From 82d6b59f0411b70ad2b5a74a24f808270e0af588 Mon Sep 17 00:00:00 2001 From: Dan Blanaru <48605845+DanBlanaru@users.noreply.github.com> Date: Thu, 11 Jun 2026 00:18:42 +0200 Subject: [PATCH] [CI/Build] Skip test_use_trtllm_attention on non-CUDA platforms (#44687) Signed-off-by: Dan Blanaru <48605845+DanBlanaru@users.noreply.github.com> Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com> Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com> --- tests/kernels/attention/test_use_trtllm_attention.py | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/tests/kernels/attention/test_use_trtllm_attention.py b/tests/kernels/attention/test_use_trtllm_attention.py index fba18fe46e..89ff86b47b 100644 --- a/tests/kernels/attention/test_use_trtllm_attention.py +++ b/tests/kernels/attention/test_use_trtllm_attention.py @@ -6,12 +6,19 @@ from unittest.mock import patch import pytest import torch +from vllm.platforms import current_platform from vllm.utils.flashinfer import ( can_use_trtllm_attention, supports_trtllm_attention, use_trtllm_attention, ) +if not current_platform.is_cuda(): + pytest.skip( + "TRTLLM attention is only supported on CUDA platforms.", + allow_module_level=True, + ) + MODEL_CONFIGS = { "Llama-3-70B": dict(num_qo_heads=64, num_kv_heads=8), "Llama-3-8B": dict(num_qo_heads=32, num_kv_heads=8),