diff --git a/.buildkite/test_areas/kernels.yaml b/.buildkite/test_areas/kernels.yaml index 159f940530..ebcb95a9d8 100644 --- a/.buildkite/test_areas/kernels.yaml +++ b/.buildkite/test_areas/kernels.yaml @@ -104,6 +104,7 @@ steps: source_file_dependencies: - csrc/quantization/ - vllm/model_executor/layers/quantization + - vllm/config/ - tests/kernels/quantization - tests/kernels/quantization/test_rocm_skinny_gemms.py - vllm/_aiter_ops.py diff --git a/tests/kernels/quantization/test_triton_scaled_mm.py b/tests/kernels/quantization/test_triton_scaled_mm.py index 1cef5eb93a..d857d495f2 100644 --- a/tests/kernels/quantization/test_triton_scaled_mm.py +++ b/tests/kernels/quantization/test_triton_scaled_mm.py @@ -60,8 +60,10 @@ def test_rocm_compressed_tensors_w8a8( vllm_runner, example_prompts, model_path, max_tokens, num_logprobs ): dtype = "bfloat16" - - with vllm_runner(model_path, dtype=dtype) as vllm_model: + # Pin to TRITON_ATTN, see https://github.com/vllm-project/vllm/issues/46179 + with vllm_runner( + model_path, dtype=dtype, attention_backend="TRITON_ATTN" + ) as vllm_model: vllm_model.generate_greedy_logprobs(example_prompts, max_tokens, num_logprobs)