mastodon 4.7.3Founder of Molecular Matters • C++ & low-level programming • Created Live++ (@liveplusplus@mastodon.gamedev.place)
https://liveplusplus.tech
Sometimes you have to profile debug builds, and as always @superluminal@mastodon.gamedev.place
is perfect for that.
My matrix multiplication was using variables on the stack, which caused a lot of stack checks.
Rewrote the code, debug performance went from 23ms to 6ms.
Two things that would be nice:
NVIDIA providing public symbols for their drivers.
NVIDIA not having their drivers bring up more than 100 threads on my machine.
RE: @liveplusplus@mastodon.gamedev.place
Of course I missed an edge case in the previous 2.11.3 update. It's really hard not to, there are so many variables to the equation.
If you don't have an anxiety disorder yet, just watching The Bear might give you one.
(Started Season 5...)
https://www.cladlabs.ai/blog/introducing-clad-labs
Very clearly shows the problem with AI being slapped onto everything these days: you don't know if it's a joke or will get acquired for a few million.