llama.cpp
ggml: move fp16/bf16 conversion optimizations to CPU backend + export conversion APIs
#13107
Merged
Merged
ggml: move fp16/bf16 conversion optimizations to CPU backend + export conversion APIs #13107
slaren
merged
3 commits into
ggml-org:master
from
SongXiaoXi:master
ggml: dynamic x86_64 feature detection for FP32 <-> FP16/BF16 conversion
c5e3b52c
move fp converter to ggml-cpu
3efb0e73
SongXiaoXi
changed the title
slaren
approved these changes
on 2025-04-26
Switch ggml_compute_forward_get_rows_f16/bf16 to new ggml_cpu_fp16/bf…
82f8630a
Login to write a write a comment.
Login via GitHub