Triton Fusion Emitter Test Failed

[amd-jianli12]

[  FAILED  ] TritonEmitterTest.FusionWithOutputContainingMoreThanInt32MaxElementsExecutesCorrectly

[amd-jianli12]

#2821

[amd-jianli12]

[2025-01-21T08:11:14.484Z] [ RUN      ] TritonEmitterTest.FusionWithOutputContainingMoreThanInt32MaxElementsExecutesCorrectly
[2025-01-21T08:11:14.484Z] 2025-01-21 08:11:12.064912: W external/local_xla/xla/service/compiler.h:209] Ignoring the buffer assignment proto provided.
[2025-01-21T08:11:14.484Z] WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
[2025-01-21T08:11:14.484Z] W0000 00:00:1737447072.209709  138864 stream_executor.h:364] Not enough memory to allocate 4294967552 on device 0 within provided limit.  limit=4294967296]
[2025-01-21T08:11:14.484Z] external/local_xla/xla/backends/gpu/codegen/triton/fusion_emitter_device_test.cc:1497: Failure
[2025-01-21T08:11:14.484Z] Value of: RunAndCompareTwoModules( std::move(triton_module), std::move(emitters_module), ErrorSpec{ 0, 0}, false)
[2025-01-21T08:11:14.484Z]   Actual: false (RESOURCE_EXHAUSTED: Failed to allocate request for 4.00GiB (4294967552B) on device ordinal 0)
[2025-01-21T08:11:14.484Z] Expected: true
[2025-01-21T08:11:14.484Z] [  FAILED  ] TritonEmitterTest.FusionWithOutputContainingMoreThanInt32MaxElementsExecutesCorrectly (147 ms)

[amd-jianli12]

@local_xla//xla/backends/gpu/codegen/triton:fusion_emitter_device_test_gpu_amd_any

[amd-jianli12]

PR-2821.txt