saymrwulf/pytorch
1
0
Fork 0
mirror of https://github.com/saymrwulf/pytorch.git synced 2026-05-14 20:57:59 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

[CI][CUDA][Dynamic Shape] xfail: DynamicShapesCodegenGPUTests.test_linspace4_dynamic_shapes_cuda (#145204)

Browse source 
python test/inductor/test_torchinductor_codegen_dynamic_shapes.py DynamicShapesCodegenGPUTests.test_linspace4_dynamic_shapes_cuda

failed to generate triton kernels, causing assert failures on 2x H100 systems (and 2x Grace H100 systems).

Failures like below:

Finline_call []                                                                                                                                                    stats [('calls_captured', 1), ('unique_graphs', 1)]
inductor [('fxgraph_cache_miss', 1)]
aot_autograd [('total', 1), ('autograd_cache_miss', 1), ('autograd_cache_saved', 1), ('ok', 1)]

FAIL: test_linspace4_dynamic_shapes_cuda (__main__.DynamicShapesCodegenGPUTests.test_linspace4_dynamic_shapes_cuda)                                       [61/1892]----------------------------------------------------------------------                                                                                             Traceback (most recent call last):                                                                                                                                   File "/usr/local/lib/python3.12/dist-packages/torch/testing/_internal/common_utils.py", line 3114, in wrapper
    method(*args, **kwargs)
  File "/opt/pytorch/pytorch/test/inductor/test_torchinductor.py", line 12212, in new_test
    return value(self)
           ^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/torch/_dynamo/testing.py", line 420, in _fn
    return fn(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^
  File "/opt/pytorch/pytorch/test/inductor/test_torchinductor.py", line 2603, in test_linspace4
    self.common(fn, (torch.Tensor([]),))
  File "/opt/pytorch/pytorch/test/inductor/test_torchinductor_codegen_dynamic_shapes.py", line 424, in common
    return check_codegen(
           ^^^^^^^^^^^^^^
  File "/opt/pytorch/pytorch/test/inductor/test_torchinductor_codegen_dynamic_shapes.py", line 82, in check_codegen
    self.assertTrue("def triton" in code, f"Failed to find triton kernel\n{code}")
AssertionError: False is not true : Failed to find triton kernel

# AOT ID: ['0_inference']                                                                                                                                 [42/1892]from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall

aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
empty_strided_p2p = torch._C._distributed_c10d._SymmetricMemory.empty_strided_p2p

async_compile.wait(globals())
del async_compile

def call(args):
    with torch.cuda._DeviceGuard(1):
        torch.cuda.set_device(1)
        buf0 = empty_strided_cuda((0, ), (1, ), torch.float32)
    return (buf0, )

def benchmark_compiled_module(times=10, repeat=10):
    from torch._dynamo.testing import rand_strided
    from torch._inductor.utils import print_performance
    fn = lambda: call([])
    return print_performance(fn, times=times, repeat=repeat)

if __name__ == "__main__":
    from torch._inductor.wrapper_benchmark import compiled_module_main
    compiled_module_main('None', benchmark_compiled_module)

To execute this test, run the following from the base repo dir:
    python test/inductor/test_torchinductor_codegen_dynamic_shapes.py DynamicShapesCodegenGPUTests.test_linspace4_dynamic_shapes_cuda

This message can be suppressed by setting PYTORCH_PRINT_REPRO_ON_FAILURE=0

Pull Request resolved: https://github.com/pytorch/pytorch/pull/145204
Approved by: https://github.com/eellison
This commit is contained in:
Wei Wang 2025-01-24 06:57:35 +00:00 committed by PyTorch MergeBot
parent 817fd14714
commit 0e98b26b28
1 changed files with 1 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
test/inductor/test_torchinductor_codegen_dynamic_shapes.py
View file

@ -169,6 +169,7 @@ test_failures = {
"test_like_rands_dynamic_shapes": TestFailure(("cpu", "cuda", "xpu")),
"test_linspace2_dynamic_shapes": TestFailure(("cpu", "cuda", "xpu")),
"test_linspace3_dynamic_shapes": TestFailure(("cpu", "cuda", "xpu")),
"test_linspace4_dynamic_shapes": TestFailure(("cpu", "cuda", "xpu")),
"test_logcumsumexp_dynamic_shapes": TestFailure(("cpu",)),
"test_logcumsumexp_zero_dim_dynamic_shapes": TestFailure(("cpu",)),
"test_max_pool2d6_dynamic_shapes": TestFailure(("cpu", "cuda", "xpu")),