Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Odd behavior with native FMA instructions when first launching Julia #35055

Closed
musm opened this issue Mar 9, 2020 · 3 comments
Closed

Odd behavior with native FMA instructions when first launching Julia #35055

musm opened this issue Mar 9, 2020 · 3 comments

Comments

@musm
Copy link
Contributor

musm commented Mar 9, 2020

I experienced weird response when it comes to FMA on Julia 1.4 .

First started Julia from the command line, without any flags (downloaded from julialang.org). Then I checked mulladd with code_native and find it's not using FMA CPU instructions and Base.Math.FMA_NATIVE also showed false... that was strange since as long as I've used julia this was always true and FMA always used native instruction.

Then I started Julia with -O3. Check thing again... muladd is correctly using native FMA, but Base.Math.FMA_NATIVE is false (makes sense since it's a precompile constant, but should be fixed regardless).

Then the strange behavior shows up. I then check julia again (without -O3 flag) and native FMA is working correctly in muladd (Base.Math.FMA_NATIVE still false).

@KristofferC
Copy link
Member

FMA_NATIVE is broken: #33011.

@musm
Copy link
Contributor Author

musm commented Mar 9, 2020

Thanks, updated issue to reflect the fact that when I first launched julia it was not using native FMA, but then ended up correctly using FMA after launching with -03.

@musm musm changed the title Odd behavior with FMA instructions and fixing FMA_NATIVE constant Odd behavior with native FMA instructions when first launching Julia Mar 9, 2020
@simonbyrne
Copy link
Contributor

Closing as duplicate of #33011.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants