ViperEkura/AstrAI
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

refactor: reorganize CUDA kernels into per-family directories

Browse Source 
- move attention kernels to csrc/kernels/attention/ and rotary to rotary/
- add shared common/mma.cuh (mma_sync, ldmatrix) and device.cuh (sm checks)
- split fp8_mm into three-layer fp8/common.h, gemm.cuh, mm.cu
- fix fused FP8 GEMM ldmatrix lane indexing to fix OOB shared reads
- update extension ops, loader, and kernel tests
This commit is contained in:
ViperEkura
2026-08-22 20:40:31 +08:00
parent cb21af38ba
commit 16a55bb474
30 changed files with 1956 additions and 1235 deletions
Download Patch File Download Diff File Expand all files Collapse all files
csrc/__init__.py
+1 -1
View File
@@ -1,2 +1,2 @@
# Source directory for CUDA kernels — build-time only.
# Compiled .so files live in astrAI/_ext/.
# Compiled .so files live in astrai/extension/lib/ (see csrc/CMakeLists.txt).