llama.cpp
Add `--no-op-offload` to improve `-ot` pp perf in MoE models like llama4 400B
#13386
Merged
Overview
Commits
4
Changes
View On GitHub
Commits
  • Add --disable-op-offload
    hjc4869 committed 180 days ago
  • Avoid negative bools in library.
    hjc4869 committed 180 days ago
  • Fix default value of ggml_backend_sched_new
    hjc4869 committed 180 days ago
  • Rename to --no-op-offload for consistency
    hjc4869 committed 178 days ago
FAQ Terms Privacy Refunds Impressum
Loading