-
Notifications
You must be signed in to change notification settings - Fork 228
Floating-Point Comparison Performance #775
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
Comments
Yeah, I think getting it to work quickly with multiprecision is the tricky part. |
We would need the existing implementation for all non-builtin non-IEEE floats IMO. But it's not just those, I suspect it's much harder than that blog post suggests once you handle all the corner cases correctly. I do remember that that code was a bitch to get correct. We should have fairly good tests though, so if Matt wants to take a shot at the faster version by all means go for it :) |
Oh don't forget the |
It's worse, behaviour can vary at runtime: https://www.intel.com/content/www/us/en/develop/documentation/cpp-compiler-developer-guide-and-reference/top/compiler-reference/floating-point-operations/understanding-floating-point-operations/setting-the-ftz-and-daz-flags.html |
eww; wouldn't that require query of the cpuid instruction on every call? |
From this write up float distance could be several times better. I can poke around when implementing
constexpr std::nexttoward
as that is the last outlier in ccmath from this proposal.The text was updated successfully, but these errors were encountered: