Историја ревизија

Аутор SHA1 Порука Датум
  Daniel Hiltgen 3c49c3ab0d Harden GPU mgmt library lookup пре 1 година
  Jeffrey Morgan c336693f07 calculate overhead based number of gpu devices (#1875) пре 1 година
  Daniel Hiltgen 1961a81f03 Set corret CUDA minimum compute capability version пре 1 година
  Jeffrey Morgan 6df83e6daa update rough cuda overhead estimate to 15% + 384MiB пре 1 година
  Jeffrey Morgan 6164f378f2 revert cuda overhead to 20% пре 1 година
  Jeffrey Morgan 6566387ae3 add `TODO` for cuda overhead пре 1 година
  Jeffrey Morgan 37708931fb update cuda overhead to 20% to fix crashes when switching between models and large context sizes пре 1 година
  Jeffrey Morgan f6cb0a553c update cuda overhead to 15% or 400MiB пре 1 година
  Jeffrey Morgan 2680078c13 fix build on linux пре 1 година
  Jeffrey Morgan f1b7e5f560 update overhead to 15% пре 1 година
  Jeffrey Morgan cb534e6ac2 use 10% vram overhead for cuda пре 1 година
  Jeffrey Morgan 08f1e18965 Offload layers to GPU based on new model size estimates (#1850) пре 1 година
  Daniel Hiltgen d74ce6bd4f Detect very old CUDA GPUs and fall back to CPU пре 1 година
  Daniel Hiltgen a2ad952440 Fix windows system memory lookup пре 1 година
  Daniel Hiltgen d966b730ac Switch windows build to fully dynamic пре 1 година
  Daniel Hiltgen 7555ea44f8 Revamp the dynamic library shim пре 1 година
  Daniel Hiltgen 1b991d0ba9 Refine build to support CPU only пре 1 година
  Daniel Hiltgen 35934b2e05 Adapted rocm support to cgo based llama.cpp пре 1 година