1. 31 Oct, 2025 1 commit
    • Weile's avatar
      Add NUM_SRCS/NUM_DSTS template parameters to GpuReduceKernel (#209) · 44140eeb
      Weile authored
        - Instantiates optimized kernels for common Transfer types:
          - Copy (1 src → 1 dst): Optimized single-source data copy
          - Read-only (1 src → 0 dst): Optimized memory read validation
          - Write-only (0 src → 1 dst): Optimized memory write/initialization
        - Compiler eliminates dead code loops for these specialized cases, improving performance by up to 7% for all-to-all workloads on MI3xx machines
        - Update CHANGELOG
      44140eeb
  2. 24 Oct, 2025 2 commits
  3. 22 Oct, 2025 1 commit
    • Weile's avatar
      Warp transfer (#203) · fbb3a80e
      Weile authored
      * add env variables
      
      * update CHANGELOG.md
      
      * add warp-level dispatch 
      fbb3a80e
  4. 04 Sep, 2025 1 commit
    • gilbertlee-amd's avatar
      v1.64.00 (#198) · a824bc1b
      gilbertlee-amd authored
      * Added BLOCKSIZES to a2asweep preset to allow sweeping over threadblock sizes
      * Fixing src initialization when using BYTE_OFFSET
      * Adding FILL_COMPRESS functionality to allow for different input data patterns
      * Updating CHANGELOG regarding GFX_BLOCKSIZE limit increase to 1024
      a824bc1b
  5. 08 Aug, 2025 1 commit
    • gilbertlee-amd's avatar
      TransferBench v1.63 (#193) · 023ce416
      gilbertlee-amd authored
      
      
      * Fixing issue with P memory type and use of DMA subexecutor
      * CMake builds require explicit opt-in by setting NIC_EXEC_ENABLE=1
      * Removing self-GPU check for DMA engine copies
      * [BUILD] Add new GPU targets and switch to amdclang++ (#187)
      * [BUILD] Add gfx950, gfx1150, and gfx1151 targets
      * [BUILD] Modify CMake to use amdclang++
      * [BUILD] Modify Makefile to use amdclang++
      * [GIT] Updated CHANGELOG and .gitignore
      * Adding HBM testing to healthcheck preset
      * Tweaking HBM tests to occur first, and provide more info during VERBOSE=1
      * Fixing timing reporting issues with NUM_SUBITERATIONS
      * [BUILD] Simplify Makefile (#190)
      * Combines steps for compilation and linking
      * Does not rebuild if no change to source code
      
      * Updating CHANGELOG
      
      ---------
      Co-authored-by: default avatarNilesh M Negi <Nilesh.Negi@amd.com>
      023ce416
  6. 09 Jun, 2025 1 commit
    • gilbertlee-amd's avatar
      TransferBench v1.62.00 (#181) · 569287a7
      gilbertlee-amd authored
      * Adding non-temporal loads and stores via GFX_TEMPORAL
      * Adding additional summary details to a2a preset
      * Add SHOW_MIN_ONLY for a2asweep preset
      * Adding new P CPU memory type which is indexed by closest GPU
      569287a7
  7. 28 Feb, 2025 1 commit
  8. 30 Jan, 2025 1 commit
  9. 21 Jan, 2025 1 commit
  10. 13 Dec, 2024 3 commits
  11. 12 Dec, 2024 1 commit
  12. 05 Dec, 2024 1 commit
  13. 28 Nov, 2024 1 commit
  14. 26 Nov, 2024 1 commit
  15. 22 Nov, 2024 1 commit
  16. 21 Nov, 2024 2 commits