llama.cpp · b12031

opencl: 초기 iq3 지원 추가 (#31338) * opencl: iq3_s AOS mv 추가, Assisted-by: Codex * opencl: iq3_s SOA mv 및 타일링된 mm 추가, Assisted-by: Codex * opencl: adreno용 iq3_s noshuffle gemm/gemv 추가, Co-authored-by: Shaofei Qi, Assisted-by: Codex * opencl: adreno용 iq3_s dp4a 추가, Co-authored-by: Shaofei Qi, Assisted-by: Codex * opencl: adreno의 iq3_s 디코딩과 배치 축소 최적화, Co-authored-by: Shaofei Qi, Assisted-by: Codex * opencl: iq3_xxs AOS mv 추가, Assisted-by: Codex * opencl: iq3_xxs SOA mv 및 타일링된 mm 추가, Assisted-by: Codex * opencl: adreno용 iq3_xxs noshuffle gemm/gemv 추가, Co-authored-by: Shaofei Qi, Assisted-by: Codex * opencl: adreno용 iq3_xxs dp4a 추가, Co-authored-by: Shaofei Qi, Assisted-by: Codex * opencl: adreno의 iq3_xxs 디코딩과 배치 축소 최적화, Co-authored-by: Sha