feat(ops): add Matmul implementations for CPU NVIDIA MetaX Iluvatar a…#59

Draft

gongchensu wants to merge 15 commits intoInfiniTensor:feat/ascend-operatorsfrom

gongchensu:feat/matmul-multi-backend

Commits on Apr 15, 2026

fix(ascend): refine framework layer — caching, naming, build fixes
zhangyue
committed
feat(base): add new operator base classes and refine existing ones
zhangyue
committed
feat(cpu): add CPU implementations for Cast, Cat, Linear, Mul
zhangyue
committed
feat(ascend): add Ascend operator kernels for all operators
zhangyue
committed
feat(ascend): add multi-implementation variants and ATB operators
zhangyue
committed
feat(ascend): add custom AscendC kernels for RmsNorm and AddRmsNorm
zhangyue
committed
test(ascend): add comprehensive tests for all Ascend operators
zhangyue
committed
style: apply clang-format and fix code convention violations (round 1)
zhangyue
committed
style: fix code convention violations (round 2)
zhangyue
committed
style: fix code convention violations (round 3)
zhangyue
committed
fix(ci): add Ascend toolkit environment variables to CI Dockerfile
zhangyue
committed
style: apply ruff format and clang-format to all modified files
zhangyue
committed
style: fix ruff F401 lint errors for side-effect imports
zhangyue
committed
refactor(test): remove duplicate rms_norm test files and unify kernel loading
zhangyue
committed

Commits on Apr 17, 2026

feat(ops): add Matmul implementations for CPU NVIDIA MetaX Iluvatar and Moore
gongchensu
committed