コミット履歴

作者 SHA1 メッセージ 日付
  Samuel Thibault 37862069c7 Permit the application to provide its own size base for performance models 13 年 前
  Samuel Thibault 86a13c15d2 deprecate cost_model, and introduce cost_function, which is provided with the whole task structure, the target arch and implementation number 13 年 前
  Samuel Thibault dfec7f881c Make sure to take control of the right GPU 13 年 前
  Samuel Thibault 1bd28b5f76 Comment on On-GPU rendering 13 年 前
  Samuel Thibault 08f4ce1f07 Fix in-place partitioning for OpenCL. OpenCL codelets are not supposed to use the ptr pointer, since in OpenCL we only have a cl_mem 13 年 前
  Cyril Roelandt 55c2efdadc Moving the multiformat example form "advanced api" to "advanced examples" 13 年 前
  Nathalie Furmento be21794fe1 doc: STARPU_GPU and gpu_funcs do not exist. Update to CUDA 13 年 前
  Samuel Thibault 63fbac4cbf Update doc concerning multiimpl 13 年 前
  Nathalie Furmento 80dc0ef0d0 doc: add missing s (Thanks Ludovic Stordeur for reporting it) 13 年 前
  Samuel Thibault c1d47302fe Tip about various compute capability variants 13 年 前
  Samuel Thibault 2f4491e804 Add can_execute member to the codelet structure, so applications can fine-tune implementations usage. Make most schedulers enforce it. 13 年 前
  Nathalie Furmento b3a6c1f52b do not use typedef for starpu_codelet, use it directly as a struct 13 年 前
  Nathalie Furmento 0ebf7ba326 Rename public opaque type starpu_data_handle as starpu_data_handle_t 13 年 前
  Samuel Thibault bf9bc759c4 Document that non-linear regression reverts to history-based on first run 13 年 前
  Samuel Thibault 0a06263c45 Only validate a regression model if we have at least 10% difference in input size. Do not validate the non-linear regression model just due to number of samples. 13 年 前
  Samuel Thibault 2d8c3bd434 mention which schedulers support choosing the implementation 13 年 前
  Samuel Thibault f1cb2d818f replace tabs with spaces 13 年 前
  Nathalie Furmento d451fc559a doc: move chapters to sub-directory chapter 13 年 前