Commit Verlauf

Autor SHA1 Nachricht Datum
  Daniel Hiltgen cff3f44f4a Fix case for NumCtx vor 10 Monaten
  Daniel Hiltgen 3518aaef33 Merge pull request #4218 from dhiltgen/auto_parallel vor 10 Monaten
  Blake Mizerany cb42e607c5 llm: speed up gguf decoding by a lot (#5246) vor 10 Monaten
  Daniel Hiltgen 9929751cc8 Disable concurrency for AMD + Windows vor 10 Monaten
  Daniel Hiltgen 17b7186cd7 Enable concurrency by default vor 1 Jahr
  Daniel Hiltgen 6f351bf586 review comments and coverage vor 11 Monaten
  Daniel Hiltgen ff4f0cbd1d Prevent multiple concurrent loads on the same gpus vor 11 Monaten
  Daniel Hiltgen fc37c192ae Refine CPU load behavior with system memory visibility vor 11 Monaten
  Daniel Hiltgen 434dfe30c5 Reintroduce nvidia nvml library for windows vor 11 Monaten
  Daniel Hiltgen 48702dd149 Harden unload for empty runners vor 11 Monaten
  Daniel Hiltgen 5e8ff556cb Support forced spreading for multi GPU vor 1 Jahr
  Michael Yang e40145a39d lint vor 11 Monaten
  Michael Yang c895a7d13f some gocritic vor 11 Monaten
  Michael Yang 04f3c12bb7 replace x/exp/slices with slices vor 11 Monaten
  Patrick Devine 4cc3be3035 Move envconfig and consolidate env vars (#4608) vor 11 Monaten
  Sang Park 4434d7f447 Correct typo in error message (#4535) vor 11 Monaten
  Daniel Hiltgen ec231a7923 Remove VRAM convergence check for windows vor 11 Monaten
  Patrick Devine 6845988807 Ollama `ps` command for showing currently loaded models (#4327) vor 11 Monaten
  Daniel Hiltgen 4142c3ef7c Always use the sorted list of GPUs vor 1 Jahr
  Jeffrey Morgan bb6fd02298 Don't clamp ctx size in `PredictServerFit` (#4317) vor 1 Jahr
  Daniel Hiltgen 354ad9254e Wait for GPU free memory reporting to converge vor 1 Jahr
  Jeffrey Morgan c9f98622b1 Skip scheduling cancelled requests, always reload unloaded runners (#4189) vor 1 Jahr
  Jeffrey Morgan dfa2f32ca0 unload in critical section (#4187) vor 1 Jahr
  Daniel Hiltgen f56aa20014 Centralize server config handling vor 1 Jahr
  Jeffrey Morgan 942c979232 allocate a large enough kv cache for all parallel requests (#4162) vor 1 Jahr
  Daniel Hiltgen 20f6c06569 Make maximum pending request configurable vor 1 Jahr
  Mark Ward 63c763685f log when the waiting for the process to stop to help debug when other tasks execute during this wait. vor 1 Jahr
  Mark Ward f4a73d57a4 fix runner expire during active use. Clearing the expire timer as it is used. Allowing the finish to assign an expire timer so that the runner will expire after no use. vor 1 Jahr
  Daniel Hiltgen d6e3b64582 Fix concurrency for CPU mode vor 1 Jahr
  Daniel Hiltgen 9b5a3c5991 Merge pull request #3914 from dhiltgen/mac_perf vor 1 Jahr