[DeepGemm] Add a flag for fast warmup (#18111)

This commit is contained in:
Baizhou Zhang
2026-02-04 14:12:13 +08:00
committed by GitHub
parent 4739f2e8d5
commit d279520ba5
3 changed files with 44 additions and 10 deletions
+3
View File
@@ -334,11 +334,14 @@ class Envs:
# DeepGemm
SGLANG_ENABLE_JIT_DEEPGEMM = EnvBool(True)
SGLANG_JIT_DEEPGEMM_PRECOMPILE = EnvBool(True)
SGLANG_JIT_DEEPGEMM_FAST_WARMUP = EnvBool(False)
SGLANG_JIT_DEEPGEMM_COMPILE_WORKERS = EnvInt(4)
SGLANG_IN_DEEPGEMM_PRECOMPILE_STAGE = EnvBool(False)
SGLANG_DG_CACHE_DIR = EnvStr(os.path.expanduser("~/.cache/deep_gemm"))
SGLANG_DG_USE_NVRTC = EnvBool(False)
SGLANG_USE_DEEPGEMM_BMM = EnvBool(False)
# DeepSeek MHA Optimization
SGLANG_CHUNKED_PREFIX_CACHE_THRESHOLD = EnvInt(8192)
# DeepEP