[Performance] Add Triton kernel for Gemma3n sparse GELU (#48498)

Signed-off-by: BWAAEEEK <jooho414@gmail.com>
Co-authored-by: Andreas Karatzas <akaratza@amd.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
This commit is contained in:
JooHo Lee
2026-09-14 07:42:15 +00:00
committed by GitHub
co-authored by Andreas Karatzas Cyrus Leung
parent e6b4e47d2d
commit 78e84261ab
15 changed files with 555 additions and 21 deletions
+3
View File
@@ -34,6 +34,9 @@ class IrOpPriorityConfig:
fused_add_rms_norm: list[str] = Field(default_factory=list)
"""Priority list for vllm.ir.ops.fused_add_rms_norm"""
gelu_and_mul_sparse: list[str] = Field(default_factory=list)
"""Priority list for vllm.ir.ops.gelu_and_mul_sparse"""
def compute_hash(self) -> str:
"""
Produces a hash unique to the pass configuration.