You can implement a LERP in two instructions with a FMA and an FNMSub (subtract instead of multiply on the third operand, and negate the multiplicand). And with AVX-512 you can do 16 of them simultaneously!
Does anyone else remember the glory days of dual GPU flagship cards?
We need some brave company to bring it back... Sell us a 5090 and a 3050 on a single board, then we have a true flagship that can actually play Batman with PhysX enabled
If you're relying on the laptop for bank transfers, overclocking it might be a very bad idea. You probably want a stable system more than one that's 5% faster, and overclocking a 16 year old laptop is going to put a lot of strain on its old capacitors and power delivery circuitry.
I've been on other forums with restrictions on new accounts, that e.g. put your threads in moderator review before putting them live. Does Anandtech software not support any of those features?
The spam threads are getting out of control, there must have been a dozen I've reported on today.
Could we do something about it? Maybe ban new accounts from making new threads until they have 50 posts, or something?
CUDA is shipped as a PTX intermediate representation, similar to how HLSL ships as an intermediate representation (soon to move to SPIR-V). They could absolutely support it on newer hardware if they chose to.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.