More quantize support #62

Job	Run time
ubuntu-focal-make	17s
ubuntu-latest-cmake	1m 22s
ubuntu-latest-cmake-mpi (mpich)	2m 6s
ubuntu-latest-cmake-mpi (libopenmpi-dev)	1m 30s
macOS-latest-cmake	2m 10s
macOS-latest-cmake-ios	55s
macOS-latest-cmake-tvos	1m 20s
macOS-latest-make	28s
macOS-latest-swift (generic/platform=macOS)	5s
ubuntu-latest-cmake-sanitizer (ADDRESS, Debug)	4m 3s
macOS-latest-swift (generic/platform=iOS)	49s
macOS-latest-swift (generic/platform=tvOS)	1s
ubuntu-latest-cmake-sanitizer (ADDRESS, Release)	3m 55s
ubuntu-latest-cmake-sanitizer (THREAD, Debug)	7m 51s
ubuntu-latest-cmake-sanitizer (THREAD, Release)	4m 36s
ubuntu-latest-cmake-sanitizer (UNDEFINED, Debug)	3m 0s
windows-latest-cmake (noavx, -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX=OFF -DLLAMA_A...	1m 14s
ubuntu-latest-cmake-sanitizer (UNDEFINED, Release)	3m 39s
windows-latest-cmake (avx2, -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON)	2m 9s
windows-latest-cmake (avx, -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX2=OFF -DBUILD_SH...	2m 14s
windows-latest-cmake-cublas (12.2.0, cublas)	15m 11s
windows-latest-cmake (avx512, -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX512=ON -DBUIL...	1m 39s
windows-latest-cmake-cublas (11.7.1, cublas)	16m 1s
windows-latest-cmake (clblast, -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_CLBLAST=ON -DBU...	1m 17s
windows-latest-cmake (openblas, -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_BLAS=ON -DBUIL...	51s
release	0s
	1h 18m 43s

Provide feedback