[RFC,TESTING] Add filecheck test infrastructure #342

ashermancinelli · 2025-07-22T21:28:48Z

Adding pattern-based checking infrastructure allows us to describe the IR and assembly expected under different conditions right next to the code that produces it. This is closer to the LLVM testing standards, and will help is iterate more quickly and add test coverage while we are in the process of vendoring in the components from numba core we need to modify in a cuda-specific way.

This PR changes a couple tests to use the filecheck patterns. For example, in numba-cuda today there tests like self.assertIn('function_name', llvm_ir). Now, we can express the same check in the docstring of the kernel, and all the checks we would like to perform may live inside the kernel they come from.

This is a request-for-comment, as filecheck tests will be meaningfully different from existing numba-cuda tests.

Note also that the filecheck that has been added to the list of dependencies is not the filecheck that is compiled as a part of the LLVM project, but is an attempt to replicate the functionality in a pure python package.
I ran some tests to see how well the usual FileCheck features work with the Python version, and it seems okay.
If we are uncomfortable with this development dependency, we can reevaluate.

TODO:

CUDA Test Case should have file check available by default
Review type-hint inconsistencies
Skip filecheck-based tests in python 3.9
Add dummy class for CUDADispatcher when running with the simulator

Another solution I considered would be to structure our tests much more like LLVM lit tests, not just using the filecheck pattern like I did in this patch.

In this alternative, I imagine test kernels would be written in their own files:

# RUN: numba-cuda-to-llvm-ir %s | filecheck %s
# ^^ the above line converts the JIT'ed kernels in this Python module to
#    LLVM IR and checks the check-comments in the module against the IR

from numba.cuda import jit

# CHECK-LABEL: foo
# CHECK: fadd
@jit((float, float))
def test_foo(a, b):
    a = b

copy-pr-bot · 2025-07-22T21:28:51Z

Auto-sync is disabled for ready for review pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

isVoid

I'm trying to understand the use of this tool. Also a broader question, the result of the compiler are in fact legal nvvm IRs, should we use NVVM as the prefix instead?

numba_cuda/numba/cuda/tests/cudapy/test_inspect.py

ashermancinelli · 2025-07-22T23:06:56Z

should we use NVVM [instead of LLVM] as the prefix instead?

Sure! Whatever will most precisely differentiate the checks from one another.

ashermancinelli · 2025-07-22T23:16:22Z

@isVoid I tried answering all your questions, but let me speak about the motivation more generally: FileCheck allows us to easily create meaningful tests (in a semi-automated way) to increase the test coverage we have so we feel more secure making larger changes. Consider this test in LLVM Flang. There are many ways to compile the same test, some of which result in the same IR. The same source code can be reused to create 6 or so test cases, some of which share the same checks. There are scripts in the LLVM project to generate and update these tests automatically.

When we vendor in components for the purpose of cuda-specific extensions, instead of vendoring in the unit tests from numba core into numba cuda directly, we can vendor in the kernels and use cuda.jit instead of the regular numba jit, and ensure the IR is the same. Then, as we change the components, we can be sure we have not changed from the IR we were getting with the non-vendored version (or if we have changed the IR, it is in the way we expect).

This is also just a suggestion/RFC. We will discuss this further before submitting any changes. Thank you for the review and please ask more questions if I left anything unclear!

numba_cuda/numba/cuda/testing.py

pyproject.toml

numba_cuda/numba/cuda/testing.py

numba_cuda/numba/cuda/tests/cudapy/test_inspect.py

gmarkall

I really like the idea here! This looks like a nice way to check the generated code.

I do think we could simplify the implementation and API, and make it a bit closer to the idioms of the rest of the test suite - see comments on the diff for these suggestions.

I also wonder - does FileCheck identify the failing line and provide error messages on stderr? I notice the Matcher class has an stderr member but I haven't dived deeper - I am looking to understand whether the framework can provide more precise error messages when failures occur.

ashermancinelli · 2025-07-23T13:32:03Z

I also wonder - does FileCheck identify the failing line and provide error messages on stderr?

It does - but please note that this is a Python re-implementation of the FileCheck executable program which is a part of the LLVM project, so I am also unfamiliar with the API of this Python module. I will investigate further and see how we can make this more ergonomic. Thank you for the review!

Adding pattern-based checking infrastructure allows us to describe the IR and assembly expected under different conditions and expressed right next to the code that produced it. This is closer to the LLVM testing standards, and will help is iterate more quickly and add test coverage while we are in the process of vendoring in components from numba core. This is a request for comment, as filecheck tests will be meaningfully different from existing numba-cuda tests.

numba_cuda/numba/cuda/testing.py

ashermancinelli · 2025-07-23T15:33:59Z

/ok to test c45924d

pyproject.toml

ashermancinelli · 2025-07-23T15:50:51Z

/ok to test df00dc7

ashermancinelli · 2025-07-23T16:10:59Z

/ok to test 0552773

isVoid · 2025-07-23T16:13:54Z

@ashermancinelli thanks for the explanation. When we write tests, we are limited by the tool that we have so that we can't test more complicated pattern. This opens up a whole range of options for us. Just a simple grab through the code base with self\.assertIn\(.*, (asm|ptx|sass|ir|llvm|llvmirs|llvmirs.*)\) we have these tests that can potentially reuse this test infra.

Details

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudadrv/test_nvvm_driver.py
  107,9:         self.assertIn(".target sm_{0}{1}".format(*arch), ptx)
  108,9:         self.assertIn("simple", ptx)
  109,9:         self.assertIn("ave", ptx)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_atomics.py
  691,17:                 self.assertIn("atom.shared.cas.b64", asm)
  693,17:                 self.assertIn("atom.cas.b64", asm)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_casting.py
  138,13:             self.assertIn(f"cvt.rni.s{size}.f16", ptx)
  160,13:             self.assertIn(f"cvt.rni.u{size}.f16", ptx)
  192,13:             self.assertIn(f"cvt.rn.f16.s{size}", ptx)
  201,13:             self.assertIn(f"cvt.rn.f16.u{size}", ptx)
  229,13:             self.assertIn(f"cvt.{postfix}.f16", ptx)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_compiler.py
  35,9:         self.assertIn(".visible .entry", ptx)
  48,9:         self.assertIn("func_retval", ptx)
  50,9:         self.assertIn(".visible .func", ptx)
  78,9:         self.assertIn("fma.rn.f32", ptx)
  79,9:         self.assertIn("div.rn.f32", ptx)
  80,9:         self.assertIn("sqrt.rn.f32", ptx)
  85,9:         self.assertIn("fma.rn.ftz.f32", ptx)
  86,9:         self.assertIn("div.approx.ftz.f32", ptx)
  87,9:         self.assertIn("sqrt.approx.ftz.f32", ptx)
  270,9:         self.assertIn(target, ptx)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_device_func.py
  129,9:         self.assertIn(fname, llvm)
  146,9:         self.assertIn(fname, ptx)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_inspect.py
  36,9:         self.assertIn("foo", llvm)
  39,9:         self.assertIn("define void @", llvm)
  44,9:         self.assertIn("foo", asm)
  46,9:         self.assertIn("Generated by NVIDIA NVVM Compiler", asm)
  70,9:         self.assertIn((intp, intp), llvmirs)
  71,9:         self.assertIn((float64, float64), llvmirs)
  74,9:         self.assertIn("foo", llvmirs[intp, intp])
  75,9:         self.assertIn("foo", llvmirs[float64, float64])
  78,9:         self.assertIn("define void @", llvmirs[intp, intp])
  79,9:         self.assertIn("define void @", llvmirs[float64, float64])
  107,9:         self.assertIn("S2R", sass)  # Special register to register
  108,9:         self.assertIn("BRA", sass)  # Branch
  109,9:         self.assertIn("EXIT", sass)  # Exit program

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_intrinsics.py
  645,9:         self.assertIn("add.f16", ptx)
  672,9:         self.assertIn("fma.rn.f16", ptx)
  697,9:         self.assertIn("sub.f16", ptx)
  722,9:         self.assertIn("mul.f16", ptx)
  767,9:         self.assertIn("neg.f16", ptx)
  790,9:         self.assertIn("abs.f16", ptx)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_libdevice.py
  165,9:         self.assertIn("ld.param", ptx)
  169,9:         self.assertIn("st.global", ptx)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_lineinfo.py
  216,9:         self.assertIn(target, ptx)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_minmax.py
  46,9:         self.assertIn(ptx_instruction, ptx)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_operator.py
  182,17:                 self.assertIn(instr, ptx)
  216,17:                 self.assertIn(instr, ptx)
  259,9:         self.assertIn("neg.f16", ptx)
  266,9:         self.assertIn("abs.f16", ptx)
  400,17:                 self.assertIn(s, ptx)
  435,17:                 self.assertIn(opstring[op], ptx)
  481,17:                 self.assertIn(ops, ptx)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_optimization.py
  70,17:                 self.assertIn(fragment, ptx)
  77,9:         self.assertIn("fma.rn.f64", ptx)

/workspace/numbast-nvshmem/numba-cuda/numba_cuda/numba/cuda/tests/nocuda/test_nvvm.py
  52,9:         self.assertIn(myconstant, ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudadrv/test_nvvm_driver.py
  107,9:         self.assertIn(".target sm_{0}{1}".format(*arch), ptx)
  108,9:         self.assertIn("simple", ptx)
  109,9:         self.assertIn("ave", ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_atomics.py
  691,17:                 self.assertIn("atom.shared.cas.b64", asm)
  693,17:                 self.assertIn("atom.cas.b64", asm)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_casting.py
  138,13:             self.assertIn(f"cvt.rni.s{size}.f16", ptx)
  160,13:             self.assertIn(f"cvt.rni.u{size}.f16", ptx)
  192,13:             self.assertIn(f"cvt.rn.f16.s{size}", ptx)
  201,13:             self.assertIn(f"cvt.rn.f16.u{size}", ptx)
  229,13:             self.assertIn(f"cvt.{postfix}.f16", ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_compiler.py
  35,9:         self.assertIn(".visible .entry", ptx)
  48,9:         self.assertIn("func_retval", ptx)
  50,9:         self.assertIn(".visible .func", ptx)
  78,9:         self.assertIn("fma.rn.f32", ptx)
  79,9:         self.assertIn("div.rn.f32", ptx)
  80,9:         self.assertIn("sqrt.rn.f32", ptx)
  85,9:         self.assertIn("fma.rn.ftz.f32", ptx)
  86,9:         self.assertIn("div.approx.ftz.f32", ptx)
  87,9:         self.assertIn("sqrt.approx.ftz.f32", ptx)
  270,9:         self.assertIn(target, ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_device_func.py
  129,9:         self.assertIn(fname, llvm)
  146,9:         self.assertIn(fname, ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_inspect.py
  36,9:         self.assertIn("foo", llvm)
  39,9:         self.assertIn("define void @", llvm)
  44,9:         self.assertIn("foo", asm)
  46,9:         self.assertIn("Generated by NVIDIA NVVM Compiler", asm)
  70,9:         self.assertIn((intp, intp), llvmirs)
  71,9:         self.assertIn((float64, float64), llvmirs)
  74,9:         self.assertIn("foo", llvmirs[intp, intp])
  75,9:         self.assertIn("foo", llvmirs[float64, float64])
  78,9:         self.assertIn("define void @", llvmirs[intp, intp])
  79,9:         self.assertIn("define void @", llvmirs[float64, float64])
  107,9:         self.assertIn("S2R", sass)  # Special register to register
  108,9:         self.assertIn("BRA", sass)  # Branch
  109,9:         self.assertIn("EXIT", sass)  # Exit program

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_intrinsics.py
  645,9:         self.assertIn("add.f16", ptx)
  672,9:         self.assertIn("fma.rn.f16", ptx)
  697,9:         self.assertIn("sub.f16", ptx)
  722,9:         self.assertIn("mul.f16", ptx)
  767,9:         self.assertIn("neg.f16", ptx)
  790,9:         self.assertIn("abs.f16", ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_libdevice.py
  165,9:         self.assertIn("ld.param", ptx)
  169,9:         self.assertIn("st.global", ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_lineinfo.py
  216,9:         self.assertIn(target, ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_minmax.py
  46,9:         self.assertIn(ptx_instruction, ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_operator.py
  182,17:                 self.assertIn(instr, ptx)
  216,17:                 self.assertIn(instr, ptx)
  259,9:         self.assertIn("neg.f16", ptx)
  266,9:         self.assertIn("abs.f16", ptx)
  400,17:                 self.assertIn(s, ptx)
  435,17:                 self.assertIn(opstring[op], ptx)
  481,17:                 self.assertIn(ops, ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/cudapy/test_optimization.py
  70,17:                 self.assertIn(fragment, ptx)
  77,9:         self.assertIn("fma.rn.f64", ptx)

/workspace/numba-cuda/numba_cuda/numba/cuda/tests/nocuda/test_nvvm.py
  52,9:         self.assertIn(myconstant, ptx)

ci/test_conda.sh

numba_cuda/numba/cuda/testing.py

Co-authored-by: Graham Markall <[email protected]>

gmarkall · 2025-07-23T16:22:25Z

numba_cuda/numba/cuda/testing.py

+        self,
+        ir_producer: CUDADispatcher,
+        signature: tuple[type, ...] | None = None,
+        check_prefixes: list[str] = ["ASM"],


This use of a list as a default argument is probably safe as it's not mutated, but it always sets off alarm bells to see a mutable default in a Python function definition. I think you could either use None as in https://docs.python-guide.org/writing/gotchas/#mutable-default-arguments, or have a tuple for the default instead (I assume filecheck will accept a tuple as well).

A tuple or frozenset would be okay with me, let me try that out. Thanks!

ashermancinelli · 2025-07-23T16:23:43Z

/ok to test 8d6dcb8

numba_cuda/numba/cuda/testing.py

ashermancinelli · 2025-07-23T21:13:08Z

The filecheck module compatible with python 3.9 is too old to be used here.

ashermancinelli · 2025-07-23T23:45:23Z

/ok to test 8fb7920

ashermancinelli · 2025-07-24T00:00:18Z

/ok to test dd6eb32

gmarkall · 2025-07-24T12:39:13Z

/ok to test

gmarkall · 2025-07-24T12:40:31Z

I think the PR is good. I am just running another check with a merge of main to make sure that all is well with pytest being the test framework before merging this PR.

- [NFC] FileCheck tests check all overloads (NVIDIA#354) - [REVIEW][NFC] Vendor in serialize to allow for future CUDA-specific refactoring and changes (NVIDIA#349) - Vendor in usecases used in testing (NVIDIA#359) - Add thirdparty tests of numba extensions (NVIDIA#348) - Support running tests in parallel (NVIDIA#350) - Add more debuginfo tests (NVIDIA#358) - [REVIEW][NFC] Vendor in the Cache, CacheImpl used by CUDACache and CUDACacheImpl to allow for future CUDA-specific refactoring and changes (NVIDIA#334) - [NFC] Vendor in Dispatcher as CUDADispatcher to allow for future CUDA-specific customization (NVIDIA#338) - Vendor in BaseNativeLowering and BaseLower for CUDA-specific customizations (NVIDIA#329) - [REVIEW] Vendor in the CompilerBase used by CUDACompiler to allow for future CUDA-specific refactoring and changes (NVIDIA#322) - Vendor in Codegen and CodeLibrary for CUDA-specific customization (NVIDIA#327) - Disable tests that deadlock due to NVIDIA#317 (NVIDIA#356) - FIX: Add type check for shape elements in DeviceNDArrayBase constructor (NVIDIA#352) - Merge pull request NVIDIA#265 from lakshayg/fp16-support - Add performance warning - Fix tests - Create and register low++ bindings for float16 - Create typing/target registries for float16 - Replace Numbast generated lower_casts - Replace Numbast generated operators - Alias __half to numba.core.types.float16 - Generate fp16 bindings using numbast - Remove existing fp16 logic - [REVIEW][NFC] Vendor in the utils and cgutils to allow for future CUDA-specific refactoring and changes (NVIDIA#340) - [RFC,TESTING] Add filecheck test infrastructure (NVIDIA#342) - Migrate test infra to pytest (NVIDIA#347) - Add .vscode to gitignore (NVIDIA#344) - [NFC] Add dev dependencies to project config (NVIDIA#341) - Allow Inspection of Link-Time Optimized PTX (NVIDIA#326) - [NFC] Vendor in DIBuilder used by CUDADIBuilder (NVIDIA#332) - Add guidance on setting up pre-commit (NVIDIA#339) - [Refactor][NFC] Vendor in MinimalCallConv (NVIDIA#333) - [Refactor][NFC] Vendor in BaseCallConv (NVIDIA#324) - [REVIEW] Vendor in CompileResult as CUDACompileResult to allow for future CUDA-specific customizations (NVIDIA#325)

- [NFC] FileCheck tests check all overloads (#354) - [REVIEW][NFC] Vendor in serialize to allow for future CUDA-specific refactoring and changes (#349) - Vendor in usecases used in testing (#359) - Add thirdparty tests of numba extensions (#348) - Support running tests in parallel (#350) - Add more debuginfo tests (#358) - [REVIEW][NFC] Vendor in the Cache, CacheImpl used by CUDACache and CUDACacheImpl to allow for future CUDA-specific refactoring and changes (#334) - [NFC] Vendor in Dispatcher as CUDADispatcher to allow for future CUDA-specific customization (#338) - Vendor in BaseNativeLowering and BaseLower for CUDA-specific customizations (#329) - [REVIEW] Vendor in the CompilerBase used by CUDACompiler to allow for future CUDA-specific refactoring and changes (#322) - Vendor in Codegen and CodeLibrary for CUDA-specific customization (#327) - Disable tests that deadlock due to #317 (#356) - FIX: Add type check for shape elements in DeviceNDArrayBase constructor (#352) - Merge pull request #265 from lakshayg/fp16-support - Add performance warning - Fix tests - Create and register low++ bindings for float16 - Create typing/target registries for float16 - Replace Numbast generated lower_casts - Replace Numbast generated operators - Alias __half to numba.core.types.float16 - Generate fp16 bindings using numbast - Remove existing fp16 logic - [REVIEW][NFC] Vendor in the utils and cgutils to allow for future CUDA-specific refactoring and changes (#340) - [RFC,TESTING] Add filecheck test infrastructure (#342) - Migrate test infra to pytest (#347) - Add .vscode to gitignore (#344) - [NFC] Add dev dependencies to project config (#341) - Allow Inspection of Link-Time Optimized PTX (#326) - [NFC] Vendor in DIBuilder used by CUDADIBuilder (#332) - Add guidance on setting up pre-commit (#339) - [Refactor][NFC] Vendor in MinimalCallConv (#333) - [Refactor][NFC] Vendor in BaseCallConv (#324) - [REVIEW] Vendor in CompileResult as CUDACompileResult to allow for future CUDA-specific customizations (#325)

isVoid reviewed Jul 22, 2025

View reviewed changes

numba_cuda/numba/cuda/tests/cudapy/test_inspect.py Show resolved Hide resolved

numba_cuda/numba/cuda/tests/cudapy/test_inspect.py Outdated Show resolved Hide resolved

numba_cuda/numba/cuda/tests/cudapy/test_inspect.py Show resolved Hide resolved

gmarkall reviewed Jul 23, 2025

View reviewed changes

numba_cuda/numba/cuda/testing.py Outdated Show resolved Hide resolved

gmarkall reviewed Jul 23, 2025

View reviewed changes

pyproject.toml Show resolved Hide resolved

gmarkall reviewed Jul 23, 2025

View reviewed changes

numba_cuda/numba/cuda/testing.py Outdated Show resolved Hide resolved

gmarkall reviewed Jul 23, 2025

View reviewed changes

numba_cuda/numba/cuda/testing.py Outdated Show resolved Hide resolved

gmarkall reviewed Jul 23, 2025

View reviewed changes

numba_cuda/numba/cuda/tests/cudapy/test_inspect.py Outdated Show resolved Hide resolved

gmarkall requested changes Jul 23, 2025

View reviewed changes

gmarkall added the 4 - Waiting on author Waiting for author to respond to review label Jul 23, 2025

ashermancinelli self-assigned this Jul 23, 2025

ashermancinelli requested review from VijayKandiah and atmnp July 23, 2025 13:32

ashermancinelli added 4 commits July 23, 2025 06:55

Fix type annotation

4e122a7

Fix type annotations

130d01d

Move filecheck utilities into a test case mixin

398233c

ashermancinelli force-pushed the ajm/filecheck branch from 0146b43 to 398233c Compare July 23, 2025 14:21

gmarkall reviewed Jul 23, 2025

View reviewed changes

numba_cuda/numba/cuda/testing.py Outdated Show resolved Hide resolved

FileCheck is available to CUDATestCase by default

c45924d

ashermancinelli added 3 - Ready for Review Ready for review by team and removed 4 - Waiting on author Waiting for author to respond to review labels Jul 23, 2025

gmarkall reviewed Jul 23, 2025

View reviewed changes

pyproject.toml Outdated Show resolved Hide resolved

Fix missing dependency

df00dc7

Install filecheck in conda CI scripts

0552773

gmarkall reviewed Jul 23, 2025

View reviewed changes

ci/test_conda.sh Show resolved Hide resolved

gmarkall reviewed Jul 23, 2025

View reviewed changes

numba_cuda/numba/cuda/testing.py Outdated Show resolved Hide resolved

Update numba_cuda/numba/cuda/testing.py

c4f2d5b

Co-authored-by: Graham Markall <[email protected]>

gmarkall reviewed Jul 23, 2025

View reviewed changes

Install filecheck in conda CI scripts

8d6dcb8

gmarkall reviewed Jul 23, 2025

View reviewed changes

numba_cuda/numba/cuda/testing.py Outdated Show resolved Hide resolved

Use self.fail instead of assert

314002e

gmarkall reviewed Jul 23, 2025

View reviewed changes

numba_cuda/numba/cuda/testing.py Show resolved Hide resolved

ashermancinelli mentioned this pull request Jul 23, 2025

[NFC] Document development dependencies and dependency groups #345

Merged

ashermancinelli added 2 - In Progress Currently a work in progress and removed 3 - Ready for Review Ready for review by team labels Jul 23, 2025

Add CUDADispatcher dummy class for simulator

8fb7920

ashermancinelli added 3 - Ready for Review Ready for review by team and removed 2 - In Progress Currently a work in progress labels Jul 23, 2025

Use dummy import for dispatcher

dd6eb32

gmarkall approved these changes Jul 24, 2025

View reviewed changes

Merge remote-tracking branch 'NVIDIA/main' into ajm/filecheck

5332139

gmarkall added 5 - Ready to merge Testing and reviews complete, ready to merge and removed 3 - Ready for Review Ready for review by team labels Jul 24, 2025

gmarkall merged commit 74439bd into NVIDIA:main Jul 24, 2025
39 checks passed

gmarkall mentioned this pull request Jul 31, 2025

Bump version to 0.18.0 #365

Merged

[RFC,TESTING] Add filecheck test infrastructure #342

[RFC,TESTING] Add filecheck test infrastructure #342

Uh oh!

Conversation

ashermancinelli commented Jul 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

copy-pr-bot bot commented Jul 22, 2025

Uh oh!

isVoid left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ashermancinelli commented Jul 22, 2025

Uh oh!

ashermancinelli commented Jul 22, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

gmarkall left a comment

Choose a reason for hiding this comment

Uh oh!

ashermancinelli commented Jul 23, 2025

Uh oh!

Uh oh!

ashermancinelli commented Jul 23, 2025

Uh oh!

Uh oh!

ashermancinelli commented Jul 23, 2025

Uh oh!

ashermancinelli commented Jul 23, 2025

Uh oh!

isVoid commented Jul 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

gmarkall Jul 23, 2025

Choose a reason for hiding this comment

Uh oh!

ashermancinelli Jul 23, 2025

Choose a reason for hiding this comment

Uh oh!

ashermancinelli commented Jul 23, 2025

Uh oh!

Uh oh!

Uh oh!

ashermancinelli commented Jul 23, 2025

Uh oh!

ashermancinelli commented Jul 23, 2025

Uh oh!

ashermancinelli commented Jul 24, 2025

Uh oh!

gmarkall commented Jul 24, 2025

Uh oh!

gmarkall commented Jul 24, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

ashermancinelli commented Jul 22, 2025 •

edited

Loading

isVoid commented Jul 23, 2025 •

edited

Loading