1. 20 Jul, 2018 2 commits
    • Daniel Krebs's avatar
      gpu: always allocate page-sized chunks, then use LinearAllocator · 26d7a13e
      Daniel Krebs authored
      This was neccessary in order to make the memory available via GDRcopy
      when multiple small allocations were made. cudaMalloc() would return
      multiple memory chunks located in the same GPU page, which GDRcopy
      pretty much dislikes (`gdrdrv:offset != 0 is not supported`).
      As a side effect, this will keep the number of BAR-mappings done
      via GDRcopy low, because they seem to be quite limited.
      26d7a13e
    • Daniel Krebs's avatar
      common/memory: let allocators own a memory block · 2ef40409
      Daniel Krebs authored
      This is useful when we sub-delegate management of a memory block
      to another allocator.
      2ef40409
  2. 11 Jul, 2018 2 commits
  3. 06 Jun, 2018 1 commit
  4. 05 Jun, 2018 2 commits
  5. 04 Jun, 2018 31 commits
  6. 16 May, 2018 2 commits