1. 14 Feb, 2023 1 commit
  2. 10 Feb, 2023 1 commit
  3. 06 Feb, 2023 1 commit
  4. 31 Jan, 2023 2 commits
  5. 19 Jan, 2023 1 commit
  6. 17 Jan, 2023 2 commits
  7. 11 Jan, 2023 1 commit
  8. 09 Jan, 2023 1 commit
  9. 11 Dec, 2022 1 commit
    • Umang Yadav's avatar
      change target flag (#1488) · b41c1f01
      Umang Yadav authored
      HIP had change in previous rocm releases to use --offload-arch instead of --cuda-gpu-arch.
      
      This should be backwards compatbile. hipRTC also supports --offload-arch.
      b41c1f01
  10. 08 Dec, 2022 2 commits
    • Charlie Lin's avatar
      Dynamic ref dot operator (#1457) · d411aa69
      Charlie Lin authored
      Extends dot MIGX operator to handle dynamic input shapes
      Only allow dot between two dynamic shapes that have exactly matching outer dimensions
      Inner dimensions must also match correspondingly
      Updates dot related tests
      Change check_shapes to use shape.ndim()
      ONNX parsers for GEMM and MatMult will be updated in a separate PR
      d411aa69
    • Charlie Lin's avatar
      Dynamic reference Softmax (#1475) · 8e7d2efe
      Charlie Lin authored
      No major changes required, use dyn_output and pass dynamic shape when calling compute_shape()
      Adds dynamic shape tests
      8e7d2efe
  11. 07 Dec, 2022 1 commit
  12. 06 Dec, 2022 2 commits
  13. 29 Nov, 2022 1 commit
  14. 20 Nov, 2022 1 commit
  15. 18 Nov, 2022 1 commit
    • Umang Yadav's avatar
      Disable Find2.0 for now (#1462) · 493bb8d5
      Umang Yadav authored
      Disabling it untill int8 fix is in mainline from MIOpen and also so that QA tests could run migraphx-driver and unittests from MIGraphX.
      493bb8d5
  16. 07 Nov, 2022 1 commit
  17. 06 Nov, 2022 1 commit
  18. 02 Nov, 2022 2 commits
  19. 28 Oct, 2022 1 commit
  20. 27 Oct, 2022 2 commits
  21. 26 Oct, 2022 1 commit
  22. 24 Oct, 2022 1 commit
  23. 19 Oct, 2022 2 commits
  24. 18 Oct, 2022 1 commit
  25. 13 Oct, 2022 2 commits
  26. 04 Oct, 2022 2 commits
  27. 03 Oct, 2022 1 commit
    • Umang Yadav's avatar
      Add output_alias and runs_on_offload_target flags for the custom ops (#1309) · c9ffb38d
      Umang Yadav authored
      Adds two methods for the custom_ops virtual class.
      
      bool runs_on_offload_target(), if the custom op runs directly on the gpu then it should be set to true. in this case, custom op expects its parameters to reside in GPU memory and writes output to the GPU memory. If it is set to false then, custom op expects it's parameter to reside on the host and puts back the result into the host memory.
      
      output_alias, if output of the custom op is aliasing the input buffer. i.e. interpreting the same input buffer with differnet shape and strides.
      
      Update as_vector() in C++ API to handle non-standard shapes. It required exposing element_index to space_index conversion method for the shape class.
      c9ffb38d
  28. 29 Sep, 2022 1 commit
  29. 28 Sep, 2022 1 commit
    • Umang Yadav's avatar
      Add compute_fp32 flag for quant_gemm tests (#1360) · 70e63960
      Umang Yadav authored
      test_gpu_pack_int8_args fails on gfx908 machine, because it doesn't set compute_fp32 flag correctly. This PR fixes the test such that it checks for the device-name, and rocblas-versions and sets this flag accordingly.
      70e63960
  30. 27 Sep, 2022 1 commit
  31. 26 Sep, 2022 1 commit