-
Notifications
You must be signed in to change notification settings - Fork 1.4k
Fix inf loop in test on arm #450
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
Fix inf loop in test on arm #450
Conversation
There was an infinite loop on the arm platform when running TMVAMulticlass.root. When compiling with the -ffast-math flag, sometimes a nan would be generated.
|
Can one of the admins verify this patch? |
|
@phsft-bot build! |
|
Awesome, thanks a lot for taking your time looking into this! |
|
Hi @ashlaban : what quantity ends up being a nan? Maybe we can avoid this with a few checks. |
|
@dpiparo The GA optimiser uses I also verified that |
|
Hi @ashlaban, |
|
Hi @dpiparo, Ah! That is good to know! I definitely think this solves one of the problems on the arm platform, could we be talking about different issues? Regarding the preprocessor protect:
So since I don't think it will matter either way, feel free to do as you feel best. One point is that the code flow will be more clear, in my mind, without any preprocessor branching. |
|
Hi, if the numerical stability of the algorithm is guaranteed, I am in favour of not cluttering the code with preprocessor statements 👍 |
|
I guarantee it. |
|
@phsft-bot build! |
There was an infinite loop on the arm platform when running TMVAMulticlass.root. When compiling with the -ffast-math flag, sometimes a nan would be generated in the GA part of the cut optimisation.
@dpiparo @martinmine Hopefully this resolves your issue. Tried it on the build machine I was given access to and it works there now. We in the TMVA team still need to revisit this part at some point, but for now I think this should be ok.