EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

Improve profiler and integrate profiler in bench_one_batch_server (#6787)

Browse Source
This commit is contained in:
Lianmin Zheng
2025-05-31 15:53:55 -07:00
committed by GitHub
parent b520d02888
commit 2d72fc47cf
25 changed files with 481 additions and 223 deletions
Download Patch File Download Diff File Expand all files Collapse all files

49
test/srt/run_suite.py
View File

@@ -16,7 +16,8 @@ suites = {
TestFile("models/lora/test_lora.py", 76),
TestFile("models/lora/test_lora_backend.py", 99),
TestFile("models/lora/test_multi_lora_backend.py", 60),
TestFile("models/test_embedding_models.py", 184),
TestFile("models/lora/test_lora_cuda_graph.py", 250),
TestFile("models/test_embedding_models.py", 73),
# TestFile("models/test_clip_models.py", 52),
TestFile("models/test_compressed_tensors_models.py", 42),
TestFile("models/test_generation_models.py", 103),
@@ -24,44 +25,43 @@ suites = {
# TestFile("models/test_grok_models.py", 60), # Disabled due to illegal memory access
TestFile("models/test_qwen_models.py", 82),
TestFile("models/test_reward_models.py", 132),
TestFile("models/test_vlm_models.py", 317),
TestFile("models/test_vlm_models.py", 437),
TestFile("test_abort.py", 51),
TestFile("test_block_int8.py", 22),
TestFile("test_create_kvindices.py", 2),
TestFile("test_chunked_prefill.py", 285),
TestFile("test_eagle_infer.py", 584),
TestFile("test_chunked_prefill.py", 313),
TestFile("test_eagle_infer.py", 619),
TestFile("test_ebnf_constrained.py", 108),
TestFile("test_enable_thinking.py", 70),
TestFile("test_embedding_openai_server.py", 141),
TestFile("test_eval_fp8_accuracy.py", 303),
TestFile("test_fa3.py", 376),
TestFile("test_fim_completion.py", 40),
TestFile("test_flashmla.py", 352),
TestFile("test_fp8_kernel.py", 8),
TestFile("test_function_call_parser.py", 10),
TestFile("test_fused_moe.py", 30),
TestFile("test_hicache.py", 116),
TestFile("test_hicache_mla.py", 254),
TestFile("test_hicache_mla.py", 127),
TestFile("test_hidden_states.py", 55),
TestFile("test_int8_kernel.py", 8),
TestFile("test_input_embeddings.py", 38),
TestFile("test_json_constrained.py", 98),
TestFile("test_large_max_new_tokens.py", 41),
TestFile("test_metrics.py", 32),
TestFile("test_mla.py", 242),
TestFile("test_mla_deepseek_v3.py", 221),
TestFile("test_mla_int8_deepseek_v3.py", 389),
TestFile("test_mla_flashinfer.py", 395),
TestFile("test_mla_fp8.py", 153),
TestFile("test_flashmla.py", 300),
TestFile("test_mla.py", 167),
TestFile("test_mla_deepseek_v3.py", 342),
TestFile("test_mla_int8_deepseek_v3.py", 429),
TestFile("test_mla_flashinfer.py", 302),
TestFile("test_mla_fp8.py", 93),
TestFile("test_no_chunked_prefill.py", 108),
TestFile("test_no_overlap_scheduler.py", 216),
TestFile("test_no_overlap_scheduler.py", 234),
TestFile("test_openai_function_calling.py", 60),
TestFile("test_openai_server.py", 149),
TestFile("test_penalty.py", 41),
TestFile("test_page_size.py", 60),
TestFile("test_pytorch_sampling_backend.py", 66),
TestFile("test_radix_attention.py", 167),
TestFile("test_radix_attention.py", 105),
TestFile("test_reasoning_content.py", 89),
TestFile("test_enable_thinking.py", 70),
TestFile("test_regex_constrained.py", 64),
TestFile("test_release_memory_occupation.py", 44),
TestFile("test_request_length_validation.py", 31),
@@ -70,13 +70,13 @@ suites = {
TestFile("test_skip_tokenizer_init.py", 117),
TestFile("test_srt_engine.py", 261),
TestFile("test_srt_endpoint.py", 130),
TestFile("test_tool_choice.py", 120),
TestFile("test_tool_choice.py", 226),
TestFile("test_torch_compile.py", 76),
TestFile("test_torch_compile_moe.py", 172),
TestFile("test_torch_native_attention_backend.py", 123),
TestFile("test_torchao.py", 70),
TestFile("test_triton_attention_kernels.py", 4),
TestFile("test_triton_attention_backend.py", 134),
TestFile("test_triton_attention_backend.py", 150),
TestFile("test_triton_moe_channel_fp8_kernel.py", 25),
TestFile("test_triton_sliding_window.py", 250),
TestFile("test_update_weights_from_disk.py", 114),
@@ -84,10 +84,9 @@ suites = {
TestFile("test_vertex_endpoint.py", 31),
TestFile("test_vision_chunked_prefill.py", 175),
TestFile("test_vlm_input_format.py", 300),
TestFile("test_vision_openai_server_a.py", 700),
TestFile("test_vision_openai_server_b.py", 700),
TestFile("test_vision_openai_server_a.py", 584),
TestFile("test_vision_openai_server_b.py", 556),
TestFile("test_w8a8_quantization.py", 46),
TestFile("models/lora/test_lora_cuda_graph.py", 250),
],
"per-commit-amd": [
TestFile("test_mla.py", 242),
@@ -119,9 +118,9 @@ suites = {
# TestFile("test_deepep_intranode.py", 50),
# TestFile("test_deepep_low_latency.py", 50),
# TestFile("test_moe_deepep_eval_accuracy_large.py", 250),
TestFile("test_disaggregation.py", 210),
TestFile("test_disaggregation_different_tp.py", 210),
TestFile("test_full_deepseek_v3.py", 250),
TestFile("test_disaggregation.py", 270),
TestFile("test_disaggregation_different_tp.py", 155),
TestFile("test_full_deepseek_v3.py", 463),
],
"per-commit-8-gpu-amd": [
TestFile("test_full_deepseek_v3.py", 250),
@@ -133,11 +132,11 @@ suites = {
TestFile("test_nightly_gsm8k_eval_amd.py"),
],
"vllm_dependency_test": [
TestFile("test_vllm_dependency.py"),
TestFile("test_awq.py"),
TestFile("test_bnb.py"),
TestFile("test_gguf.py", 78),
TestFile("test_gptqmodel_dynamic.py", 72),
TestFile("test_bnb.py"),
TestFile("test_vllm_dependency.py"),
],
}