Linux kernel creator Linus Torvalds hopes to see the end of Intel's 512-bit vector extensions and admits he has an "irrational hatred" of the floating-point (FP) benchmarks used to prove their value.
Torvalds fired off his criticism of Intel's Advanced Vector Extensions 512 (AVX-512) instructions in a mailing list chat. He was responding to Phoronix's article on GNU Compiler Collection 11 lacking support for AVX-512 in the compiler instructions Intel has enabled for Alder Lake, its 2021 processors for the desktop. Intel's future Xeon Sapphire Roads processors still do support AVX-512.
"I hope AVX-512 dies a painful death, and that Intel starts fixing real problems instead of trying to create magic instructions to then create benchmarks that they can look good on," wrote Torvalds.
SEE: An IT pro's guide to robotic process automation (free PDF) (TechRepublic)
Torvalds takes exception to Intel's focus on FP benchmarks and its processors' performance on supercomputers, or high-performance computers (HPCs)
"I hope Intel gets back to basics: gets their process working again, and concentrate more on regular code that isn't HPC or some other pointless special case."
He notes that "in the heyday of x86", Intel's rivals always outperformed it on FP loads.
"Intel's FP performance sucked (relatively speaking), and it mattered not one iota. Because absolutely nobody cares outside of benchmarks," Torvalds said.
"The same is largely true of AVX-512 now – and in the future. Yes, you can find things that care. No, those things don't sell machines in the big picture."
He continued his criticism by saying AVX512 has real downsides.
"I'd much rather see that transistor budget used on other things that are much more relevant. Even if it's still FP math (in the GPU, rather than AVX-512). Or just give me more cores (with good single-thread performance, but without the garbage like AVX-512) like AMD did."
Web performance firm Cloudflare has written about the performance impact of AVX-512. It advised customers who don't need AVX-512 for high-performance tasks to disable AVX-512 execution on the server and desktop to avoid its "accidental" throttling.
"I want my power limits to be reached with regular integer code, not with some AVX-512 power virus that takes away top frequency (because people ended up using it for memcpy!) and takes away cores (because those useless garbage units take up space)," continued Torvalds.
"Yes, yes, I'm biased. I absolutely detest FP benchmarks, and I realize other people care deeply. I just think AVX-512 is exactly the wrong thing to do. It's a pet peeve of mine. It's a prime example of something Intel has done wrong, partly by just increasing the fragmentation of the market.
"Stop with the special-case garbage, and make all the core common stuff that everybody cares about run as well as you humanly can."
In a separate email, Torvalds emphasized that he had an "irrational hatred of vector units and FP benchmarks".
"I think they are largely a complete waste of transistors and effort, and I think the amount of time spent on them – both by hardware people and by software people trying to use them – has been largely time wasted," he wrote.
"So I'm exaggerating and overstating things to the point of half-kidding. But only half. I'm taking a fairly extreme standpoint, and I know my hatred isn't really rational, but just a personal quirk and just pure unadulterated opinionated ranting."