Skip to content

test(hipblaslt): skip gfx1250-strict where the tests can't run #13464

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
archana-ramalingam merged 2 commits into from
Oct 12, 2026
Merged

test(hipblaslt): skip gfx1250-strict where the tests can't run #13464

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3fdb7f2
test(hipblaslt): skip gfx1250-strict where the tests can't run
archana-ramalingam Oct 10, 2026
d51015f
test(hipblaslt): limit the bf8 MX rows to gfx950 per row, not by split
archana-ramalingam Oct 12, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 7 additions & 6 deletions projects/hipblaslt/clients/tests/data/matmul_gtest.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4850,15 +4850,16 @@ Tests:
# shape here.
###############################################################################

# Every fp8 format pair, each solution launched and checked twice.
# Every fp8 format pair, each solution launched and checked twice. gfx1250 has MX
# kernels for f8 x f8 only, so the rows with bf8 run on gfx950 alone.
- name: matmul_fast_check_mx_fp8
category: pre_checkin
function:
matmul:
- { a_type: f8_r, b_type: f8_r, c_type: f32_r, d_type: f32_r, compute_type: c_f32_r, scaleA: 3, scaleB: 3, scale_type: f32_r}
- { a_type: bf8_r, b_type: bf8_r, c_type: f32_r, d_type: f32_r, compute_type: c_f32_r, scaleA: 3, scaleB: 3, scale_type: f32_r}
- { a_type: f8_r, b_type: bf8_r, c_type: f32_r, d_type: f32_r, compute_type: c_f32_r, scaleA: 3, scaleB: 3, scale_type: f32_r}
- { a_type: bf8_r, b_type: f8_r, c_type: f32_r, d_type: f32_r, compute_type: c_f32_r, scaleA: 3, scaleB: 3, scale_type: f32_r}
- { a_type: bf8_r, b_type: bf8_r, c_type: f32_r, d_type: f32_r, compute_type: c_f32_r, scaleA: 3, scaleB: 3, scale_type: f32_r, gpu_arch: '950'}
- { a_type: f8_r, b_type: bf8_r, c_type: f32_r, d_type: f32_r, compute_type: c_f32_r, scaleA: 3, scaleB: 3, scale_type: f32_r, gpu_arch: '950'}
- { a_type: bf8_r, b_type: f8_r, c_type: f32_r, d_type: f32_r, compute_type: c_f32_r, scaleA: 3, scaleB: 3, scale_type: f32_r, gpu_arch: '950'}
matrix_size:
- { M: 256, N: 256, K: 512 }
- { M: 1040, N: 528, K: 1024 }
Expand Down Expand Up @@ -4894,13 +4895,13 @@ Tests:
fast_check: 1
gpu_arch: '(950|1250)'

# Large K, where the scale tensors span many blocks.
# Large K, where the scale tensors span many blocks. f8 x bf8 on gfx950 alone, as above.
- name: matmul_fast_check_mx_fp8_large_k
category: nightly
function:
matmul:
- { a_type: f8_r, b_type: f8_r, c_type: f32_r, d_type: f32_r, compute_type: c_f32_r, scaleA: 3, scaleB: 3, scale_type: f32_r}
- { a_type: f8_r, b_type: bf8_r, c_type: f32_r, d_type: f32_r, compute_type: c_f32_r, scaleA: 3, scaleB: 3, scale_type: f32_r}
- { a_type: f8_r, b_type: bf8_r, c_type: f32_r, d_type: f32_r, compute_type: c_f32_r, scaleA: 3, scaleB: 3, scale_type: f32_r, gpu_arch: '950'}
matrix_size:
- { M: 3072, N: 3072, K: 16384 }
- { M: 2560, N: 16384, K: 2560 }
Expand Down
2 changes: 1 addition & 1 deletion projects/hipblaslt/tensilelite/tensilelite/Tests/common/streamk/sk_queue_selfclean.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@
# that do not block evenly over the queues.

TestParameters:
marks: [skip-gfx900, skip-gfx906, skip-gfx908, skip-gfx90a, skip-gfx940, skip-gfx941, skip-gfx1010, skip-gfx1011, skip-gfx1012, skip-gfx1030, skip-gfx1100, skip-gfx1101, skip-gfx1102, skip-gfx1200, skip-gfx1201, skip-gfx1250] # NumXCD is 8 only on gfx942/gfx950; elsewhere the single queue is trivially balanced
marks: [skip-gfx900, skip-gfx906, skip-gfx908, skip-gfx90a, skip-gfx940, skip-gfx941, skip-gfx1010, skip-gfx1011, skip-gfx1012, skip-gfx1030, skip-gfx1100, skip-gfx1101, skip-gfx1102, skip-gfx1200, skip-gfx1201, skip-gfx1250, skip-gfx1250-strict] # NumXCD is 8 only on gfx942/gfx950; elsewhere the single queue is trivially balanced

GlobalParameters:
# DynamicWorkQueue hybrid sub-path; the static sub-path has no work queue.
Expand Down
2 changes: 1 addition & 1 deletion projects/hipblaslt/tensilelite/tensilelite/Tests/custom/custom_rocblas_gemv_bf16_tn_m1.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
# SPDX-License-Identifier: MIT

TestParameters:
marks: [skip-gfx900, skip-gfx906, skip-gfx908, skip-gfx90a, skip-gfx940, skip-gfx941, skip-gfx942, skip-gfx1010, skip-gfx1011, skip-gfx1012, skip-gfx1030, skip-gfx1100, skip-gfx1101, skip-gfx1102, skip-gfx1200, skip-gfx1201, skip-gfx1250] # wvSpltK BF16: gfx950 (v_dot2c_f32_bf16)
marks: [skip-gfx900, skip-gfx906, skip-gfx908, skip-gfx90a, skip-gfx940, skip-gfx941, skip-gfx942, skip-gfx1010, skip-gfx1011, skip-gfx1012, skip-gfx1030, skip-gfx1100, skip-gfx1101, skip-gfx1102, skip-gfx1200, skip-gfx1201, skip-gfx1250, skip-gfx1250-strict] # wvSpltK BF16: gfx950 (v_dot2c_f32_bf16)

GlobalParameters:
MinimumRequiredVersion: 5.0.0
Expand Down
2 changes: 1 addition & 1 deletion projects/hipblaslt/tensilelite/tensilelite/Tests/custom/custom_rocblas_gemv_bf16_tn_m2.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
# SPDX-License-Identifier: MIT

TestParameters:
marks: [skip-gfx900, skip-gfx906, skip-gfx908, skip-gfx90a, skip-gfx940, skip-gfx941, skip-gfx942, skip-gfx1010, skip-gfx1011, skip-gfx1012, skip-gfx1030, skip-gfx1100, skip-gfx1101, skip-gfx1102, skip-gfx1200, skip-gfx1201, skip-gfx1250] # wvSpltK BF16: gfx950 (v_dot2c_f32_bf16)
marks: [skip-gfx900, skip-gfx906, skip-gfx908, skip-gfx90a, skip-gfx940, skip-gfx941, skip-gfx942, skip-gfx1010, skip-gfx1011, skip-gfx1012, skip-gfx1030, skip-gfx1100, skip-gfx1101, skip-gfx1102, skip-gfx1200, skip-gfx1201, skip-gfx1250, skip-gfx1250-strict] # wvSpltK BF16: gfx950 (v_dot2c_f32_bf16)

GlobalParameters:
MinimumRequiredVersion: 5.0.0
Expand Down
2 changes: 1 addition & 1 deletion projects/hipblaslt/tensilelite/tensilelite/Tests/custom/custom_rocblas_gemv_bf16_tn_m4.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
# SPDX-License-Identifier: MIT

TestParameters:
marks: [skip-gfx900, skip-gfx906, skip-gfx908, skip-gfx90a, skip-gfx940, skip-gfx941, skip-gfx942, skip-gfx1010, skip-gfx1011, skip-gfx1012, skip-gfx1030, skip-gfx1100, skip-gfx1101, skip-gfx1102, skip-gfx1200, skip-gfx1201, skip-gfx1250] # wvSpltK BF16: gfx950 (v_dot2c_f32_bf16)
marks: [skip-gfx900, skip-gfx906, skip-gfx908, skip-gfx90a, skip-gfx940, skip-gfx941, skip-gfx942, skip-gfx1010, skip-gfx1011, skip-gfx1012, skip-gfx1030, skip-gfx1100, skip-gfx1101, skip-gfx1102, skip-gfx1200, skip-gfx1201, skip-gfx1250, skip-gfx1250-strict] # wvSpltK BF16: gfx950 (v_dot2c_f32_bf16)

GlobalParameters:
MinimumRequiredVersion: 5.0.0
Expand Down
Loading