5, stacked on #1, #2, #5] Generalize the fast path to support double-precision accumulation: - Template ShadowBuffer<AccumT>: storage, pointer, and element access are all AccumT. Float/Double inputs zero-copy when AccumT matches; sub-float types go through float then widen. - Template loadTo<AccumT, SrcType> and storeFrom<AccumT, DstType> (renamed from loadToFloat/storeFromFloat). - Rename solveCPUFastInF32 → solveCPUFast<AccumT, MathOpAccumT>: all tile registers, inner reduction, epilogue, alpha/beta extraction, bias reading, and activation args use AccumT. - Add Double to isFastPathEligible's supported input/output types. - SolveGemmCPU dispatch: route Double to solveCPUFast<double>. - Add 10 f64 fast-path tests (transpose combos, Beta, Bias, AllFeatures, TN_AllFeatures, ScaleAB Scalar/Vector). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

## Motivation Fork PRs fail CI when `RUN_AITER_TESTS` or `RUN_FA_TESTS` is enabled. The docker scripts run `git clone -b "$CK_*_BRANCH" https://github.com/ROCm/rocm-libraries.git`, but a fork's branch doesn't exist upstream: ``` fatal: Remote branch <fork-branch> not found in upstream origin ``` Example: [PR #6529 build #4](http://micimaster.amd.com/blue/organizations/jenkins/rocm-libraries-folder%2FComposable%20Kernel/detail/PR-6529/4/pipeline). ## Technical Details **`Jenkinsfile`** — for PRs, use the upstream-visible PR ref instead of the head branch name: ```groovy CURRENT_BRANCH_NAME = env.CHANGE_ID ? "refs/pull/${env.CHANGE_ID}/head" : (env.CHANGE_BRANCH ? env.CHANGE_BRANCH : env.BRANCH_NAME) ``` **`Dockerfile.aiter` / `Dockerfile.fa`** — `git clone -b <ref>` only accepts branches (`refs/heads/*`) and tags (`refs/tags/*`), so it can't resolve `refs/pull/N/head`. Switch to `git fetch`, which accepts any refspec (and still works for plain branch names): ```sh mkdir rocm-libraries && cd rocm-libraries git init -q git remote add origin https://github.com/ROCm/rocm-libraries.git git fetch --depth 1 --filter=blob:none origin "$CK_*_BRANCH" git sparse-checkout init --cone git sparse-checkout set projects/composablekernel git checkout FETCH_HEAD ``` `git checkout FETCH_HEAD` lands in detached HEAD, which breaks the existing `git branch -m "$CK_*_BRANCH"` (and that name isn't a valid local branch anyway). Decouple the local branch name from the upstream ref: - Replace `git init` + `git branch -m` with `git init -b "$LOCAL_BRANCH"` (requires git ≥ 2.28, satisfied by base images) - `LOCAL_BRANCH="ck-import-${ROCM_LIBRARIES_SHA}"` in the rocm-libraries path; `LOCAL_BRANCH="$CK_*_BRANCH"` in the fallback - Downstream `git clone -b ... ../ck` uses `$LOCAL_BRANCH` ## Test Plan Manually trigger a build on this PR with `RUN_AITER_TESTS=true` and `RUN_FA_TESTS=true`; both docker images should build end-to-end. ## Test Result [jenkins / rocm-libraries-folder/Composable Kernel / PR-6701 / #3](http://micimaster.amd.com/blue/organizations/jenkins/rocm-libraries-folder%2FComposable%20Kernel/detail/PR-6701/3/pipeline/) ## Submission Checklist - [x] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

# Add gfx950 MXFP4 Subtile-based kernel implementation ## Summary This PR is a follow-up to #6499 ([hipblaslt] Add support for gfx950 mxfp4) and adds the **Subtile-based kernel implementation (`UseSubtileImpl=1`)** for hipBLASLt on **gfx950**. It introduces a new tile-decomposed code generation path optimized for **MXFP4** and **BF16** GEMMs, plus the solution-selection plumbing, validation, Origami logic yamls, and unit tests needed to make it production-usable. ## Motivation PR #6499 brought MX data type support online for gfx950, but the existing TensileLite codegen path leaves significant performance on the table for MXFP4-heavy workloads. The Subtile path restructures global-read / local-read / MFMA / store scheduling at a finer granularity, which **greatly improves MXFP4 GEMM performance when using `HIPBLASLT_MATMUL_MATRIX_SCALE_BLK32_UE8M0_32_8_EXT`** (added to the hipBLASLt CHANGELOG). ## What's included ### 1. New Subtile-based kernel components (Tensile) New modules under `projects/hipblaslt/tensilelite/Tensile/Components/`: * `SubtileBasedKernel.py` (~1850 LOC) — entry point and orchestration of the subtile codegen path; replaces large portions of the standard prefetch / unroll / store flow when `UseSubtileImpl=1`. * `SubtileBasedLogicalScheduler.py` (~2415 LOC) — logical scheduler that builds the subtile-grained instruction graph (GR loads, LR offsets, MFMA tiles, scale loads, stores) from kernel parameters. * `SubtileBasedInstructionScheduler.py` (~433 LOC) — converts the logical schedule to an emit order respecting wave / register / hazard constraints. * `SubtileBasedInstructionEmitter.py` (~216 LOC) — instruction emission helpers shared by the subtile components. ### 2. Kernel writer / common changes * **`KernelWriter.py`**, **`KernelWriterAssembly.py`**: integration points for the subtile path — prefetch, GR offset calculation, LR offset calculation, post-loop, MFMA macro accounting, optimized `storeD`, LDS buffer swap, MX FP4 scale emit, `SrdMXSA/B+2` handling, sgpr allocation / overflow guards, computeLoadSrd fix. * **`SolutionStructs/Solution.py`**, **`SolutionStructs/Problem.py`**: introduces the `UseSubtileImpl` parameter, MX-related reject conditions for non-Subtile paths on gfx950, and additional valid GEMM type combinations for MX inputs. * **`Common/ValidParameters.py`**, **`Common/RequiredParameters.py`**, **`Common/GlobalParameters.py`**: `UseSubtileImpl` registration and defaults. * **`Components/StreamK.py`**: subtile-aware StreamK fixup (incl. import union with the `BufferLoadB32` cache-coherence change from #6837). * **`Components/GlobalWriteBatch.py`**: optimized global write batching for the subtile path (~670 LOC of changes). * **`Components/ComputeStoreVgprs.py`**, **`Components/LSU.py`**, **`Components/WorkGroupMappingAlgos.py`**, **`AsmStoreState.py`**, **`KernelWriterModules.py`**: minor adjustments needed by the subtile pipeline. ### 3. rocisa / host / client * **`rocisa/rocisa/include/container.hpp`**: helpers needed by the new emitter. * **`tensile_host.cpp`**, **`include/Tensile/TensorDescriptor.hpp`**: small fixups for the subtile path and gfx950 build. * **`client/include/DataInitialization.hpp`**, **`client/src/DataInitialization.cpp`**, **`client/src/Reference.cpp`**, **`client/src/ReferenceValidator.cpp`**, **`client/include/TypedId.hpp`**: MX scale init and reference paths used by the new tests. * **`clients/common/include/testing_matmul.hpp`**, **`clients/common/include/norm.hpp`**, **`clients/common/include/hipblaslt_datatype2string.hpp`**, **`clients/common/src/mxDataGen.cpp`**: wiring for batched (>1) testing and MX init. ### 4. Origami / solution selection (gfx950 MXFP4) New auto-tuned logic yamls under `projects/hipblaslt/library/.../Tensile/Logic/asm_full/gfx950/gfx950/Origami/` covering the FP4 SS / HS / BS variants in three layouts: * `Origami/` (default) * `Origami/Origami_nta4/` (no-transpose-A FP4) * `Origami/Origami_ntb4/` (no-transpose-B FP4) (9 new `gfx950_Cijk_Alik_Bljk_F4{SS,HS,BS}_MXA32_MXB32_*_UserArgs.yaml` files in total.) ### 5. New tests **End-to-end gfx950 GEMM yamls** in `Tensile/Tests/common/gemm/gfx950/`: * `subtile_bf16.yaml`, `subtile_mxfp4.yaml` * `mx32f4_tn.yaml`, `mx32f8_tn.yaml` * `mxfp4_mxfp4_{fp32,bf16}_tn_act{,_groupgemm}.yaml` * `mxfp4_fp8_{fp32,bf16}_tn_act{,_groupgemm}.yaml` * `fp8_mxfp4_{fp32,bf16}_tn_act{,_groupgemm}.yaml` **StreamK + MX:** `Tensile/Tests/common/streamk/sk_mx32f4_quick.yaml`, `sk_mx32f8_quick.yaml`. **New unit tests** (`Tensile/Tests/unit/`): * `test_SubtileBasedLogicalScheduler.py` (~1735 LOC) * `test_SubtileBasedSchedulerRef.py` (~596 LOC) * `test_gr_lr_roundtrip.py` (~571 LOC) * `test_storeD_roundtrip.py` (~2420 LOC) * `test_graTileAssignment.py` (~354 LOC) * `test_lraTileAssignment.py` (~360 LOC) * `conftest.py`, `gpu_test_helpers.py` shared fixtures (~601 LOC) **New gtest:** `tensilelite/tests/MXScalePadding_test.cpp`. ### 6. Misc / hardening * Reject conditions: gfx950 MX + non-Subtile, DepthU constraints, GroupGEMM not yet supported with StreamK + MX, AssertSummationElementMultiple=256 for subtile MXFP4, missing-mxblock check for non-MX types. * Skip rocRoller for FP4-A/FP4-B with pre-swizzled scale layout (#42). * `forceDenorm=False` in `generateMXInput` (#11). * Several rebase fixes, copyright/year header updates, and review-comment fixes to `KernelWriter` / `KernelWriterAssembly`. ### 7. CHANGELOG Greatly improved MXFP4 GEMM performance when using HIPBLASLT_MATMUL_MATRIX_SCALE_BLK32_UE8M0_32_8_EXT ## How to use Set `UseSubtileImpl: 1` on a gfx950 MX-FP4 solution (see the new `subtile_mxfp4.yaml` / `mx32f4_tn.yaml` for canonical configs). The path is opt-in — non-MX and non-gfx950 kernels are unaffected. ## Backwards compatibility / risk * All new behavior is gated on `UseSubtileImpl=1` and gfx950. Existing solutions on other architectures or non-MX paths are unchanged. * `GroupGEMM + StreamK + MX` is intentionally rejected for now (TODO). * New Origami yamls only add solutions; nothing existing is modified. ## Test plan * New gtests + unit tests run automatically in CI (Tensilelite Python unit suite, `MXDataGen_test`, `MXScalePadding_test`). * New end-to-end gfx950 GEMM and StreamK yamls are added to the common test buckets. * Manual: run the gfx950 MXFP4 subtile suites (`pytest -k gfx950` after building Tensile, plus `tensilelite-client --yaml subtile_mxfp4.yaml` for sanity). ## Notes for reviewers * This branch was rebased onto current `develop` (post-#6499) by skipping the `users/nakajee/gfx950_mx_rebase_merge` history (which #6499 squash-merged) and replaying only the subtile-specific work as a single squashed commit. The actual code changes in this PR are limited to the files listed above (24 added, 56 modified; ~+170k / −2.6k including generated logic yamls). * The largest reviewable diffs are: * `Tensile/Components/SubtileBased{Kernel,LogicalScheduler,InstructionScheduler,InstructionEmitter}.py` (new files) * `Tensile/KernelWriter.py`, `Tensile/KernelWriterAssembly.py` * `Tensile/SolutionStructs/{Problem,Solution}.py` * `Tensile/Components/{GlobalWriteBatch,StreamK}.py` * `clients/common/include/testing_matmul.hpp` * `client/src/DataInitialization.cpp` * Description of all commits that were squashed for this feature branch: Subtile implementation for gfx950 MX FP4 --- 272f88d: Add sample subtile impl --- Author: brianshi <brianshi@amd.com> --- 60ecede: GR Offset calculation (#1) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> --- be69c1d: Enable post-loop code generation, and add some subroutines --- Author: b-shi <brianshi@amd.com> --- 646d102: LR offset calculation (#2) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> --- 71f4bca: Add GR load emit logic, and misc fixes (#3) --- Author: b-shi <brianshi@amd.com> --- 1fd0db9: Emit LR + init ACCVGPR (#4) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> --- 9d406b9: Add loop and ptr update code --- Author: b-shi <brianshi@amd.com> --- b6127bc: Update GR/LR offset calculation to fully support 2x2, 1x4, 4x1 waveConfigs (#7) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> --- 89ec87c: Account for valuC macro value in SK WS store code --- Author: b-shi <brianshi@amd.com> --- 6edf53d: Rebase fix --- Author: b-shi <brianshi@amd.com> --- 34e79fc: Enable fp4 (#8) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> --- d5a5c57: [Tensilelite] Add MX FP4 scale offset computation for subtile-based kernel (#6) --- Author: Archana Ramalingam <98564406+archana-ramalingam@users.noreply.github.com> --- 7a8a85a: Add lds buffer swap logic --- Author: b-shi <brianshi@amd.com> --- d24a8fe: Add optimized storeD code (#9) --- Author: b-shi <brianshi@amd.com> --- a45c20c: Fix MX scale tensor initialization: set forceDenorm=false in generateMXInput (#11) --- Author: T.J. Alumbaugh <T.J.Alumbaugh@amd.com> --- f945268: [Tensilelite] Enable the MX FP4 scale emit code in the subtile-based kernel (#10) --- Author: Archana Ramalingam <98564406+archana-ramalingam@users.noreply.github.com> --- cf37df4: Use fixed value for SrdMXSA/B+2 (#14) --- Author: Koji Nakajima <75698246+nakajee@users.noreply.github.com> --- f0c8dbc: Merge subtile_mx_f4_schedule to subtile_mx branch (#16) --- Author: b-shi <brianshi@amd.com> --- 543796f: Enable DU > 256, and reduce sgpr allocation (#18) --- Author: b-shi <brianshi@amd.com> --- c65bdb0: Add missing mxblock check for non-mx data types --- Author: b-shi <brianshi@amd.com> --- d64d226: Introduce UseSubtileImpl parameter (#20) --- Author: b-shi <brianshi@amd.com> Squash commits 20-35 from subtile_mx branch --- e4780da: Enable FixSrd2 for A/B (#23) --- Author: b-shi <brianshi@amd.com> * Enable FixSrd2 for A/B * Address comments from PR --------- --- e4c64a7: Add nt libs --- Author: b-shi <brianshi@amd.com> --- cd13ec1: [Tensilelite] Pad MX scale tensor dimensions for unaligned problem sizes (#21) --- Author: Archana Ramalingam <98564406+archana-ramalingam@users.noreply.github.com> * Add scale padding * Add tests * Remove redundant pre-swizzle path * Remove code from conflict * Fix reverted mxdatagen path for tensile tests * Add diverse test cases for scale padding in MXScalePadding_test and subtile.yaml - Expanded test cases to include non-multiple-of-32, even non-multiple-of-16, and odd dimensions. --- d87938f: Split subtile.yaml into subtile_bf16.yaml and subtile_mxfp4.yaml (#22) --- Author: James Newling <james.newling@gmail.com> Replace the 'monolithic' subtile.yaml with two focused test files. All original test coverage is preserved. Two new FP4 groups added. BF16 coverage (subtile_bf16.yaml, tests are essentially unchanged): # | Description | Dest | MIs | PGR | DU | SK | Sizes --+--------------------+------+-----+-----+---------+-----+------ 0 | BF16 TN main | b | 19 | 0 | 64 | 0,3 | 11 1 | BF16 TN large DU | b | 4 | 0 | 128,192 | 0,3 | 7 2 | BSS (f32 output) | s | 6 | 0 | 64 | 0,3 | 9 3 | BF16 bias | b | 2 | 0 | 64 | 0 | 1 FP4 coverage (subtile_mxfp4.yaml): # | Description | Dest | MIs | PGR | DU | SK | Sizes | Status --+--------------------+------+-----+-----+-----+-----+-------+-------- 0 | FP4 TN main | b | 15 | 0 | 256 | 0,3 | 23 | from original 1 | FP4 TN large DU | b | 4 | 0 | 512 | 0,3 | 13 | from original 2 | F4SS (f32 output) | s | 5 | 0 | 256 | 0,3 | 13 | from original 3 | FP4 bias | b | 2 | 0 | 256 | 0 | 1 | from original 4 | FP4 PGR=2 | b | 13 | 2 | 256 | 0 | 5 | new 5 | FP4 expanded MIWT | b | 24 | 0 | 256 | 0 | 5 | new 6 | PGR=2 WG 4x1/1x4 | | 6 | 2 | 256 | 0 | 1 | known failures (commented) Run times on gfx950 (8x MI350X): File | NEV=-1 | NEV=0 -------------------+--------+------ subtile_bf16.yaml | 23s | 23s subtile_mxfp4.yaml | 37s | 40s Where NEV is number of elements to validate. I (James) have checked these numbers, and weirdly it is true that NEV=0 is a bit faster than NEV=-1 for mxfp4. --- af04f0d: Dependency based instruction scheduling (#19) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> * Revert to single partition * Start using dependencies * as is * start using separate EmittedModules * remove reduntant wait * Add _extractPathsFromBeforeDeps * Continue simplification * Simplifying * Add more rules * cleanup * Add fp4 test * fix test * Add tests * Remove after field on emittedmodule * Refactoring instructionSchedule * Add comments * cleanup modules vs ops * Refactoring print functions * Test cleanup * Add more tests * Replace subgroup by partition * Remove unused unroll param * Add high level notes * Simplify NLL and NGLL GR removal * Add some comments * Force instruction insertion if no slots available * Fix test after rebase * Move scale before A/B and track inflight count * Fine-grain vmcnt calculation * Separate counts for scaleA and B * Avoid using m0 update and buffer_lod on same MFMA slot to avoid scalar instruction serialization * Fix test * Add vmcnt test * Fix duplicated loads for 1x4 and 4x1 * Fix placement in reverse order * Fix regression on PGR0 * add fallback to numMFMA=1 --- 3ec902b: Add some 1x4 and 4x1 origami solutions --- Author: b-shi <brianshi@amd.com> --- c5000d3: Fix typo --- Author: b-shi <brianshi@amd.com> --- 226ed84: [hipblaslt] Refactor Srd2 calculation for useFixedSrd2 (#30) --- Author: Koji Nakajima <75698246+nakajee@users.noreply.github.com> --- abf19d4: [Tensilelite] UseSubtileImpl: subtile-aligned edge check for store path (#29) --- Author: b-shi <brianshi@amd.com> * [Tensilelite] UseSubtileImpl: subtile-aligned edge check, OOB guard, and refactoring - Replace Size%MT edge check with subtile-aligned check: NonEdge paired store when trailing rows/cols are a multiple of the subtile block size (waveGroupM rows for M, 16 cols for N). Non-last workgroups always take NonEdge. - Add per-wave OOB guard (subtileM32ValidBlocksSgpr / subtileN16ValidBlocksSgpr) to skip stores outside valid M/N tile bounds in the NonEdge path. - Refactor duplicated OOB guard into _emitSubtileOobGuard helper; refactor M/N guard SGPR computation into _emitSubtileMGuard / _emitSubtileNGuard. - Fix orphan scalar store blockIdxM (was tt0, now (tt0*MatrixInstM)//mBlockSize). - Add quick-exit and edge/non-edge header comments to generated ASM. * Add some bias tests, combine M/N guard to single routine * Add OOB check for C loads, update storeD unit tests to check OOB, simplify quick exit checks * Address more PR comments: add M group skip, and skip to store end. simplified loadC OOB mask --------- --- 637881a: Fix unit tests & remove legacy code for subtile interleaving (#33) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> * Fix gr_lr_roundtrip test * Use non-interleaved version as ref code * Fix scheduler test * Removed legacy interleaved mode for LR/GR offset calculation --- e9cb889: Fix MX FP4 scale buffer allocation and initialization for batched GEMM (#25) --- Author: Archana Ramalingam <98564406+archana-ramalingam@users.noreply.github.com> * Fix bacth count issue * Add batch count tests * Fix bacth count issue * Address PR review: clarify FP4-specific byte stride and add non-aligned batched tests - Updated comments on dataBatchBytes computation to clarify FP4 packing assumption (2 elements/byte) and flag that non-FP4 block-scaling types would require updating this conversion. - Added batched test cases with non-multiple-of-32 M/N dimensions: FP4 DU=256: [48,48,2] and [33,65,2] FP4 DU=512: [63,63,2] BF16: [50,100,2] --------- --- a43247b: Update some test yamls (#31) --- Author: b-shi <brianshi@amd.com> --- e2f69c8: Add f4bs origami library with activation function support. Refactor sgpr allocation to reduce sgpr usage in post loop. Store code-path reorganization (#32) --- Author: b-shi <brianshi@amd.com> * Free swap/localwritebase sgprs before post-loop * Defer sgpr allocation to remove holds in sgpr pool. Add Origami library logic files for Cijk_Alik_Bljk_F4BS_MXA32_MXB32 (base, nta4, ntb4 variants). * Remove uneeded alignment and comment * Add more epilogue tests * Remove older origami library for f4bs * Reorder post-loop code blocks to after persistant loop Misc fixes * Fix build issues, relax longjump sgpr requirements * Fix GSU0 branch logic --------- --- 3f034bf: Add F4HS and F4SS Origami library logic for FP4→F16 and FP4→F32 GEMM (#35) --- Author: Majedul Sujon <85503863+msujon-AMD@users.noreply.github.com> * Add F4HS and F4SS Origami library logic for FP4→F16 and FP4→F32 GEMM - Add 6 new yaml files (F4HS, F4SS) across Origami, Origami_nta4, Origami_ntb4 - Update F4BS yaml files: AssertSummationElementMultiple 32→256 for K%256 enforcement - Add ("F4", "F4", "H", "S") to _validGEMMTypes and _HPATypes in Problem.py * Add F4HS test cases to subtile_mxfp4.yaml Add two new benchmark problem blocks for FP4→F16 (F4HS): - No-bias block: same wavetile and problem size coverage as F4SS - Bias epilogue block: BiasDataTypeList [s, h], relu/none activations * Add F4HS (FP4->Half) type support to Tensile client Add TypedGemm_F4_H_S typedef and corresponding reference CPU solver case so F4HS (FP4 input, Float16 output, Float compute) problems can be validated by the benchmark client. --------- --- d0bc8fd: Rewrite subtile-based scheduler. Fix DU>64 & enable very large MT (#36) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> * Initial support for DU>256 * Renaming * add option to do DU=512 in the tests * blocked K-major for scale * Change scaleSet swap logic * Update print functions * Put scales after values for avoid race conditions * Fix tests * more test * tweak printschedule display * Add PGR2 in the yaml tests * Add new scaleGROp * comment out failing tests * Revert "comment out failing tests" This reverts commit 1f5802c. * Draft new logical scheduler * Refactoring * Add more test on step1 * Add more tests on step1 * add bf16 320x320 test * reduce step1 code * Simplify step1 logic * validate some step1 test * Fix partition 2x2 test * more step1 test * 320x320 BF16 test * Add test DU512 + partition2x2 * Simplify step1 code * Add step2 tests * Fix multi-partition step2 * Add step2 du512, 2x2 partition test * Use common algo for all numPartitions * Draft for step3 tests * remove useless tests * New GR algo (draft) * [Step3] Add more test * Iteration on GR * Display ordered GR list with granularities * More test * Add some comments * Disable by default debug logs * Getting rid of step naming * Start remove AnnotatedOp (still there in group pass) * Split dependency Ops * Add todo on place_GRs pass * Valid test_annotate_deps_1x1_partition_DU256 * Test output looking better (still WIP) * single dep for LR tooo * Add remove_cross_deps pass * Fix bugs in dependency pass * insert_gr_lr_inc pass * Add group_lr_gr pass * Add emit pass * Quick port of instruction Emit code * Move emit function to separate file * Refactoring instructionEmitter * Port vgprTile tracking * Reworking second pass (WIP) * Display unrolling requirement * Unrolling check on 2nd pass * Generic validation for assign_vgpr pass * Fix unroll * Add inst schedule in standalone mode * Use lrGran for vgprTile size calculation * Fix bug in emit pass (missing depencency) * PreMFMA path + non-duplication scale load * missing globalReadLDSBufferSwap for GR_INC scales * add wairlr_sync on all LR->GR dep * add waitgr_sync op * remove_unnecessary_gr_deps * Change LR dispatch algo a bit to avoid too many waitgr_sync * Avoid duplicated loads in emitter * Fix bug on gr_emit code * GrInc pass. fix duplicated insertion for B * Fix missing LR_inc for SA/SB * preloop, NLL, NGLL * Simplify preloop * minor changes * Move unroll logic to scheduler * minor changes * Fix unroll id bug on NLL / NGLL * Disable post GRINC for now * Remove commented code * Handle 1x4, 4x1 gr read gran * Fix vmcnt computation * Use correct grCount mapping * Revert in emit logic on buffer_load for PGR0 needs * Add bf16 version in standalone test * Fix LR_Inc insertion on DU>64 * Add subIterK/Partition comment to codegen * Fix issue in GrInc placement * Remove last_mt * Fix LR MT index bug with muli-partition * Disable early LDS size check when subtileImpl is on * Add pass to remove redundant LR deps + fixed issue on dependency annotation pass * Remove more LR redundant deps * Only insert wait_lr_sync on deps * Simple algo to select partition config * Remove HC value for partitions... * Take into account all inflight GR (all tensors) * Fix tests and regressions on gr counts * Fix grCount merge calculation * Better display of dependencies * Add remove_wait_lr_sync after grouping * Add temporary non reg file * Change merge logic on GR grouping pass * Fix non necessary wait_lr_sync * Downgrade some waitlr_sync to sync + added 384x256 no reg test * non reg test 320x320 * Add larger MT * non reg test for fp4 256x256 * Moving out instructionScheduler * Remove old scheduler * Renaming scheduler * Re-work test * Add larger MT test cases * Rename non-ref test * Re-add standalone mode * Refactor DepOp * Remove dead code * Remove MFMATileSize class * Remove from_til_info * Avoid redundant tensor list creation * Remove hardcode granularities in vgrpTile allocation pass. Simplify code. * Re-enable # PGR=2 WG 4x1/1x4, K > DU tests * Remove unused GRScaleOp * DepRef renaming * Get rid of MT string representation * Remove TODO * EmmitedModule simplication * Use explicit pass dependencies * Renaming LogicalScheduler * Remove old test_InterleavingScheduler.py file * Commenting failing test for now * Remove debug logs * Disable lds padding when using UseSubtileImpl --- e8e8c09: Fix LR-GR dependency issue when DU>64 (#40) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> * Fix and simplify logic for remove_unnecessary_lr_deps * Add new ref tests for 128x128x(128,64) --- 4aa441a: Rebase fix --- Author: b-shi <brianshi@amd.com> --- 5ba911e: Skip rocRoller for FP4-A/FP4-B + pre-swizzled scale layout (#42) --- Author: Archana Ramalingam <98564406+archana-ramalingam@users.noreply.github.com> --- 9c74998: Rebase fix --- Author: b-shi <brianshi@amd.com> --- 842b149: Addressed review comments for KernelWriter and KernelWriterAssembly --- Author: Koji Nakajima <knakajim@amd.com> --- dce43b1: Fix computeLoadSrd issue --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- c075bbf: Fix preSolution CPU re-sync regressing subtile_mxfp4.yaml --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- ced840f: Fix computeLoadSrd issue (#43) --- Author: bnemanich <brad.nemanich@amd.com> --- bc2f6dd: Small update for gfx950 mx tests + more - enable UseSubtileImpl for all gfx950 non subtile mx tests - skip all gfx950 mxfp8 - use MXScaleFormat=1 as default - set AssertSummationElementMultiple=256 for subtile mxfp4 - fix isSwizzledSubtile in computeLoadSrd --- Author: Koji Nakajima <knakajim@amd.com> --- 5c794b7: Fix gsuasb.yaml failures --- Author: b-shi <brianshi@amd.com> --- 727f8db: tensilelite: add solution reject conditions for UseSubtileImpl=1 (#38) --- Author: Majedul Sujon <85503863+msujon-AMD@users.noreply.github.com> --- 8928fbb: Add more reject conditions for Subtile --- Author: Koji Nakajima <knakajim@amd.com> --- 6e63ab6: Fix kringshift test failures --- Author: b-shi <brianshi@amd.com> --- b3e9724: Update reject condtion for DepthU in subtile case. Plus, update DepthU setting for gfx950 mx test cases --- Author: Koji Nakajima <knakajim@amd.com> --- 5ab6009: Fix build errors --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- 4a4edca: Update more mxfp4 tensilelite test cases --- Author: Koji Nakajima <knakajim@amd.com> --- bbbc553: Update change log --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- 6476c04: Add more reject conditions for gfx950 subtile --- Author: Koji Nakajima <knakajim@amd.com> --- c5828c4: Updated gfx950 mxfp4 test cases - add StreamK setting - skip groupgemm tests for now (groupgemm does not support streamK) --- Author: Koji Nakajima <knakajim@amd.com> --- f1fc2f1: Fix hipblaslt build error of gfx950 --- Author: Koji Nakajima <knakajim@amd.com> --- 70cea1b: Updated subtile_mxfp4.yaml (add StreamK) --- Author: Koji Nakajima <knakajim@amd.com> --- c1c9b2a: Add uninit lsc,lsp, etc.. fields for subtile --- Author: b-shi <brianshi@amd.com> --- c0c1f72: Fixed merge error in testing_matmul.hpp --- Author: Koji Nakajima <knakajim@amd.com> --- 191e0cb: Add missed batch_count >1 changes --- Author: archana-ramalingam <Archana.Ramalingam@amd.com> --- 01c52f8: Addressed PR comments --- Author: Koji Nakajima <knakajim@amd.com> --- 4e89c91: Reduce mxfp4 test time --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- 3dac20f: Prevent overflow for wgmxcc sgpr allocation --- Author: b-shi <brianshi@amd.com> --- 18dec79: Fix error with problem type --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- 9e69ffd: Add a reject conditoin for gfx950 mx + non Subtile --- Author: Koji Nakajima <knakajim@amd.com> --- 0eed3ba: Add more valid GEMM types --- Author: Brad Nemanich <brad.nemanich@amd.com> --- 8b5514e: Fix missing b build error --- Author: archana-ramalingam <Archana.Ramalingam@amd.com> --- f981ff5: Fix 1250 tests --- Author: Brad Nemanich <brad.nemanich@amd.com> --- d1e69d9: Add more FP4 tests --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- e3a688f: Add MXScaleFormat: 1 to all gfx950 mx test yaml --- Author: Koji Nakajima <knakajim@amd.com> --- aaef3f5: Add DataTypeMXSA,B setting in gfx950 mxfp4 logic yaml --- Author: Koji Nakajima <knakajim@amd.com> --- 861ef8e: Add DataTypeMXSA,B setting in gfx950 mxfp4 logic yaml (nta4,ntb4) --- Author: Koji Nakajima <knakajim@amd.com> Co-authored-by: Archana Ramalingam <Archana.Ramalingam@amd.com> Co-authored-by: Brad Nemanich <Brad.Nemanich@amd.com> Co-authored-by: Brian Shi <Brian.Shi@amd.com> Co-authored-by: James Newling <James.Newling@amd.com> Co-authored-by: Koji Nakajima <Koji.Nakajima@amd.com> Co-authored-by: Majedul Sujon <Majed.Sujon@amd.com> Co-authored-by: Sebastien Vince <Sebastien.Vince@amd.com> Co-authored-by: T.J. Alumbaugh <T.J.Alumbaugh@amd.com> ## Submission Checklist - [ ] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests. --------- Co-authored-by: Archana Ramalingam <Archana.Ramalingam@amd.com> Co-authored-by: Brad Nemanich <Brad.Nemanich@amd.com> Co-authored-by: Brian Shi <Brian.Shi@amd.com> Co-authored-by: James Newling <James.Newling@amd.com> Co-authored-by: Koji Nakajima <Koji.Nakajima@amd.com> Co-authored-by: Majedul Sujon <Majed.Sujon@amd.com> Co-authored-by: Sebastien Vince <Sebastien.Vince@amd.com> Co-authored-by: T.J. Alumbaugh <T.J.Alumbaugh@amd.com>

…rows in 8t9 audit DZL_SCOPE_REASSESSMENT.md §1.1 SCC: adds source-level confirmation that CommonInstruction (parent of every SCC-touching SOPx class) has only dst/dst1/srcs fields, with getDstParams()/getSrcParams() bodies shown verbatim from instruction.hpp:382. Rules out the q9j-style "binding gap" hypothesis at the source level, not just by Python dir() probing. ROCISA_WORKAROUNDS_AUDIT.md: adds a "Status update (2026-05-07)" preamble that flags which rows have been resolved or reframed by subsequent investigations: - Row #3 (_OPERAND_RULES, ~600 LoC estimate) → reassessed by q9j reassessment to ~12 lines C++ binding + ~260 LoC validator migration (the C++ API already exists). - Rows #7, #8, #19 (VCC machinery) → REMOVED by uraq; no rocisa-side replacement planned. VCC dataflow tracking is permanently dropped from the validator's scope. - MFMA acc special-casing → RECLASSIFIED to q9j Category A (acc and acc2 are already in the C++ getDstParams/getSrcParams partition). - Category C narrowed from SCC/VCC/m0/acc to SCC/m0 only. The audit's table and prose are preserved as the original artifact; the preamble lets a reader find the current canonical reframing for each affected row without rewriting the historical record.

…ty and CLI Data Integrity checks #1-#3 tagged as *proposed* — none are implemented today. CLI flags marked as not yet existing. Prevents readers from confusing proposed behavior with current state. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…7008) ## Motivation Several memory leaks were detected in MIOpen gtests using ASan. Some of the tests were blacklisted and others were not. This change looks to fix all of the low hanging fruit, which are the majority of the leaks found. This includes all of the critical leaks (>100MB) that were reported. Some other leaks were identified as needing a larger refactor to resolve. After fixing the w_supertensor.cpp leak, the supertensor tests hit virtual memory area limit errors. These changes are to enable running them with ASan, but only with a subset of the tests in order to not hit limits. The test coverage being lost is fairly negligible and the full tests are still run without ASan enabled. ## Technical Details Here is a summary of the files looked at and the changes made: | File | Status | Notes | |------|--------|-------| | `test/gtest/gpu_mha_forward.cpp` | Fixed | Added `miopenDestroySolution()` loop after using solutions. Fixes MHA forward solution descriptor leaks (report #11, #13). | | `test/gtest/gpu_mha_backward.cpp` | Fixed | Added `miopenDestroySolution()` loop after using solutions. Fixes MHA backward solution descriptor leaks (report #9, #12). | | `test/gtest/mha_find20.cpp` | Fixed | Added `miopenDestroySolution()` loop in both `MhaForward` and `MhaBackward` tests. Fixes MHA Find2.0 solution leak (report #16). | | `test/gtest/gtest_desc_guard.hpp` | Fixed | New shared header introduced by the refactor. Provides a single `DescGuard<DescType, CreateFn, DestroyFn>` template (with `TensorDescGuard`, `ConvDescGuard`, `DropoutDescGuard`, `RNNDescGuard` aliases), a `HandleGuard` RAII wrapper for `miopenHandle_t`, and the `DestroyInternalRnnDropoutDesc(rnnDesc)` helper used by every RNN/LSTM/GRU test to free the internal `DropoutDescriptor` that `miopenCreateRNNDescriptor` allocates and `miopenSetRNNDescriptor*` then leaks. Replaces the per-file ad-hoc guard structs from the initial implementation. | | `test/gtest/w_supertensor.cpp` | Fixed | Switched raw descriptors to the shared `RNNDescGuard` / `TensorDescGuard` from `gtest_desc_guard.hpp`. Added a class-local `DestroyDropoutDesc()` (called from `TearDown` and before `miopenSetRNNDescriptor`) to prevent the `miopenSetRNNDescriptor` overwrite leak. Reduced test parameter space under ASan to avoid OOM. Removed unused `seqLen` parameter and dead `param_dev_out`/`bias_dev_out` allocations. | | `test/gtest/lstm.hpp` | Fixed | Switched `rnnDesc` → `RNNDescGuard`, `DropoutDesc` → `DropoutDescGuard`, and `mio_handle` → `HandleGuard` (which now owns the `miopenDestroy` call), all from the shared `gtest_desc_guard.hpp`. Hoisted `dropout_state_buf` so it can be `hipFree`d at the end of the dropout path. Added `DestroyInternalRnnDropoutDesc(rnnDesc)` before `miopenSetRNNDescriptor*` and (in the non-dropout path only) at end of `Run`, which frees the internal `DropoutDescriptor` that the Set call would otherwise leak. Fixes ~615 MB dropout leaks and ~16.5 KB non-dropout descriptor leaks (report #3, #4, #6, #17-#22). | | `test/gtest/gru_test.cpp` | Fixed | In the GRU test class: switched `rnnDesc` → `RNNDescGuard`, `DropoutDesc` → `DropoutDescGuard`, `mio_handle` → `HandleGuard`, and added `DestroyInternalRnnDropoutDesc(rnnDesc)` before `miopenSetRNNDescriptor*` and (non-dropout path only) at end of `Run` plus `hipFree(dropout_state_buf)` for the dropout path. In the in-file `GRUFwdCPUVerify` / `GRUBwdDataCPUVerify` helpers: converted the raw `dropout_inputTensor` / `dropout_outputTensor` declarations to `TensorDescGuard` (mirroring the `cpu_rnn.hpp` change for the LSTM/RNN helpers). | | `test/gtest/softmax_find20.cpp` | Fixed | Changed `Finalize()` to take the `std::vector<miopenSolution_t>&` and destroy each solution via `miopenDestroySolution()` before destroying the problem. Updated all 6 `TEST(...)` callers to pass the solutions vector. Fixes Find2.0 softmax solution/kernel leaks (report #25-#27). | | `test/gtest/rnn_seq_api.hpp` | Fixed | Hoisted `dropout_state_buf` so the dropout path can `hipFree` it at the end. Added `DestroyInternalRnnDropoutDesc(&rnnDesc)` before `miopenSetRNNDescriptor*` and (non-dropout path only) at end of the run to free the internal `DropoutDescriptor` allocations leaked by `RNNDescriptor` copy-assignment. Same pattern as the LSTM/GRU fixes (report #29-#30). | | `test/cpu_rnn.hpp` | Fixed | Converted 6 raw `miopenTensorDescriptor_t dropout_input/outputTensor` declarations across the LSTM/RNN CPU verification helpers (`LSTMFwdCPUVerify`, `LSTMBwdDataCPUVerify`, `RNNFwdTrainCPUVerify`, `RNNBwdDataCPUVerify`, `GRUFwdCPUVerify`, `GRUBwdDataCPUVerify`) to the shared `TensorDescGuard` from `gtest_desc_guard.hpp`. Removed the redundant `miopenCreateTensorDescriptor` calls and updated 12 `miopen::deref(...)` sites to `.get()`. (Note: the GRU helpers in this header are stale duplicates; the live ones are inside `test/gtest/gru_test.cpp` and were updated there too.) Fixes the LSTM/GRU CPU-verify tensor descriptor leaks (report #1, #2, #14, #15, #32, #33). | | `test/gtest/rnn_vanilla_common.hpp` | Fixed | Added `DestroyInternalRnnDropoutDesc(rnnDesc)` calls before `miopenSetRNNDescriptor*` and (non-dropout path only) at end of the run. The `RNNDescGuard` / `DropoutDescGuard` usage was already in place from an earlier commit and now resolves to the shared definitions in `gtest_desc_guard.hpp`. Same pattern as LSTM/GRU/rnn_seq_api fixes (report #14, #15, #32, #33). | | `test/gtest/graphapi_gtest_common.hpp` | Skipped | File no longer exists in the codebase. The GraphApi test infrastructure has been removed; only the leak report (a stale snapshot) still references it. No fix possible against the current source tree. | | `test/gtest/graphapi_execution_plan.cpp` | Skipped | File no longer exists in the codebase (GraphApi removed). The leak it represented was largely an external hipblaslt bug anyway; the remaining test-side portion is not fixable against the current source. | | `test/gtest/na_train.cpp` / `na_inference.cpp` / `na_*_find2.cpp` | Skipped | Leaks come from the internal MIOpen solver/kernel cache living on the global singleton handle, which is never destroyed. Not easily fixable without redesigning global handle lifecycle. Should be suppressed in the ASan suppression file. | | hipblaslt / rocblaslt (external) | Skipped | `SolutionCache::addKernel` and `preloadCustomKernels` leak via `_rocblaslt_handle` constructor. Called from `miopen::Handle::CreateHipblasLtHandle`. This is an upstream bug in hipblaslt/rocblaslt, not fixable in MIOpen. Affects suites that initialize a handle (report Category 2, Category 7). | | CLR / HIP runtime (external) | Skipped | `amd::Context` and `amd::roc::Device` global initialization leaks from `rocclr/platform/context.cpp`. HIP runtime internals, not fixable in MIOpen. | | `src/hipoc/hipoc_program.cpp` | Skipped | `HIPOCProgramImpl` objects leak during kernel compilation/caching (line ~178). This is an internal MIOpen kernel cache lifecycle issue that requires deeper architectural changes to fix. Contributes small amounts to MHA and Softmax Find2.0 leaks. | | `test/gtest/conv_api.cpp` | Fixed | Already clean against current source — `miopenDestroyConvolutionDescriptor(conv_desc)` call exists (line 24) inside the test loop. ASAN run reports no leaks. The leak report was based on a stale snapshot. The hipblaslt handle init portion is tracked under the external-skipped row. | | `test/gtest/log_test.cpp` (CPU_LOG_TEST_FUSION / CPU_LOG_TEST_NEG) | Fixed | Already clean against current source — `Tensor`, `Conv`, `CreateCBAFusionPlan`, `CreateBNormFusionPlan` all have proper destructors that call the corresponding `miopenDestroy*` APIs in `log.cpp`. ASAN run on `CPU_LOG_TEST_*` (11 tests across log_test.cpp + log_test_neg.cpp) reports no leaks. The hipblaslt handle init portion is tracked separately under the external-skipped row. | | `test/gtest/fusion_test.cpp` (CPU_FusionCreateOpConvForward) | Fixed | File renamed from `fusion.cpp` to `fusion_test.cpp`. Already clean against current source — uses `TensorDescGuard`/`ConvDescGuard` for tensor/conv descriptors and calls `miopenDestroyFusionPlan(fusionPlanDesc)` on the fusion plan (line 195). ASAN run on `CPU_FusionCreateOpConvForward_FP32.*` reports no leaks. | | `test/gtest/deterministic_conv_api.cpp` | Fixed | Already clean against current source — uses `ConvDescGuard` (line 66) for the conv descriptor. ASAN run on `*CPU_DeterministicConvApi*` reports no leaks. | | `test/gtest/fusion_aux.cpp` (GPU_FusionAux) | Fixed | Already clean against current source — uses `ConvDescGuard` plus stack-allocated internal C++ objects (`miopen::TensorDescriptor`, `miopen::FusionPlanDescriptor`) which have proper destructors. The `convoOp` handle is owned by the fusion plan. ASAN run on `*GPU_FusionAux*` reports no leaks. | | `test/gtest/backend_api.cpp` (CPU_BackendApi) | Skipped | File no longer exists in the codebase. The backend API test infrastructure (part of the removed GraphApi suite) was removed; no fix possible against the current source tree. | ### High-level notes New shared infrastructure (test/gtest/gtest_desc_guard.hpp) - DescGuard<DescType, CreateFn, DestroyFn> — a single RAII template parameterized on the descriptor type and its create/destroy entry points. Aliases provide TensorDescGuard, ConvDescGuard, DropoutDescGuard, and RNNDescGuard, replacing the four near-identical guard structs that were copy-pasted across test files in the initial implementation. - HandleGuard — separate RAII wrapper for miopenHandle_t (couldn't reuse the template because miopenCreateWithStream takes an extra hipStream_t argument). Supports lazy create(stream) so callers that only need a handle in the dropout branch can default-construct one and populate it conditionally. - DestroyInternalRnnDropoutDesc(rnnDesc) — frees the internal DropoutDescriptor that miopenCreateRNNDescriptor allocates and that miopenSetRNNDescriptor* then orphans. Replaces the equivalent inline blocks that LSTM/GRU/RNN tests were each carrying. The header documents the two call-sites: before each Set* (always safe) and at end-of-run only on the non-dropout path (the dropout path aliases the user-owned descriptor, so freeing would double-free). Recurring patterns enabled by the refactor - The "leak from Set* overwriting the default-constructed internal dropout descriptor" fix collapsed from per-file code to a one-line helper call, applied uniformly across lstm.hpp, gru_test.cpp, rnn_seq_api.hpp, and rnn_vanilla_common.hpp. - mio_handle ownership in LSTM/GRU is now expressed via HandleGuard rather than a manual miopenDestroy at the end of the dropout branch — eliminates a class of forgotten-cleanup bugs. - dropout_state_buf is consistently hoisted out of the dropout if block so an end-of-run hipFree can release it; deletion of the buffer pairs visibly with its allocation. Notable non-RNN change - softmax_find20.cpp was the only Find2.0 leak fix in this commit: Finalize() now takes the solutions vector and calls miopenDestroySolution() for each before destroying the problem. Same shape applied to all 6 tests in the file. ## Test Plan Run the tests beforehand to observe the ASan leak errors and then again afterward to verify the fixes have resolved the problem. ## Test Result List from ROCM-21512: | # | Test Name | Status | Leak Status | |---:|------------------------------------------------------|------------------------------------------------|-----------------| | 1 | Smoke/GPU_RNNVanillaDropout_FP32 | PASSED (4 tests) | No leaks | | 2 | Smoke/GPU_RNNVanillaDropout_FP16 | PASSED (4 tests) | No leaks | | 3 | Full/GPU_LSTM_dropout_FP32 | PASSED (4 tests) | No leaks | | 4 | Full/GPU_LSTM_dropout_FP16 | PASSED (4 tests) | No leaks | | 5 | CPU_GraphApiExecutionPlanBuilder_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 6 | Full/GPU_LSTM_dropout_FP64 | REMOVED (PR #5750, 2026-03-26) | n/a (deleted) | | 7 | Unit/CPU_GraphApiPointwise_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 8 | Full/GPU_LstmMSRnn_FP32 | PASSED (1152 tests) | No leaks | | 9 | Smoke/GPU_Bwd_Mha_FP32 | PASSED (12 tests) | No leaks | | 10 | Full/GPU_LstmMSRnn_FP16 | PASSED (864 tests) | No leaks | | 11 | Smoke/GPU_Fwd_Mha_FP32 | PASSED (15 tests) | No leaks | | 12 | Full/GPU_Bwd_Mha_FP32 | PASSED (6 tests) | No leaks | | 13 | Full/GPU_Fwd_Mha_FP32 | PASSED (7 tests) | No leaks | | 14 | Full/GPU_RNNVanilla_FP32 | PASSED (96 tests) | No leaks | | 15 | Full/GPU_RNNVanilla_FP16 | PASSED (96 tests) | No leaks | | 16 | GPU_TestMhaFind20_FP32 | PASSED (2 tests) | No leaks | | 17 | Full/GPU_LSTM_FP32 | PASSED (32 tests) | No leaks | | 18 | Full/GPU_LSTM_FP16 | PASSED (32 tests) | No leaks | | 19 | Full/GPU_LSTM_extra_FP32 | PASSED (30 tests) | No leaks | | 20 | Full/GPU_LSTM_extra_FP16 | PASSED (30 tests) | No leaks | | 21 | Full/GPU_DeepBench_LSTM_FP16 | PASSED (22 tests) | No leaks | | 22 | Full/GPU_DeepBench_LSTM_FP32 | PASSED (22 tests) | No leaks | | 23 | CPU_LOG_TEST_FUSION_NONE | PASSED (2 tests) | No leaks | | 24 | CPU_LOG_TEST_NEG_NONE | PASSED (4 tests) | No leaks | | 25 | GPU_SoftmaxFind20_BFP16 | PASSED (2 tests) | No leaks | | 26 | GPU_SoftmaxFind20_FP16 | PASSED (2 tests) | No leaks | | 27 | GPU_SoftmaxFind20_FP32 | PASSED (2 tests) | No leaks | | 28 | CPU_ConvApi_NONE | PASSED (1 test) | No leaks | | 29 | Full/GPU_RNNSeqApi_FP16 | PASSED (16 tests) | No leaks | | 30 | Full/GPU_RNNSeqApi_FP32 | PASSED (16 tests) | No leaks | | 31 | UnitVAN/CPU_GraphApiRng_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 32 | Smoke/GPU_RNNVanilla_FP16 | PASSED (4 tests) | No leaks | | 33 | Smoke/GPU_RNNVanilla_FP32 | PASSED (4 tests) | No leaks | | 34 | CPU_FusionCreateOpConvForward_FP32 | PASSED (1 test) | No leaks | | 35 | CPU_GraphApiOperationReduction_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 36 | Unit2IV1/CPU_GraphApiOperationPointwiseBwd_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 37 | Unit2IV1/CPU_GraphApiOperationPointwiseTwoInput | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 38 | Unit2IV2/CPU_GraphApiOperationPointwiseBwd_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 39 | Unit2IV2/CPU_GraphApiOperationPointwiseTwoInput | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 40 | UnitVAB/CPU_GraphApiRng_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 41 | Smoke/GPU_RNNVanillaDropout_FP16 (duplicate of #2) | (see #2) | (see #2) | | 42 | CPU_GraphApiOperationGraphDescriptor_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 43 | UnitVA/CPU_GraphApiVariantPack_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 44 | UnitVAU/CPU_GraphApiRng_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 45 | CPU_GraphApiOperationReshape_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 46 | Smoke/CPU_DeterministicConvApi_NONE | PASSED (1 test) | No leaks | | 47 | Smoke/GPU_FusionAux_FP32 | PASSED (1 test) | No leaks | | 48 | CPU_GraphApiEngineHeur_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 49 | Unit/CPU_GraphApiReduction_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 50 | CPU_GraphApiEngineCfg_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 51 | Unit/CPU_GraphApiMatMul_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 52 | CPU_BackendApi_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 53 | UnitIV/CPU_GraphApiOperationPointwiseOneInput | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 54 | Unit3IV/CPU_GraphApiOperationPointwiseThreeInput | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 55 | UnitVA/CPU_GraphApiOperationMatmul_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 56 | UnitVA/CPU_GraphApiOperationRng_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | Outcome | Count | |------------------------------------|------:| | Passed, no leaks | 39 | | Passed, leaks detected | 0 | | Failed | 0 | | Crashed / timed out | 0 | | Removed — GraphAPI purge (#5603) | 15 | | Removed — FP64 LSTM purge (#5750) | 1 | | Duplicate (not re-run) | 1 | | **Total rows** | **56** | ## Risk Assessment Low --------- Co-authored-by: JonathanLichtnerAMD <195780826+JonathanLichtnerAMD@users.noreply.github.com>

## Motivation Fork PRs fail CI when `RUN_AITER_TESTS` or `RUN_FA_TESTS` is enabled. The docker scripts run `git clone -b "$CK_*_BRANCH" https://github.com/ROCm/rocm-libraries.git`, but a fork's branch doesn't exist upstream: ``` fatal: Remote branch <fork-branch> not found in upstream origin ``` Example: [PR #6529 build #4](http://micimaster.amd.com/blue/organizations/jenkins/rocm-libraries-folder%2FComposable%20Kernel/detail/PR-6529/4/pipeline). ## Technical Details **`Jenkinsfile`** — for PRs, use the upstream-visible PR ref instead of the head branch name: ```groovy CURRENT_BRANCH_NAME = env.CHANGE_ID ? "refs/pull/${env.CHANGE_ID}/head" : (env.CHANGE_BRANCH ? env.CHANGE_BRANCH : env.BRANCH_NAME) ``` **`Dockerfile.aiter` / `Dockerfile.fa`** — `git clone -b <ref>` only accepts branches (`refs/heads/*`) and tags (`refs/tags/*`), so it can't resolve `refs/pull/N/head`. Switch to `git fetch`, which accepts any refspec (and still works for plain branch names): ```sh mkdir rocm-libraries && cd rocm-libraries git init -q git remote add origin https://github.com/ROCm/rocm-libraries.git git fetch --depth 1 --filter=blob:none origin "$CK_*_BRANCH" git sparse-checkout init --cone git sparse-checkout set projects/composablekernel git checkout FETCH_HEAD ``` `git checkout FETCH_HEAD` lands in detached HEAD, which breaks the existing `git branch -m "$CK_*_BRANCH"` (and that name isn't a valid local branch anyway). Decouple the local branch name from the upstream ref: - Replace `git init` + `git branch -m` with `git init -b "$LOCAL_BRANCH"` (requires git ≥ 2.28, satisfied by base images) - `LOCAL_BRANCH="ck-import-${ROCM_LIBRARIES_SHA}"` in the rocm-libraries path; `LOCAL_BRANCH="$CK_*_BRANCH"` in the fallback - Downstream `git clone -b ... ../ck` uses `$LOCAL_BRANCH` ## Test Plan Manually trigger a build on this PR with `RUN_AITER_TESTS=true` and `RUN_FA_TESTS=true`; both docker images should build end-to-end. ## Test Result [jenkins / rocm-libraries-folder/Composable Kernel / PR-6701 / #3](http://micimaster.amd.com/blue/organizations/jenkins/rocm-libraries-folder%2FComposable%20Kernel/detail/PR-6701/3/pipeline/) ## Submission Checklist - [x] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

# Add gfx950 MXFP4 Subtile-based kernel implementation ## Summary This PR is a follow-up to #6499 ([hipblaslt] Add support for gfx950 mxfp4) and adds the **Subtile-based kernel implementation (`UseSubtileImpl=1`)** for hipBLASLt on **gfx950**. It introduces a new tile-decomposed code generation path optimized for **MXFP4** and **BF16** GEMMs, plus the solution-selection plumbing, validation, Origami logic yamls, and unit tests needed to make it production-usable. ## Motivation PR #6499 brought MX data type support online for gfx950, but the existing TensileLite codegen path leaves significant performance on the table for MXFP4-heavy workloads. The Subtile path restructures global-read / local-read / MFMA / store scheduling at a finer granularity, which **greatly improves MXFP4 GEMM performance when using `HIPBLASLT_MATMUL_MATRIX_SCALE_BLK32_UE8M0_32_8_EXT`** (added to the hipBLASLt CHANGELOG). ## What's included ### 1. New Subtile-based kernel components (Tensile) New modules under `projects/hipblaslt/tensilelite/Tensile/Components/`: * `SubtileBasedKernel.py` (~1850 LOC) — entry point and orchestration of the subtile codegen path; replaces large portions of the standard prefetch / unroll / store flow when `UseSubtileImpl=1`. * `SubtileBasedLogicalScheduler.py` (~2415 LOC) — logical scheduler that builds the subtile-grained instruction graph (GR loads, LR offsets, MFMA tiles, scale loads, stores) from kernel parameters. * `SubtileBasedInstructionScheduler.py` (~433 LOC) — converts the logical schedule to an emit order respecting wave / register / hazard constraints. * `SubtileBasedInstructionEmitter.py` (~216 LOC) — instruction emission helpers shared by the subtile components. ### 2. Kernel writer / common changes * **`KernelWriter.py`**, **`KernelWriterAssembly.py`**: integration points for the subtile path — prefetch, GR offset calculation, LR offset calculation, post-loop, MFMA macro accounting, optimized `storeD`, LDS buffer swap, MX FP4 scale emit, `SrdMXSA/B+2` handling, sgpr allocation / overflow guards, computeLoadSrd fix. * **`SolutionStructs/Solution.py`**, **`SolutionStructs/Problem.py`**: introduces the `UseSubtileImpl` parameter, MX-related reject conditions for non-Subtile paths on gfx950, and additional valid GEMM type combinations for MX inputs. * **`Common/ValidParameters.py`**, **`Common/RequiredParameters.py`**, **`Common/GlobalParameters.py`**: `UseSubtileImpl` registration and defaults. * **`Components/StreamK.py`**: subtile-aware StreamK fixup (incl. import union with the `BufferLoadB32` cache-coherence change from #6837). * **`Components/GlobalWriteBatch.py`**: optimized global write batching for the subtile path (~670 LOC of changes). * **`Components/ComputeStoreVgprs.py`**, **`Components/LSU.py`**, **`Components/WorkGroupMappingAlgos.py`**, **`AsmStoreState.py`**, **`KernelWriterModules.py`**: minor adjustments needed by the subtile pipeline. ### 3. rocisa / host / client * **`rocisa/rocisa/include/container.hpp`**: helpers needed by the new emitter. * **`tensile_host.cpp`**, **`include/Tensile/TensorDescriptor.hpp`**: small fixups for the subtile path and gfx950 build. * **`client/include/DataInitialization.hpp`**, **`client/src/DataInitialization.cpp`**, **`client/src/Reference.cpp`**, **`client/src/ReferenceValidator.cpp`**, **`client/include/TypedId.hpp`**: MX scale init and reference paths used by the new tests. * **`clients/common/include/testing_matmul.hpp`**, **`clients/common/include/norm.hpp`**, **`clients/common/include/hipblaslt_datatype2string.hpp`**, **`clients/common/src/mxDataGen.cpp`**: wiring for batched (>1) testing and MX init. ### 4. Origami / solution selection (gfx950 MXFP4) New auto-tuned logic yamls under `projects/hipblaslt/library/.../Tensile/Logic/asm_full/gfx950/gfx950/Origami/` covering the FP4 SS / HS / BS variants in three layouts: * `Origami/` (default) * `Origami/Origami_nta4/` (no-transpose-A FP4) * `Origami/Origami_ntb4/` (no-transpose-B FP4) (9 new `gfx950_Cijk_Alik_Bljk_F4{SS,HS,BS}_MXA32_MXB32_*_UserArgs.yaml` files in total.) ### 5. New tests **End-to-end gfx950 GEMM yamls** in `Tensile/Tests/common/gemm/gfx950/`: * `subtile_bf16.yaml`, `subtile_mxfp4.yaml` * `mx32f4_tn.yaml`, `mx32f8_tn.yaml` * `mxfp4_mxfp4_{fp32,bf16}_tn_act{,_groupgemm}.yaml` * `mxfp4_fp8_{fp32,bf16}_tn_act{,_groupgemm}.yaml` * `fp8_mxfp4_{fp32,bf16}_tn_act{,_groupgemm}.yaml` **StreamK + MX:** `Tensile/Tests/common/streamk/sk_mx32f4_quick.yaml`, `sk_mx32f8_quick.yaml`. **New unit tests** (`Tensile/Tests/unit/`): * `test_SubtileBasedLogicalScheduler.py` (~1735 LOC) * `test_SubtileBasedSchedulerRef.py` (~596 LOC) * `test_gr_lr_roundtrip.py` (~571 LOC) * `test_storeD_roundtrip.py` (~2420 LOC) * `test_graTileAssignment.py` (~354 LOC) * `test_lraTileAssignment.py` (~360 LOC) * `conftest.py`, `gpu_test_helpers.py` shared fixtures (~601 LOC) **New gtest:** `tensilelite/tests/MXScalePadding_test.cpp`. ### 6. Misc / hardening * Reject conditions: gfx950 MX + non-Subtile, DepthU constraints, GroupGEMM not yet supported with StreamK + MX, AssertSummationElementMultiple=256 for subtile MXFP4, missing-mxblock check for non-MX types. * Skip rocRoller for FP4-A/FP4-B with pre-swizzled scale layout (#42). * `forceDenorm=False` in `generateMXInput` (#11). * Several rebase fixes, copyright/year header updates, and review-comment fixes to `KernelWriter` / `KernelWriterAssembly`. ### 7. CHANGELOG Greatly improved MXFP4 GEMM performance when using HIPBLASLT_MATMUL_MATRIX_SCALE_BLK32_UE8M0_32_8_EXT ## How to use Set `UseSubtileImpl: 1` on a gfx950 MX-FP4 solution (see the new `subtile_mxfp4.yaml` / `mx32f4_tn.yaml` for canonical configs). The path is opt-in — non-MX and non-gfx950 kernels are unaffected. ## Backwards compatibility / risk * All new behavior is gated on `UseSubtileImpl=1` and gfx950. Existing solutions on other architectures or non-MX paths are unchanged. * `GroupGEMM + StreamK + MX` is intentionally rejected for now (TODO). * New Origami yamls only add solutions; nothing existing is modified. ## Test plan * New gtests + unit tests run automatically in CI (Tensilelite Python unit suite, `MXDataGen_test`, `MXScalePadding_test`). * New end-to-end gfx950 GEMM and StreamK yamls are added to the common test buckets. * Manual: run the gfx950 MXFP4 subtile suites (`pytest -k gfx950` after building Tensile, plus `tensilelite-client --yaml subtile_mxfp4.yaml` for sanity). ## Notes for reviewers * This branch was rebased onto current `develop` (post-#6499) by skipping the `users/nakajee/gfx950_mx_rebase_merge` history (which #6499 squash-merged) and replaying only the subtile-specific work as a single squashed commit. The actual code changes in this PR are limited to the files listed above (24 added, 56 modified; ~+170k / −2.6k including generated logic yamls). * The largest reviewable diffs are: * `Tensile/Components/SubtileBased{Kernel,LogicalScheduler,InstructionScheduler,InstructionEmitter}.py` (new files) * `Tensile/KernelWriter.py`, `Tensile/KernelWriterAssembly.py` * `Tensile/SolutionStructs/{Problem,Solution}.py` * `Tensile/Components/{GlobalWriteBatch,StreamK}.py` * `clients/common/include/testing_matmul.hpp` * `client/src/DataInitialization.cpp` * Description of all commits that were squashed for this feature branch: Subtile implementation for gfx950 MX FP4 --- 272f88d: Add sample subtile impl --- Author: brianshi <brianshi@amd.com> --- 60ecede: GR Offset calculation (#1) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> --- be69c1d: Enable post-loop code generation, and add some subroutines --- Author: b-shi <brianshi@amd.com> --- 646d102: LR offset calculation (#2) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> --- 71f4bca: Add GR load emit logic, and misc fixes (#3) --- Author: b-shi <brianshi@amd.com> --- 1fd0db9: Emit LR + init ACCVGPR (#4) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> --- 9d406b9: Add loop and ptr update code --- Author: b-shi <brianshi@amd.com> --- b6127bc: Update GR/LR offset calculation to fully support 2x2, 1x4, 4x1 waveConfigs (#7) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> --- 89ec87c: Account for valuC macro value in SK WS store code --- Author: b-shi <brianshi@amd.com> --- 6edf53d: Rebase fix --- Author: b-shi <brianshi@amd.com> --- 34e79fc: Enable fp4 (#8) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> --- d5a5c57: [Tensilelite] Add MX FP4 scale offset computation for subtile-based kernel (#6) --- Author: Archana Ramalingam <98564406+archana-ramalingam@users.noreply.github.com> --- 7a8a85a: Add lds buffer swap logic --- Author: b-shi <brianshi@amd.com> --- d24a8fe: Add optimized storeD code (#9) --- Author: b-shi <brianshi@amd.com> --- a45c20c: Fix MX scale tensor initialization: set forceDenorm=false in generateMXInput (#11) --- Author: T.J. Alumbaugh <T.J.Alumbaugh@amd.com> --- f945268: [Tensilelite] Enable the MX FP4 scale emit code in the subtile-based kernel (#10) --- Author: Archana Ramalingam <98564406+archana-ramalingam@users.noreply.github.com> --- cf37df4: Use fixed value for SrdMXSA/B+2 (#14) --- Author: Koji Nakajima <75698246+nakajee@users.noreply.github.com> --- f0c8dbc: Merge subtile_mx_f4_schedule to subtile_mx branch (#16) --- Author: b-shi <brianshi@amd.com> --- 543796f: Enable DU > 256, and reduce sgpr allocation (#18) --- Author: b-shi <brianshi@amd.com> --- c65bdb0: Add missing mxblock check for non-mx data types --- Author: b-shi <brianshi@amd.com> --- d64d226: Introduce UseSubtileImpl parameter (#20) --- Author: b-shi <brianshi@amd.com> Squash commits 20-35 from subtile_mx branch --- e4780da: Enable FixSrd2 for A/B (#23) --- Author: b-shi <brianshi@amd.com> * Enable FixSrd2 for A/B * Address comments from PR --------- --- e4c64a7: Add nt libs --- Author: b-shi <brianshi@amd.com> --- cd13ec1: [Tensilelite] Pad MX scale tensor dimensions for unaligned problem sizes (#21) --- Author: Archana Ramalingam <98564406+archana-ramalingam@users.noreply.github.com> * Add scale padding * Add tests * Remove redundant pre-swizzle path * Remove code from conflict * Fix reverted mxdatagen path for tensile tests * Add diverse test cases for scale padding in MXScalePadding_test and subtile.yaml - Expanded test cases to include non-multiple-of-32, even non-multiple-of-16, and odd dimensions. --- d87938f: Split subtile.yaml into subtile_bf16.yaml and subtile_mxfp4.yaml (#22) --- Author: James Newling <james.newling@gmail.com> Replace the 'monolithic' subtile.yaml with two focused test files. All original test coverage is preserved. Two new FP4 groups added. BF16 coverage (subtile_bf16.yaml, tests are essentially unchanged): # | Description | Dest | MIs | PGR | DU | SK | Sizes --+--------------------+------+-----+-----+---------+-----+------ 0 | BF16 TN main | b | 19 | 0 | 64 | 0,3 | 11 1 | BF16 TN large DU | b | 4 | 0 | 128,192 | 0,3 | 7 2 | BSS (f32 output) | s | 6 | 0 | 64 | 0,3 | 9 3 | BF16 bias | b | 2 | 0 | 64 | 0 | 1 FP4 coverage (subtile_mxfp4.yaml): # | Description | Dest | MIs | PGR | DU | SK | Sizes | Status --+--------------------+------+-----+-----+-----+-----+-------+-------- 0 | FP4 TN main | b | 15 | 0 | 256 | 0,3 | 23 | from original 1 | FP4 TN large DU | b | 4 | 0 | 512 | 0,3 | 13 | from original 2 | F4SS (f32 output) | s | 5 | 0 | 256 | 0,3 | 13 | from original 3 | FP4 bias | b | 2 | 0 | 256 | 0 | 1 | from original 4 | FP4 PGR=2 | b | 13 | 2 | 256 | 0 | 5 | new 5 | FP4 expanded MIWT | b | 24 | 0 | 256 | 0 | 5 | new 6 | PGR=2 WG 4x1/1x4 | | 6 | 2 | 256 | 0 | 1 | known failures (commented) Run times on gfx950 (8x MI350X): File | NEV=-1 | NEV=0 -------------------+--------+------ subtile_bf16.yaml | 23s | 23s subtile_mxfp4.yaml | 37s | 40s Where NEV is number of elements to validate. I (James) have checked these numbers, and weirdly it is true that NEV=0 is a bit faster than NEV=-1 for mxfp4. --- af04f0d: Dependency based instruction scheduling (#19) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> * Revert to single partition * Start using dependencies * as is * start using separate EmittedModules * remove reduntant wait * Add _extractPathsFromBeforeDeps * Continue simplification * Simplifying * Add more rules * cleanup * Add fp4 test * fix test * Add tests * Remove after field on emittedmodule * Refactoring instructionSchedule * Add comments * cleanup modules vs ops * Refactoring print functions * Test cleanup * Add more tests * Replace subgroup by partition * Remove unused unroll param * Add high level notes * Simplify NLL and NGLL GR removal * Add some comments * Force instruction insertion if no slots available * Fix test after rebase * Move scale before A/B and track inflight count * Fine-grain vmcnt calculation * Separate counts for scaleA and B * Avoid using m0 update and buffer_lod on same MFMA slot to avoid scalar instruction serialization * Fix test * Add vmcnt test * Fix duplicated loads for 1x4 and 4x1 * Fix placement in reverse order * Fix regression on PGR0 * add fallback to numMFMA=1 --- 3ec902b: Add some 1x4 and 4x1 origami solutions --- Author: b-shi <brianshi@amd.com> --- c5000d3: Fix typo --- Author: b-shi <brianshi@amd.com> --- 226ed84: [hipblaslt] Refactor Srd2 calculation for useFixedSrd2 (#30) --- Author: Koji Nakajima <75698246+nakajee@users.noreply.github.com> --- abf19d4: [Tensilelite] UseSubtileImpl: subtile-aligned edge check for store path (#29) --- Author: b-shi <brianshi@amd.com> * [Tensilelite] UseSubtileImpl: subtile-aligned edge check, OOB guard, and refactoring - Replace Size%MT edge check with subtile-aligned check: NonEdge paired store when trailing rows/cols are a multiple of the subtile block size (waveGroupM rows for M, 16 cols for N). Non-last workgroups always take NonEdge. - Add per-wave OOB guard (subtileM32ValidBlocksSgpr / subtileN16ValidBlocksSgpr) to skip stores outside valid M/N tile bounds in the NonEdge path. - Refactor duplicated OOB guard into _emitSubtileOobGuard helper; refactor M/N guard SGPR computation into _emitSubtileMGuard / _emitSubtileNGuard. - Fix orphan scalar store blockIdxM (was tt0, now (tt0*MatrixInstM)//mBlockSize). - Add quick-exit and edge/non-edge header comments to generated ASM. * Add some bias tests, combine M/N guard to single routine * Add OOB check for C loads, update storeD unit tests to check OOB, simplify quick exit checks * Address more PR comments: add M group skip, and skip to store end. simplified loadC OOB mask --------- --- 637881a: Fix unit tests & remove legacy code for subtile interleaving (#33) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> * Fix gr_lr_roundtrip test * Use non-interleaved version as ref code * Fix scheduler test * Removed legacy interleaved mode for LR/GR offset calculation --- e9cb889: Fix MX FP4 scale buffer allocation and initialization for batched GEMM (#25) --- Author: Archana Ramalingam <98564406+archana-ramalingam@users.noreply.github.com> * Fix bacth count issue * Add batch count tests * Fix bacth count issue * Address PR review: clarify FP4-specific byte stride and add non-aligned batched tests - Updated comments on dataBatchBytes computation to clarify FP4 packing assumption (2 elements/byte) and flag that non-FP4 block-scaling types would require updating this conversion. - Added batched test cases with non-multiple-of-32 M/N dimensions: FP4 DU=256: [48,48,2] and [33,65,2] FP4 DU=512: [63,63,2] BF16: [50,100,2] --------- --- a43247b: Update some test yamls (#31) --- Author: b-shi <brianshi@amd.com> --- e2f69c8: Add f4bs origami library with activation function support. Refactor sgpr allocation to reduce sgpr usage in post loop. Store code-path reorganization (#32) --- Author: b-shi <brianshi@amd.com> * Free swap/localwritebase sgprs before post-loop * Defer sgpr allocation to remove holds in sgpr pool. Add Origami library logic files for Cijk_Alik_Bljk_F4BS_MXA32_MXB32 (base, nta4, ntb4 variants). * Remove uneeded alignment and comment * Add more epilogue tests * Remove older origami library for f4bs * Reorder post-loop code blocks to after persistant loop Misc fixes * Fix build issues, relax longjump sgpr requirements * Fix GSU0 branch logic --------- --- 3f034bf: Add F4HS and F4SS Origami library logic for FP4→F16 and FP4→F32 GEMM (#35) --- Author: Majedul Sujon <85503863+msujon-AMD@users.noreply.github.com> * Add F4HS and F4SS Origami library logic for FP4→F16 and FP4→F32 GEMM - Add 6 new yaml files (F4HS, F4SS) across Origami, Origami_nta4, Origami_ntb4 - Update F4BS yaml files: AssertSummationElementMultiple 32→256 for K%256 enforcement - Add ("F4", "F4", "H", "S") to _validGEMMTypes and _HPATypes in Problem.py * Add F4HS test cases to subtile_mxfp4.yaml Add two new benchmark problem blocks for FP4→F16 (F4HS): - No-bias block: same wavetile and problem size coverage as F4SS - Bias epilogue block: BiasDataTypeList [s, h], relu/none activations * Add F4HS (FP4->Half) type support to Tensile client Add TypedGemm_F4_H_S typedef and corresponding reference CPU solver case so F4HS (FP4 input, Float16 output, Float compute) problems can be validated by the benchmark client. --------- --- d0bc8fd: Rewrite subtile-based scheduler. Fix DU>64 & enable very large MT (#36) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> * Initial support for DU>256 * Renaming * add option to do DU=512 in the tests * blocked K-major for scale * Change scaleSet swap logic * Update print functions * Put scales after values for avoid race conditions * Fix tests * more test * tweak printschedule display * Add PGR2 in the yaml tests * Add new scaleGROp * comment out failing tests * Revert "comment out failing tests" This reverts commit 1f5802c. * Draft new logical scheduler * Refactoring * Add more test on step1 * Add more tests on step1 * add bf16 320x320 test * reduce step1 code * Simplify step1 logic * validate some step1 test * Fix partition 2x2 test * more step1 test * 320x320 BF16 test * Add test DU512 + partition2x2 * Simplify step1 code * Add step2 tests * Fix multi-partition step2 * Add step2 du512, 2x2 partition test * Use common algo for all numPartitions * Draft for step3 tests * remove useless tests * New GR algo (draft) * [Step3] Add more test * Iteration on GR * Display ordered GR list with granularities * More test * Add some comments * Disable by default debug logs * Getting rid of step naming * Start remove AnnotatedOp (still there in group pass) * Split dependency Ops * Add todo on place_GRs pass * Valid test_annotate_deps_1x1_partition_DU256 * Test output looking better (still WIP) * single dep for LR tooo * Add remove_cross_deps pass * Fix bugs in dependency pass * insert_gr_lr_inc pass * Add group_lr_gr pass * Add emit pass * Quick port of instruction Emit code * Move emit function to separate file * Refactoring instructionEmitter * Port vgprTile tracking * Reworking second pass (WIP) * Display unrolling requirement * Unrolling check on 2nd pass * Generic validation for assign_vgpr pass * Fix unroll * Add inst schedule in standalone mode * Use lrGran for vgprTile size calculation * Fix bug in emit pass (missing depencency) * PreMFMA path + non-duplication scale load * missing globalReadLDSBufferSwap for GR_INC scales * add wairlr_sync on all LR->GR dep * add waitgr_sync op * remove_unnecessary_gr_deps * Change LR dispatch algo a bit to avoid too many waitgr_sync * Avoid duplicated loads in emitter * Fix bug on gr_emit code * GrInc pass. fix duplicated insertion for B * Fix missing LR_inc for SA/SB * preloop, NLL, NGLL * Simplify preloop * minor changes * Move unroll logic to scheduler * minor changes * Fix unroll id bug on NLL / NGLL * Disable post GRINC for now * Remove commented code * Handle 1x4, 4x1 gr read gran * Fix vmcnt computation * Use correct grCount mapping * Revert in emit logic on buffer_load for PGR0 needs * Add bf16 version in standalone test * Fix LR_Inc insertion on DU>64 * Add subIterK/Partition comment to codegen * Fix issue in GrInc placement * Remove last_mt * Fix LR MT index bug with muli-partition * Disable early LDS size check when subtileImpl is on * Add pass to remove redundant LR deps + fixed issue on dependency annotation pass * Remove more LR redundant deps * Only insert wait_lr_sync on deps * Simple algo to select partition config * Remove HC value for partitions... * Take into account all inflight GR (all tensors) * Fix tests and regressions on gr counts * Fix grCount merge calculation * Better display of dependencies * Add remove_wait_lr_sync after grouping * Add temporary non reg file * Change merge logic on GR grouping pass * Fix non necessary wait_lr_sync * Downgrade some waitlr_sync to sync + added 384x256 no reg test * non reg test 320x320 * Add larger MT * non reg test for fp4 256x256 * Moving out instructionScheduler * Remove old scheduler * Renaming scheduler * Re-work test * Add larger MT test cases * Rename non-ref test * Re-add standalone mode * Refactor DepOp * Remove dead code * Remove MFMATileSize class * Remove from_til_info * Avoid redundant tensor list creation * Remove hardcode granularities in vgrpTile allocation pass. Simplify code. * Re-enable # PGR=2 WG 4x1/1x4, K > DU tests * Remove unused GRScaleOp * DepRef renaming * Get rid of MT string representation * Remove TODO * EmmitedModule simplication * Use explicit pass dependencies * Renaming LogicalScheduler * Remove old test_InterleavingScheduler.py file * Commenting failing test for now * Remove debug logs * Disable lds padding when using UseSubtileImpl --- e8e8c09: Fix LR-GR dependency issue when DU>64 (#40) --- Author: sebvince <115461989+sebvince@users.noreply.github.com> * Fix and simplify logic for remove_unnecessary_lr_deps * Add new ref tests for 128x128x(128,64) --- 4aa441a: Rebase fix --- Author: b-shi <brianshi@amd.com> --- 5ba911e: Skip rocRoller for FP4-A/FP4-B + pre-swizzled scale layout (#42) --- Author: Archana Ramalingam <98564406+archana-ramalingam@users.noreply.github.com> --- 9c74998: Rebase fix --- Author: b-shi <brianshi@amd.com> --- 842b149: Addressed review comments for KernelWriter and KernelWriterAssembly --- Author: Koji Nakajima <knakajim@amd.com> --- dce43b1: Fix computeLoadSrd issue --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- c075bbf: Fix preSolution CPU re-sync regressing subtile_mxfp4.yaml --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- ced840f: Fix computeLoadSrd issue (#43) --- Author: bnemanich <brad.nemanich@amd.com> --- bc2f6dd: Small update for gfx950 mx tests + more - enable UseSubtileImpl for all gfx950 non subtile mx tests - skip all gfx950 mxfp8 - use MXScaleFormat=1 as default - set AssertSummationElementMultiple=256 for subtile mxfp4 - fix isSwizzledSubtile in computeLoadSrd --- Author: Koji Nakajima <knakajim@amd.com> --- 5c794b7: Fix gsuasb.yaml failures --- Author: b-shi <brianshi@amd.com> --- 727f8db: tensilelite: add solution reject conditions for UseSubtileImpl=1 (#38) --- Author: Majedul Sujon <85503863+msujon-AMD@users.noreply.github.com> --- 8928fbb: Add more reject conditions for Subtile --- Author: Koji Nakajima <knakajim@amd.com> --- 6e63ab6: Fix kringshift test failures --- Author: b-shi <brianshi@amd.com> --- b3e9724: Update reject condtion for DepthU in subtile case. Plus, update DepthU setting for gfx950 mx test cases --- Author: Koji Nakajima <knakajim@amd.com> --- 5ab6009: Fix build errors --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- 4a4edca: Update more mxfp4 tensilelite test cases --- Author: Koji Nakajima <knakajim@amd.com> --- bbbc553: Update change log --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- 6476c04: Add more reject conditions for gfx950 subtile --- Author: Koji Nakajima <knakajim@amd.com> --- c5828c4: Updated gfx950 mxfp4 test cases - add StreamK setting - skip groupgemm tests for now (groupgemm does not support streamK) --- Author: Koji Nakajima <knakajim@amd.com> --- f1fc2f1: Fix hipblaslt build error of gfx950 --- Author: Koji Nakajima <knakajim@amd.com> --- 70cea1b: Updated subtile_mxfp4.yaml (add StreamK) --- Author: Koji Nakajima <knakajim@amd.com> --- c1c9b2a: Add uninit lsc,lsp, etc.. fields for subtile --- Author: b-shi <brianshi@amd.com> --- c0c1f72: Fixed merge error in testing_matmul.hpp --- Author: Koji Nakajima <knakajim@amd.com> --- 191e0cb: Add missed batch_count >1 changes --- Author: archana-ramalingam <Archana.Ramalingam@amd.com> --- 01c52f8: Addressed PR comments --- Author: Koji Nakajima <knakajim@amd.com> --- 4e89c91: Reduce mxfp4 test time --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- 3dac20f: Prevent overflow for wgmxcc sgpr allocation --- Author: b-shi <brianshi@amd.com> --- 18dec79: Fix error with problem type --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- 9e69ffd: Add a reject conditoin for gfx950 mx + non Subtile --- Author: Koji Nakajima <knakajim@amd.com> --- 0eed3ba: Add more valid GEMM types --- Author: Brad Nemanich <brad.nemanich@amd.com> --- 8b5514e: Fix missing b build error --- Author: archana-ramalingam <Archana.Ramalingam@amd.com> --- f981ff5: Fix 1250 tests --- Author: Brad Nemanich <brad.nemanich@amd.com> --- d1e69d9: Add more FP4 tests --- Author: Brad Nemanich <Brad.Nemanich@amd.com> --- e3a688f: Add MXScaleFormat: 1 to all gfx950 mx test yaml --- Author: Koji Nakajima <knakajim@amd.com> --- aaef3f5: Add DataTypeMXSA,B setting in gfx950 mxfp4 logic yaml --- Author: Koji Nakajima <knakajim@amd.com> --- 861ef8e: Add DataTypeMXSA,B setting in gfx950 mxfp4 logic yaml (nta4,ntb4) --- Author: Koji Nakajima <knakajim@amd.com> Co-authored-by: Archana Ramalingam <Archana.Ramalingam@amd.com> Co-authored-by: Brad Nemanich <Brad.Nemanich@amd.com> Co-authored-by: Brian Shi <Brian.Shi@amd.com> Co-authored-by: James Newling <James.Newling@amd.com> Co-authored-by: Koji Nakajima <Koji.Nakajima@amd.com> Co-authored-by: Majedul Sujon <Majed.Sujon@amd.com> Co-authored-by: Sebastien Vince <Sebastien.Vince@amd.com> Co-authored-by: T.J. Alumbaugh <T.J.Alumbaugh@amd.com> ## Submission Checklist - [ ] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests. --------- Co-authored-by: Archana Ramalingam <Archana.Ramalingam@amd.com> Co-authored-by: Brad Nemanich <Brad.Nemanich@amd.com> Co-authored-by: Brian Shi <Brian.Shi@amd.com> Co-authored-by: James Newling <James.Newling@amd.com> Co-authored-by: Koji Nakajima <Koji.Nakajima@amd.com> Co-authored-by: Majedul Sujon <Majed.Sujon@amd.com> Co-authored-by: Sebastien Vince <Sebastien.Vince@amd.com> Co-authored-by: T.J. Alumbaugh <T.J.Alumbaugh@amd.com>

…7008) ## Motivation Several memory leaks were detected in MIOpen gtests using ASan. Some of the tests were blacklisted and others were not. This change looks to fix all of the low hanging fruit, which are the majority of the leaks found. This includes all of the critical leaks (>100MB) that were reported. Some other leaks were identified as needing a larger refactor to resolve. After fixing the w_supertensor.cpp leak, the supertensor tests hit virtual memory area limit errors. These changes are to enable running them with ASan, but only with a subset of the tests in order to not hit limits. The test coverage being lost is fairly negligible and the full tests are still run without ASan enabled. ## Technical Details Here is a summary of the files looked at and the changes made: | File | Status | Notes | |------|--------|-------| | `test/gtest/gpu_mha_forward.cpp` | Fixed | Added `miopenDestroySolution()` loop after using solutions. Fixes MHA forward solution descriptor leaks (report #11, #13). | | `test/gtest/gpu_mha_backward.cpp` | Fixed | Added `miopenDestroySolution()` loop after using solutions. Fixes MHA backward solution descriptor leaks (report #9, #12). | | `test/gtest/mha_find20.cpp` | Fixed | Added `miopenDestroySolution()` loop in both `MhaForward` and `MhaBackward` tests. Fixes MHA Find2.0 solution leak (report #16). | | `test/gtest/gtest_desc_guard.hpp` | Fixed | New shared header introduced by the refactor. Provides a single `DescGuard<DescType, CreateFn, DestroyFn>` template (with `TensorDescGuard`, `ConvDescGuard`, `DropoutDescGuard`, `RNNDescGuard` aliases), a `HandleGuard` RAII wrapper for `miopenHandle_t`, and the `DestroyInternalRnnDropoutDesc(rnnDesc)` helper used by every RNN/LSTM/GRU test to free the internal `DropoutDescriptor` that `miopenCreateRNNDescriptor` allocates and `miopenSetRNNDescriptor*` then leaks. Replaces the per-file ad-hoc guard structs from the initial implementation. | | `test/gtest/w_supertensor.cpp` | Fixed | Switched raw descriptors to the shared `RNNDescGuard` / `TensorDescGuard` from `gtest_desc_guard.hpp`. Added a class-local `DestroyDropoutDesc()` (called from `TearDown` and before `miopenSetRNNDescriptor`) to prevent the `miopenSetRNNDescriptor` overwrite leak. Reduced test parameter space under ASan to avoid OOM. Removed unused `seqLen` parameter and dead `param_dev_out`/`bias_dev_out` allocations. | | `test/gtest/lstm.hpp` | Fixed | Switched `rnnDesc` → `RNNDescGuard`, `DropoutDesc` → `DropoutDescGuard`, and `mio_handle` → `HandleGuard` (which now owns the `miopenDestroy` call), all from the shared `gtest_desc_guard.hpp`. Hoisted `dropout_state_buf` so it can be `hipFree`d at the end of the dropout path. Added `DestroyInternalRnnDropoutDesc(rnnDesc)` before `miopenSetRNNDescriptor*` and (in the non-dropout path only) at end of `Run`, which frees the internal `DropoutDescriptor` that the Set call would otherwise leak. Fixes ~615 MB dropout leaks and ~16.5 KB non-dropout descriptor leaks (report #3, #4, #6, #17-#22). | | `test/gtest/gru_test.cpp` | Fixed | In the GRU test class: switched `rnnDesc` → `RNNDescGuard`, `DropoutDesc` → `DropoutDescGuard`, `mio_handle` → `HandleGuard`, and added `DestroyInternalRnnDropoutDesc(rnnDesc)` before `miopenSetRNNDescriptor*` and (non-dropout path only) at end of `Run` plus `hipFree(dropout_state_buf)` for the dropout path. In the in-file `GRUFwdCPUVerify` / `GRUBwdDataCPUVerify` helpers: converted the raw `dropout_inputTensor` / `dropout_outputTensor` declarations to `TensorDescGuard` (mirroring the `cpu_rnn.hpp` change for the LSTM/RNN helpers). | | `test/gtest/softmax_find20.cpp` | Fixed | Changed `Finalize()` to take the `std::vector<miopenSolution_t>&` and destroy each solution via `miopenDestroySolution()` before destroying the problem. Updated all 6 `TEST(...)` callers to pass the solutions vector. Fixes Find2.0 softmax solution/kernel leaks (report #25-#27). | | `test/gtest/rnn_seq_api.hpp` | Fixed | Hoisted `dropout_state_buf` so the dropout path can `hipFree` it at the end. Added `DestroyInternalRnnDropoutDesc(&rnnDesc)` before `miopenSetRNNDescriptor*` and (non-dropout path only) at end of the run to free the internal `DropoutDescriptor` allocations leaked by `RNNDescriptor` copy-assignment. Same pattern as the LSTM/GRU fixes (report #29-#30). | | `test/cpu_rnn.hpp` | Fixed | Converted 6 raw `miopenTensorDescriptor_t dropout_input/outputTensor` declarations across the LSTM/RNN CPU verification helpers (`LSTMFwdCPUVerify`, `LSTMBwdDataCPUVerify`, `RNNFwdTrainCPUVerify`, `RNNBwdDataCPUVerify`, `GRUFwdCPUVerify`, `GRUBwdDataCPUVerify`) to the shared `TensorDescGuard` from `gtest_desc_guard.hpp`. Removed the redundant `miopenCreateTensorDescriptor` calls and updated 12 `miopen::deref(...)` sites to `.get()`. (Note: the GRU helpers in this header are stale duplicates; the live ones are inside `test/gtest/gru_test.cpp` and were updated there too.) Fixes the LSTM/GRU CPU-verify tensor descriptor leaks (report #1, #2, #14, #15, #32, #33). | | `test/gtest/rnn_vanilla_common.hpp` | Fixed | Added `DestroyInternalRnnDropoutDesc(rnnDesc)` calls before `miopenSetRNNDescriptor*` and (non-dropout path only) at end of the run. The `RNNDescGuard` / `DropoutDescGuard` usage was already in place from an earlier commit and now resolves to the shared definitions in `gtest_desc_guard.hpp`. Same pattern as LSTM/GRU/rnn_seq_api fixes (report #14, #15, #32, #33). | | `test/gtest/graphapi_gtest_common.hpp` | Skipped | File no longer exists in the codebase. The GraphApi test infrastructure has been removed; only the leak report (a stale snapshot) still references it. No fix possible against the current source tree. | | `test/gtest/graphapi_execution_plan.cpp` | Skipped | File no longer exists in the codebase (GraphApi removed). The leak it represented was largely an external hipblaslt bug anyway; the remaining test-side portion is not fixable against the current source. | | `test/gtest/na_train.cpp` / `na_inference.cpp` / `na_*_find2.cpp` | Skipped | Leaks come from the internal MIOpen solver/kernel cache living on the global singleton handle, which is never destroyed. Not easily fixable without redesigning global handle lifecycle. Should be suppressed in the ASan suppression file. | | hipblaslt / rocblaslt (external) | Skipped | `SolutionCache::addKernel` and `preloadCustomKernels` leak via `_rocblaslt_handle` constructor. Called from `miopen::Handle::CreateHipblasLtHandle`. This is an upstream bug in hipblaslt/rocblaslt, not fixable in MIOpen. Affects suites that initialize a handle (report Category 2, Category 7). | | CLR / HIP runtime (external) | Skipped | `amd::Context` and `amd::roc::Device` global initialization leaks from `rocclr/platform/context.cpp`. HIP runtime internals, not fixable in MIOpen. | | `src/hipoc/hipoc_program.cpp` | Skipped | `HIPOCProgramImpl` objects leak during kernel compilation/caching (line ~178). This is an internal MIOpen kernel cache lifecycle issue that requires deeper architectural changes to fix. Contributes small amounts to MHA and Softmax Find2.0 leaks. | | `test/gtest/conv_api.cpp` | Fixed | Already clean against current source — `miopenDestroyConvolutionDescriptor(conv_desc)` call exists (line 24) inside the test loop. ASAN run reports no leaks. The leak report was based on a stale snapshot. The hipblaslt handle init portion is tracked under the external-skipped row. | | `test/gtest/log_test.cpp` (CPU_LOG_TEST_FUSION / CPU_LOG_TEST_NEG) | Fixed | Already clean against current source — `Tensor`, `Conv`, `CreateCBAFusionPlan`, `CreateBNormFusionPlan` all have proper destructors that call the corresponding `miopenDestroy*` APIs in `log.cpp`. ASAN run on `CPU_LOG_TEST_*` (11 tests across log_test.cpp + log_test_neg.cpp) reports no leaks. The hipblaslt handle init portion is tracked separately under the external-skipped row. | | `test/gtest/fusion_test.cpp` (CPU_FusionCreateOpConvForward) | Fixed | File renamed from `fusion.cpp` to `fusion_test.cpp`. Already clean against current source — uses `TensorDescGuard`/`ConvDescGuard` for tensor/conv descriptors and calls `miopenDestroyFusionPlan(fusionPlanDesc)` on the fusion plan (line 195). ASAN run on `CPU_FusionCreateOpConvForward_FP32.*` reports no leaks. | | `test/gtest/deterministic_conv_api.cpp` | Fixed | Already clean against current source — uses `ConvDescGuard` (line 66) for the conv descriptor. ASAN run on `*CPU_DeterministicConvApi*` reports no leaks. | | `test/gtest/fusion_aux.cpp` (GPU_FusionAux) | Fixed | Already clean against current source — uses `ConvDescGuard` plus stack-allocated internal C++ objects (`miopen::TensorDescriptor`, `miopen::FusionPlanDescriptor`) which have proper destructors. The `convoOp` handle is owned by the fusion plan. ASAN run on `*GPU_FusionAux*` reports no leaks. | | `test/gtest/backend_api.cpp` (CPU_BackendApi) | Skipped | File no longer exists in the codebase. The backend API test infrastructure (part of the removed GraphApi suite) was removed; no fix possible against the current source tree. | ### High-level notes New shared infrastructure (test/gtest/gtest_desc_guard.hpp) - DescGuard<DescType, CreateFn, DestroyFn> — a single RAII template parameterized on the descriptor type and its create/destroy entry points. Aliases provide TensorDescGuard, ConvDescGuard, DropoutDescGuard, and RNNDescGuard, replacing the four near-identical guard structs that were copy-pasted across test files in the initial implementation. - HandleGuard — separate RAII wrapper for miopenHandle_t (couldn't reuse the template because miopenCreateWithStream takes an extra hipStream_t argument). Supports lazy create(stream) so callers that only need a handle in the dropout branch can default-construct one and populate it conditionally. - DestroyInternalRnnDropoutDesc(rnnDesc) — frees the internal DropoutDescriptor that miopenCreateRNNDescriptor allocates and that miopenSetRNNDescriptor* then orphans. Replaces the equivalent inline blocks that LSTM/GRU/RNN tests were each carrying. The header documents the two call-sites: before each Set* (always safe) and at end-of-run only on the non-dropout path (the dropout path aliases the user-owned descriptor, so freeing would double-free). Recurring patterns enabled by the refactor - The "leak from Set* overwriting the default-constructed internal dropout descriptor" fix collapsed from per-file code to a one-line helper call, applied uniformly across lstm.hpp, gru_test.cpp, rnn_seq_api.hpp, and rnn_vanilla_common.hpp. - mio_handle ownership in LSTM/GRU is now expressed via HandleGuard rather than a manual miopenDestroy at the end of the dropout branch — eliminates a class of forgotten-cleanup bugs. - dropout_state_buf is consistently hoisted out of the dropout if block so an end-of-run hipFree can release it; deletion of the buffer pairs visibly with its allocation. Notable non-RNN change - softmax_find20.cpp was the only Find2.0 leak fix in this commit: Finalize() now takes the solutions vector and calls miopenDestroySolution() for each before destroying the problem. Same shape applied to all 6 tests in the file. ## Test Plan Run the tests beforehand to observe the ASan leak errors and then again afterward to verify the fixes have resolved the problem. ## Test Result List from ROCM-21512: | # | Test Name | Status | Leak Status | |---:|------------------------------------------------------|------------------------------------------------|-----------------| | 1 | Smoke/GPU_RNNVanillaDropout_FP32 | PASSED (4 tests) | No leaks | | 2 | Smoke/GPU_RNNVanillaDropout_FP16 | PASSED (4 tests) | No leaks | | 3 | Full/GPU_LSTM_dropout_FP32 | PASSED (4 tests) | No leaks | | 4 | Full/GPU_LSTM_dropout_FP16 | PASSED (4 tests) | No leaks | | 5 | CPU_GraphApiExecutionPlanBuilder_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 6 | Full/GPU_LSTM_dropout_FP64 | REMOVED (PR #5750, 2026-03-26) | n/a (deleted) | | 7 | Unit/CPU_GraphApiPointwise_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 8 | Full/GPU_LstmMSRnn_FP32 | PASSED (1152 tests) | No leaks | | 9 | Smoke/GPU_Bwd_Mha_FP32 | PASSED (12 tests) | No leaks | | 10 | Full/GPU_LstmMSRnn_FP16 | PASSED (864 tests) | No leaks | | 11 | Smoke/GPU_Fwd_Mha_FP32 | PASSED (15 tests) | No leaks | | 12 | Full/GPU_Bwd_Mha_FP32 | PASSED (6 tests) | No leaks | | 13 | Full/GPU_Fwd_Mha_FP32 | PASSED (7 tests) | No leaks | | 14 | Full/GPU_RNNVanilla_FP32 | PASSED (96 tests) | No leaks | | 15 | Full/GPU_RNNVanilla_FP16 | PASSED (96 tests) | No leaks | | 16 | GPU_TestMhaFind20_FP32 | PASSED (2 tests) | No leaks | | 17 | Full/GPU_LSTM_FP32 | PASSED (32 tests) | No leaks | | 18 | Full/GPU_LSTM_FP16 | PASSED (32 tests) | No leaks | | 19 | Full/GPU_LSTM_extra_FP32 | PASSED (30 tests) | No leaks | | 20 | Full/GPU_LSTM_extra_FP16 | PASSED (30 tests) | No leaks | | 21 | Full/GPU_DeepBench_LSTM_FP16 | PASSED (22 tests) | No leaks | | 22 | Full/GPU_DeepBench_LSTM_FP32 | PASSED (22 tests) | No leaks | | 23 | CPU_LOG_TEST_FUSION_NONE | PASSED (2 tests) | No leaks | | 24 | CPU_LOG_TEST_NEG_NONE | PASSED (4 tests) | No leaks | | 25 | GPU_SoftmaxFind20_BFP16 | PASSED (2 tests) | No leaks | | 26 | GPU_SoftmaxFind20_FP16 | PASSED (2 tests) | No leaks | | 27 | GPU_SoftmaxFind20_FP32 | PASSED (2 tests) | No leaks | | 28 | CPU_ConvApi_NONE | PASSED (1 test) | No leaks | | 29 | Full/GPU_RNNSeqApi_FP16 | PASSED (16 tests) | No leaks | | 30 | Full/GPU_RNNSeqApi_FP32 | PASSED (16 tests) | No leaks | | 31 | UnitVAN/CPU_GraphApiRng_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 32 | Smoke/GPU_RNNVanilla_FP16 | PASSED (4 tests) | No leaks | | 33 | Smoke/GPU_RNNVanilla_FP32 | PASSED (4 tests) | No leaks | | 34 | CPU_FusionCreateOpConvForward_FP32 | PASSED (1 test) | No leaks | | 35 | CPU_GraphApiOperationReduction_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 36 | Unit2IV1/CPU_GraphApiOperationPointwiseBwd_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 37 | Unit2IV1/CPU_GraphApiOperationPointwiseTwoInput | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 38 | Unit2IV2/CPU_GraphApiOperationPointwiseBwd_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 39 | Unit2IV2/CPU_GraphApiOperationPointwiseTwoInput | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 40 | UnitVAB/CPU_GraphApiRng_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 41 | Smoke/GPU_RNNVanillaDropout_FP16 (duplicate of #2) | (see #2) | (see #2) | | 42 | CPU_GraphApiOperationGraphDescriptor_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 43 | UnitVA/CPU_GraphApiVariantPack_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 44 | UnitVAU/CPU_GraphApiRng_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 45 | CPU_GraphApiOperationReshape_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 46 | Smoke/CPU_DeterministicConvApi_NONE | PASSED (1 test) | No leaks | | 47 | Smoke/GPU_FusionAux_FP32 | PASSED (1 test) | No leaks | | 48 | CPU_GraphApiEngineHeur_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 49 | Unit/CPU_GraphApiReduction_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 50 | CPU_GraphApiEngineCfg_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 51 | Unit/CPU_GraphApiMatMul_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 52 | CPU_BackendApi_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 53 | UnitIV/CPU_GraphApiOperationPointwiseOneInput | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 54 | Unit3IV/CPU_GraphApiOperationPointwiseThreeInput | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 55 | UnitVA/CPU_GraphApiOperationMatmul_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | 56 | UnitVA/CPU_GraphApiOperationRng_NONE | REMOVED (PR #5603, 2026-03-26) | n/a (deleted) | | Outcome | Count | |------------------------------------|------:| | Passed, no leaks | 39 | | Passed, leaks detected | 0 | | Failed | 0 | | Crashed / timed out | 0 | | Removed — GraphAPI purge (#5603) | 15 | | Removed — FP64 LSTM purge (#5750) | 1 | | Duplicate (not re-run) | 1 | | **Total rows** | **56** | ## Risk Assessment Low --------- Co-authored-by: JonathanLichtnerAMD <195780826+JonathanLichtnerAMD@users.noreply.github.com>

Change shared libs name

d079aeb

jayhawk-commits requested review from BradPepersAMD, BrianHarrisonAMD, JonathanLichtnerAMD and adickin-amd as code owners April 29, 2025 17:17

jayhawk-commits closed this May 2, 2025

jayhawk-commits deleted the miopenCI branch May 2, 2025 19:59

assistant-librarian Bot pushed a commit that referenced this pull request May 13, 2025

Zenguyen/block sort missing unit test (#3)

ce5a2b2

* finished sort keys no size tests * completed sort pair no size functions * completed sort pair with size functions * updated changelog

jayhawk-commits self-assigned this May 17, 2025

jayhawk-commits added the migration Tasks or issues tied to migration to this monorepo label May 17, 2025

jayhawk-commits pushed a commit that referenced this pull request Jun 8, 2025

Enable gfx12 support (#3) (#148)

13b1510

* Enable gfx12 support [ROCm/hipRAND commit: 881a6bf]

idass1990 pushed a commit that referenced this pull request Jun 13, 2025

Remove hip dependency (#3)

577eae1

Removes the cmake dependency that was added for hip. This was preventing cuda builds of hipBLAS from building. [ROCm/hipBLAS-common commit: eb8d0fa]

This was referenced Jun 30, 2025

Invalid instruction oops for gfx1151 on Fedora rawhide #458

Open

kernel oops for gfx1201 on Fedora Rawhide #459

Open

assistant-librarian Bot mentioned this pull request Jul 14, 2025

rocsparse-test fails to run on Fedora #627

Closed

assistant-librarian Bot mentioned this pull request Jul 23, 2025

hipblas-test double free error on Fedora. #807

Closed

evbo mentioned this pull request Aug 14, 2025

[Feature] Tensile gfx1032 support #1202

Open

assistant-librarian Bot mentioned this pull request Sep 18, 2025

Kernrl oops on illegal instruction for gfx1151 on Fedora Rawhide #1675

Closed

ibrahimw1 added a commit that referenced this pull request Nov 25, 2025

Integrate PR comments #3

0284d57

evetsso pushed a commit to evetsso/rocm-libraries that referenced this pull request Dec 31, 2025

fix endif logic error ROCm#3

fc81799

Vsevolod1983 mentioned this pull request Jan 26, 2026

[MIOpen] Convert bn_3d_peract_test from ctest to Google Test #2447

Closed

4 tasks

pdhirajkumarprasad mentioned this pull request Mar 25, 2026

Adding multiple enhancement to improve the execution time as well as debugging #5740

Open

1 task

nakajee pushed a commit to nakajee/rocm-libraries that referenced this pull request Mar 31, 2026

Add GR load emit logic, and misc fixes (ROCm#3)

b36af61

sebvince referenced this pull request in sebvince/rocm-libraries Apr 3, 2026

Add GR load emit logic, and misc fixes (#3)

535a82f

pdhirajkumarprasad mentioned this pull request Apr 7, 2026

Fix code generation when UseBeta is false #6202

Open

1 task

sebvince referenced this pull request in sebvince/rocm-libraries Apr 9, 2026

Add GR load emit logic, and misc fixes (#3)

8d1929f

NolanHannaAMD mentioned this pull request May 1, 2026

[MIOpen] [ROCM-21512] [ALMIOPEN-1814] Fix MIOpen ASan memory leaks #7008

Merged

bnemanich mentioned this pull request May 3, 2026

Add gfx950 MXFP4 and BF16 Subtile-based kernel implementation #7017

Merged

1 task

poyenc mentioned this pull request May 14, 2026

[CK] Fix RDNA3 FMHA tile-load paths #7016

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Change shared libs name#3

Change shared libs name#3
jayhawk-commits wants to merge 1 commit into
mockupfrom
miopenCI

jayhawk-commits commented Apr 29, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

jayhawk-commits commented Apr 29, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant