1. 14 Dec, 2020 1 commit
    • Paul Fultz II's avatar
      Use dnnl for cpu backend (#688) · 406afeb8
      Paul Fultz II authored
      
      
      * Add flag to enable cpu backend
      
      * Make buffers shared
      
      * Enable optimizations
      
      * Add onednn
      
      * Formatting
      
      * Formatting
      
      * Add dnnl header
      
      * Formatting
      
      * Rewrite rnn first
      
      * Formatting
      
      * Call reference implementation
      
      * Formatting
      
      * Make literal data shared
      
      * Formatting
      
      * Add convolution
      
      * Formatting
      
      * Compensate for dilation
      
      * Formatting
      
      * Use name/make_op instead
      
      * Formatting
      
      * Rename gemm header
      
      * Formatting
      
      * Add dnnl convolution/gemm operators
      
      * Formatting
      
      * Add eliminate_contiguous
      
      * Add faster pointwise operators
      
      * Formatting
      
      * Formatting
      
      * Formatting
      
      * Add dnnl op class
      
      * Formatting
      
      * Add add op
      
      * Formatting
      
      * Add concat operator
      
      * Formatting
      
      * Add more ops
      
      * Create descriptor during finalization
      
      * Formatting
      
      * Dont rewrite pooling
      
      * Enable memory coloring
      
      * Formatting
      
      * Add output aliases
      
      * Formatting
      
      * Fix errors
      
      * Formatting
      
      * Convert literals
      
      * Add missing file
      
      * Remove batch_norm
      
      * Formatting
      
      * Use strides
      
      * Formatting
      
      * Add some debug checks
      
      * Formatting
      
      * Fix big in adjusting shape for gemm
      
      * Formatting
      
      * Fix fallback dot operator
      
      * Zero initialize buffers
      
      * Add suport for group convolutions
      
      * Formatting
      
      * Make adjust allocation target independent
      
      * Formatting
      
      * Enable adjust_allocation for gpu/cpu
      
      * Formatting
      
      * Add copy to allocation model
      
      * Formatting
      
      * Add copy operator
      
      * Formatting
      
      * Better handling of output parameters in adjust_allocation
      
      * Formatting
      
      * Build with dnnl
      
      * Make dnnl required
      
      * Fix compile error
      
      * Tidy fixes
      
      * Formatting
      
      * Tidy fixes
      
      * Formatting
      
      * Fix more tidy issues
      
      * Formatting
      
      * Add mul op
      
      * Add mul op
      
      * Set c compiler to clang as well
      
      * Compensate for normalized compute shape
      
      * Formatting
      
      * Fix cppcheck errors
      
      * Formatting
      
      * Add onednn library to hcc
      
      * Guard clang pragmas
      
      * Disable cpu mode for gcc for now
      
      * Leave it enabled it for gcc 7
      
      * Fix cppcheck suppresion
      
      * Fix compile error on gcc 5
      
      * Remove unused code
      Co-authored-by: default avatarShucai Xiao <shucai.xiao@amd.com>
      Co-authored-by: default avatarmvermeulen <5479696+mvermeulen@users.noreply.github.com>
      406afeb8
  2. 08 Dec, 2020 1 commit
    • Paul Fultz II's avatar
      Refactor to use make_op almost everywhere (#696) · 8d21fdc9
      Paul Fultz II authored
      * Load op when serializing
      
      * Formatting
      
      * Add missing clip field
      
      * Use make_op almost everywhere
      
      * Formatting
      
      * More make ops for rnns
      
      * Get rid of spaces
      
      * Formatting
      
      * Remove operators headers
      
      * Formatting
      
      * Remove unused op headers
      
      * Increase line threshold
      8d21fdc9
  3. 02 Dec, 2020 1 commit
  4. 16 Nov, 2020 1 commit
    • Shucai Xiao's avatar
      Normalize ops (#667) · 8443ecd1
      Shucai Xiao authored
      
      
      * add a pass to normalize ops
      
      * clang format
      
      * add unit tests
      
      * clang format
      
      * code backup
      
      * clang format
      
      * code backup
      
      * clang format
      
      * add support for slice in the normalize_op function
      
      * clang format
      
      * add operation method api for whether we need to call normalize_op
      
      * clang format
      
      * fix review comments
      
      * clang format
      
      * rename a function namejJ
      
      * clang format
      
      * change compute_shape to normalize_compute_shape for corresponding operators
      
      * clang format
      
      * remove unnecessary code
      
      * fix various issues
      
      * clang format
      
      * add attributes to operators having axis attributes
      
      * clang format
      
      * fixed jenkins build error
      
      * clang format
      
      * fix a bug related to slice
      
      * clang format
      
      * code backup
      
      * clang format
      
      * code backup
      
      * clang format
      
      * rename a file
      
      * fix cppcheck error
      
      * some code refinement
      
      * clang format
      
      * change attributes to enum
      
      * clang format
      
      * refine the enum
      
      * clang format
      
      * remove unnecessary code
      
      * add unit tests for more code coverage and fixed a bug
      
      * clang format
      
      * remove unnecessary changes
      
      * change normalize_axes to normalize
      
      * clang format
      
      * revert back the changes in broadcast.hpp
      
      * rename normalize_axes to normalize
      
      * fix review comments
      
      * clang format
      
      * Add flag to enable cpu backend
      
      * Make buffers shared
      
      * Enable optimizations
      
      * Formatting
      
      * Try to avoid ambiguous assign in value class
      
      * fixed a build error
      
      * clang format
      
      * add the normalize_ops pass to the ref target
      
      * refactor program to module to normalize_ops pass
      Co-authored-by: default avatarPaul <pfultz2@yahoo.com>
      Co-authored-by: default avatarmvermeulen <5479696+mvermeulen@users.noreply.github.com>
      8443ecd1
  5. 16 Sep, 2020 1 commit
  6. 19 Aug, 2020 1 commit
    • Paul Fultz II's avatar
      Add to_value/from_value to operation class (#605) · d1caaaa1
      Paul Fultz II authored
      * Add initial serialization
      
      * Formatting
      
      * Add unit tests
      
      * Formatting
      
      * Add tests for serialization
      
      * Formatting
      
      * Use or not and
      
      * Add value test
      
      * Formatting
      
      * Add more tests
      
      * Add shape serialization
      
      * Formatting
      
      * Add serializtion for literal and argument
      
      * Formatting
      
      * Add from and to value to operatation
      
      * Formatting
      
      * Serialize empty types
      
      * Formatting
      
      * Tidy fixes
      
      * Formatting
      
      * Fix tidy issues
      
      * Formatting
      
      * Reformat value type macro
      
      * Formatting
      
      * Handle enum types
      
      * Formatting
      
      * Use const ref
      
      * Update
      
      * Add tests for to_value/from_value
      
      * Formatting
      
      * Add more tests
      
      * Rewrite test to avoid redundant assignment
      d1caaaa1
  7. 20 Nov, 2019 2 commits
  8. 04 May, 2019 1 commit
  9. 02 May, 2019 4 commits
  10. 05 Feb, 2019 1 commit
  11. 16 Jan, 2019 2 commits
  12. 03 Jan, 2019 2 commits
  13. 11 Dec, 2018 1 commit
  14. 14 Nov, 2018 1 commit
  15. 02 Nov, 2018 2 commits
  16. 19 Oct, 2018 1 commit
  17. 18 Oct, 2018 5 commits
  18. 16 Sep, 2018 2 commits
  19. 15 Sep, 2018 3 commits
  20. 18 Aug, 2018 2 commits
  21. 31 Jul, 2018 1 commit
  22. 20 Jul, 2018 2 commits
  23. 18 Jul, 2018 2 commits