A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 13:30 UTC
Fencing Kernels Tests for enabling on AMD (#16929)
Signed-off-by: Alexei V. Ivanov <alexei.ivanov@amd.com>
Alexei-V-Ivanov-AMD committed
Apr 22, 2025 at 11:32 UTC
5536b30a4c7877d75758d21bdaf39b3a59aa2dc2
2 files changed
+8
-1
.buildkite/scripts/hardware_ci/run-amd-test.sh
+7
@@ -98,6 +98,13 @@ if [[ $commands == *" kernels "* ]]; then
98
--ignore=kernels/test_machete_mm.py \
99
--ignore=kernels/test_mha_attn.py \
100
--ignore=kernels/test_block_fp8.py \
101
+ --ignore=kernels/test_cutlass_moe.py \
102
+ --ignore=kernels/test_mamba_ssm_ssd.py \
103
+ --ignore=kernels/test_attention.py \
104
+ --ignore=kernels/test_block_int8.py \
105
+ --ignore=kernels/test_fused_quant_layernorm.py \
106
+ --ignore=kernels/test_int8_kernel.py \
107
+ --ignore=kernels/test_triton_moe_ptpc_fp8.py \
108
--ignore=kernels/test_permute_cols.py"
109
fi
110
.buildkite/test-pipeline.yaml
+1
-1
@@ -314,7 +314,7 @@ steps:
314
- pytest -v -s compile/test_full_graph.py
315
316
- label: Kernels Test %N # 1h each
317
- # mirror_hardwares: [amd]
317
+ mirror_hardwares: [amd]
318
source_file_dependencies:
319
- csrc/
320
- vllm/attention