From 059a34468c6c2753ed792f8bb6df38940a788968 Mon Sep 17 00:00:00 2001 From: QI JUN <22017000+QiJune@users.noreply.github.com> Date: Fri, 4 Apr 2025 16:19:02 +0800 Subject: [PATCH] fix deepseek multi gpu tests timeout (#3285) Signed-off-by: junq <22017000+QiJune@users.noreply.github.com> --- tests/integration/test_lists/test-db/l0_dgx_h100.yml | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/tests/integration/test_lists/test-db/l0_dgx_h100.yml b/tests/integration/test_lists/test-db/l0_dgx_h100.yml index 666d293fc6..0b2707465e 100644 --- a/tests/integration/test_lists/test-db/l0_dgx_h100.yml +++ b/tests/integration/test_lists/test-db/l0_dgx_h100.yml @@ -15,8 +15,12 @@ l0_dgx_h100: tests: # ------------- PyTorch tests --------------- - unittest/_torch/multi_gpu - - unittest/_torch/multi_gpu_modeling -k "deepseek and not (tp1 and pp1) and nextn0" - - unittest/_torch/multi_gpu_modeling -k "deepseek and not (tp1 and pp1) and not nextn0" + - unittest/_torch/multi_gpu_modeling -k "deepseek and tp4_pp1_ep1_nextn0" + - unittest/_torch/multi_gpu_modeling -k "deepseek and tp4_pp1_ep1_nextn2" + - unittest/_torch/multi_gpu_modeling -k "deepseek and tp4_pp1_ep4_nextn0" + - unittest/_torch/multi_gpu_modeling -k "deepseek and tp4_pp1_ep4_nextn2" + - unittest/_torch/multi_gpu_modeling -k "deepseek and tp2_pp2_ep1_nextn0" + - unittest/_torch/multi_gpu_modeling -k "deepseek and tp2_pp2_ep1_nextn2" - unittest/_torch/multi_gpu_modeling -k "llama and not (tp1 and pp1)" - unittest/_torch/auto_deploy/unit/multigpu - disaggregated/test_disaggregated.py::test_disaggregated_multi_gpu_with_mpirun[TinyLlama-1.1B-Chat-v1.0]