feat: x86 matmul and transpose support (SDPA llama runnable)#641

Merged

oreomaker merged 3 commits intoUbiquitousLearning:mainfrom

Feb 20, 2026

Commits on Feb 19, 2026

feat(cpu): add GGUF Q4_0 support for CPU embedding operation
oreomaker
committed
feat(cpu): add x86 transpose kernel implementation with SIMD optimization
oreomaker
committed
feat(cpu): add x86 BLAS SGEMM implementation with AVX optimization
oreomaker
committed