Added support for ARM64/OpenCL (RC5-72) and ARM64 (OGR-NG & ANSI RC5-72) on macOS 11 (Apple Silicon) #16
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Added .DS_Store to .gitignore
It doesn't seem arm64 & OpenCL was originally planned. Added an exception for ARM64 & OpenCL.
Update to configure to include macosx-arm64-opencl. Although arm64 support is macOS 11 is technically in macOS 11 not Mac OS X, it has more in common. Consider adding a new OS to the list and updating.
Added OPENCL and ARM64 as a vaild combination.
Added a (bad) assumption that if you're using an Apple arm64 processor with OpenCL it's not iOS.
Turn OGR back on in configure
Determined arm64 in add_sources is not required with OpenCL.
Tidied up macOS defintions to a single line.
Added ARM64 & ARM to CPU list rather than directly under APPLE
Fixed spacing
OSX: leading underscore required on assembler functions that are going to be global and visible to C/C++ code.
Added maxosx-arm64 support for OGR-NG cores, and ANSI RC5-72 cores. ScalarFusion not currently functional.
ScalarFusion not currently passing tests for MacOS ARM64.
ShadowLive
changed the title
|
The current ARM64 core shipping in the private branch is MONIKA, which has replaced the (slow) ScalarFusion. MONIKA's performance is shaky on some cores (it works well on Kryo, poorly on licensable cores) but if you'd like to get in touch about making it work, feel free to hop on the IRC channel. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Added support for Apple Silicon ARM64/OPENCL (RC5-72).
ARM64 Native (this fork)
[Nov 28 10:37:15 UTC] RC5-72 benchmark summary :
Default core : #-1 (undefined) 0 keys/sec
Fastest core : #1 (CL 1-pipe) 952,068,808 keys/sec
Existing AMD64 under Rosetta 2
[Nov 28 10:39:10 UTC] RC5-72 benchmark summary :
Default core : #-1 (undefined) 0 keys/sec
Fastest core : #1 (CL 1-pipe) 881,571,409 keys/sec
Added support for Apple Silicon ARM64 (OGR-NG).
ARM64 Native (this fork)
[Nov 30 13:12:32 UTC] OGR-NG: Benchmark for core #0 (FLEGE 2.0)
0.00:00:16.98 [109,275,989 nodes/sec]
Existing AMD64 under Rosetta 2
[Nov 30 13:13:43 UTC] OGR-NG benchmark summary :
Default core : #2 (cj-asm-sse2) 60,832,450 nodes/sec
Fastest core : #1 (cj-asm-generic) 88,987,193 nodes/sec
Added support for Apple Silicon ARM64 (RC5-72 ANSI).
Slower than AMD64 under Rosetta 2 as ScalarFusion Core not functional on Apple ARM64 at this time.
ARM64 Native (this fork)
[Nov 30 13:15:16 UTC] RC5-72 benchmark summary :
Default core : #-1 (undefined) 0 keys/sec
Fastest core : #0 (ANSI 4-pipe) 7,213,400 keys/sec
Existing AMD64 under Rosetta 2
[Nov 30 13:18:03 UTC] RC5-72 benchmark summary :
Default core : #3 (GO 2-pipe d) 11,192,248 keys/sec
Fastest core : #1 (KBE-64 3-pipe) 11,864,297 keys/sec
Notes:
Confirmed it compiles using Clang 12.
Apple clang version 12.0.0 (clang-1200.0.32.27)
Target: arm64-apple-darwin20.1.0
Confirmed all tests passed using './dnetc -test'
Updated temperature.cpp to allow ARM64 but have not confirmed temperature can be read on Apple Silicon.
Still to do: