The Volkswagon emissions testing model
FP8 is ~100 tflops faster when the kernel name has "cutlass" in it
11–20 of 114 posts
Re: FP8 is ~100 tflops faster when the kernel name has "cutlass" in it
#12Let's hope for Nvidia this is an innocent optimization only valid for internal kernels that cannot be applied in general.
Re: FP8 is ~100 tflops faster when the kernel name has "cutlass" in it
#13is 100 tflops a lot?
It's like 5-10% here
Re: FP8 is ~100 tflops faster when the kernel name has "cutlass" in it
#14Intel's quest to move from "trusted by default / the reference" to "check for scam" is getting worse every release. And it's 100% self inflicted. How weird.
NVIDIA-inflicted in this case.
Re: FP8 is ~100 tflops faster when the kernel name has "cutlass" in it
#15Intel's quest to move from "trusted by default / the reference" to "check for scam" is getting worse every release. And it's 100% self inflicted. How weird.
In my understanding of the PR, it rather seems that it is NVidia is the company that is cheating. :-)
Re: FP8 is ~100 tflops faster when the kernel name has "cutlass" in it
#16is 100 tflops a lot?
5060 ti +~15%
Re: FP8 is ~100 tflops faster when the kernel name has "cutlass" in it
#17I have small experience with compilers and llvm but youd be shocked how many things rely on names and parsing names
If you have hundreds of passes that are complex and rely on various "contracts" like type names or some shit, then really crazy things like this can happen unintentionally and not maliciously
Re: FP8 is ~100 tflops faster when the kernel name has "cutlass" in it
#18I have small experience with compilers and llvm but youd be shocked how many things rely on names and parsing names If you have hundreds of passes that are complex and rely on various "contracts" like type names or some shit, then really crazy things like this can happen unintentionally and not maliciously
Web-developers are well aware of this too. Sincerely, Mozilla/5.0 (X11; Linux x86_64; rv:139.0) Gecko/20100101 Firefox/139.0
Re: FP8 is ~100 tflops faster when the kernel name has "cutlass" in it
#19And what’s the downside of using that kernel name? It can’t just be that it’s faster and nothing else. Unless they included lots of sleep(x) calls.
Re: FP8 is ~100 tflops faster when the kernel name has "cutlass" in it
#20In `libnvidia-nvvm.so` the string `cutlass` appears right after `Memory Dependence Analysis` and `memdep`. Perhaps it acts as an optimization attribute of some sort, where the compiler is allowed to make assumptions about the kernel's behavior that are not valid in general?
Thats very likely imo