minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

Enable PDL for various kernels in DSV32/GLM5 (#23965)

Browse Source 
Co-authored-by: b8zhong <b8zhong@users.noreply.github.com>
This commit is contained in:
Brayden Zhong
2026-05-09 03:42:56 -07:00
committed by GitHub
co-authored by b8zhong
parent d5564c2a96
commit 05d1ab51e8
8 changed files with 71 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files
sgl-kernel/csrc/elementwise/pos_enc.cuh
+4 -4
View File
@@ -105,7 +105,7 @@ __global__ void BatchQKApplyRotaryPosIdsCosSinCacheEnhancedHeadParallelismKernel
const uint32_t bdy = blockDim.y;
#if (defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 900))
asm volatile("griddepcontrol.wait;");
cudaGridDependencySynchronize();
#endif
vec_t<float, vec_size> cos, sin;
@@ -184,7 +184,7 @@ __global__ void BatchQKApplyRotaryPosIdsCosSinCacheEnhancedHeadParallelismKernel
}
#if (defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 900))
asm volatile("griddepcontrol.launch_dependents;");
cudaTriggerProgrammaticLaunchCompletion();
#endif
}
@@ -229,7 +229,7 @@ __global__ void BatchQKApplyRotaryPosIdsCosSinCacheEnhancedKernel(
const uint32_t bdy = blockDim.y;
#if (defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 900))
asm volatile("griddepcontrol.wait;");
cudaGridDependencySynchronize();
#endif
vec_t<float, vec_size> cos, sin;
@@ -310,7 +310,7 @@ __global__ void BatchQKApplyRotaryPosIdsCosSinCacheEnhancedKernel(
}
#if (defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 900))
asm volatile("griddepcontrol.launch_dependents;");
cudaTriggerProgrammaticLaunchCompletion();
#endif
}