diff --git a/test/srt/nightly/test_deepseek_v31_perf.py b/test/srt/nightly/test_deepseek_v31_perf.py index a618ad190..58614350c 100644 --- a/test/srt/nightly/test_deepseek_v31_perf.py +++ b/test/srt/nightly/test_deepseek_v31_perf.py @@ -25,8 +25,8 @@ class TestNightlyDeepseekV31Performance(unittest.TestCase): "--trust-remote-code", "--tp", "8", - "--dp", - "8", + "--model-loader-extra-config", + '{"enable_multithread_load": true}', ], }, { @@ -35,8 +35,6 @@ class TestNightlyDeepseekV31Performance(unittest.TestCase): "--trust-remote-code", "--tp", "8", - "--dp", - "8", "--speculative-algorithm", "EAGLE", "--speculative-num-steps", @@ -47,6 +45,8 @@ class TestNightlyDeepseekV31Performance(unittest.TestCase): "4", "--mem-frac", "0.7", + "--model-loader-extra-config", + '{"enable_multithread_load": true}', ], }, ] diff --git a/test/srt/nightly/test_deepseek_v32_perf.py b/test/srt/nightly/test_deepseek_v32_perf.py index cf27dd99e..f7ed778c0 100644 --- a/test/srt/nightly/test_deepseek_v32_perf.py +++ b/test/srt/nightly/test_deepseek_v32_perf.py @@ -25,8 +25,8 @@ class TestNightlyDeepseekV32Performance(unittest.TestCase): "--trust-remote-code", "--tp", "8", - "--dp", - "8", + "--model-loader-extra-config", + '{"enable_multithread_load": true}', ], }, { @@ -35,8 +35,6 @@ class TestNightlyDeepseekV32Performance(unittest.TestCase): "--trust-remote-code", "--tp", "8", - "--dp", - "8", "--speculative-algorithm", "EAGLE", "--speculative-num-steps", @@ -47,6 +45,8 @@ class TestNightlyDeepseekV32Performance(unittest.TestCase): "4", "--mem-frac", "0.7", + "--model-loader-extra-config", + '{"enable_multithread_load": true}', ], }, { @@ -55,14 +55,14 @@ class TestNightlyDeepseekV32Performance(unittest.TestCase): "--trust-remote-code", "--tp", "8", - "--dp", - "8", "--attention-backend", "nsa", "--nsa-prefill-backend", "flashmla_sparse", "--nsa-decode-backend", "flashmla_kv", + "--model-loader-extra-config", + '{"enable_multithread_load": true}', ], }, ]