Skip to content

Latest commit

 

History

History
40 lines (33 loc) · 1004 Bytes

README.MD

File metadata and controls

40 lines (33 loc) · 1004 Bytes

IRIS


Purpose of this project is to provide software implementation for vectorizing intrinsics available on ARM and x86 processors.

Additionally:

  • Ability to target and test software that uses ARM NEON intrinsics on x86 machines and vice versa
  • Ability to target and test ARM/x86 intrinsics from newer architectures when corresponding hardware is not yet available ( April 2017 - SVE(ARMv8-A) or AVX512(Intel Cannonlake))

Intrinsics to be supported:

  • ARM
    • NEON
    • SVE
  • x86
    • MMX
    • SSE
    • SSE2
    • SSE3
    • SSE4.1
    • SSE4.2
    • AVX
    • AVX2
    • FMA
    • AVX-512
      • AVX-512F
      • AVX-512BW
      • AVX-512CD
      • AVX-512DQ
      • AVX-512ER
      • AVX-512IFMA52
      • AVX-512PF
      • AVX-512VBMI
      • AVX-512VL
    • KNC

Current implementation status is available here: Status Summary

License: Apache 2.0