Closed gauteh closed 2 years ago
Hi @gauteh 👋
Thank you very much for your contribution.
Unfortunately, this change has merge conflicts. This is due to my rework of the repo to use earthly. Do fix the merge conflict it seems like you only have to disregard the changes to the comparison.py
since this script doesn't exist any more. It's all managed in the Earthfile
now.
Also on another note, these are the new results:
So, I don't know if Rust needs optimizations any more 😅
Nice :D That looks good. I've updated the PR. Changing the target-cpu and using a f32 is a significantly faster. Using fold
over for
doesn't really make a difference, but maybe its a bit more idiomatic. Would be interesting to see if this PR makes a difference on the new charts!
https://github.com/niklas-heer/speed-comparison/pull/24 proposes to make the C and C++ computations more precise, instead, by using C's "doubles" (IEEE754 binary64 on most machines, AKA f64
).
Thank you very much for your contribution @gauteh :+1: Also thank you very much for the discussion and input @workingjubilee :slightly_smiling_face:
Thanks 🙂!
man. 10. okt. 2022, 16:00 skrev Niklas Heer @.***>:
Thank you very much for your contribution @gauteh https://github.com/gauteh 👍 Also thank you very much for the discussion and input @workingjubilee https://github.com/workingjubilee 🙂
— Reply to this email directly, view it on GitHub https://github.com/niklas-heer/speed-comparison/pull/20#issuecomment-1273358842, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAN36YYHZACVH6LQRUXR6DWCQOOPANCNFSM6AAAAAAQ3ARPB4 . You are receiving this because you were mentioned.Message ID: @.***>
fixes #12.