Language Selection

English French German Italian Portuguese Spanish

Linus Torvalds: "I Hope AVX512 Dies A Painful Death"

Filed under
Development
Linux

Linux creator Linus Torvalds had some choice words today on Advanced Vector Extensions 512 (AVX-512) found on select Intel processors.

In a mailing list discussion stemming from the Phoronix article this week on the compiler instructions Intel is enabling for Alder Lake (and Sapphire Rapids), Linus Torvalds chimed in. The Alder Lake instructions being flipped on in GCC right now make no mention of AVX-512 but only AVX2 and others, likely due to Intel pursuing the subset supported by both the small and large cores in this new hybrid design being pursued.

Read more

Also: The Linux Team Approves New Neutral Terminology

background on AVX-512

In Slashdot now

  • Linus Torvalds Hopes Intel's AVX-512 'Dies A Painful Death'

    I hope AVX512 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.

    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.

    I've said this before, and I'll say it again: in the heyday of x86, when Intel was laughing all the way to the bank and killing all their competition, absolutely everybody else did better than Intel on FP loads. Intel's FP performance sucked (relatively speaking), and it matter not one iota.

    Because absolutely nobody cares outside of benchmarks.

    The same is largely true of AVX512 now - and in the future...

More on AVX-512

Linux founder tells Intel to stop inventing 'magic instructions'

  • Linux founder tells Intel to stop inventing 'magic instructions' and 'start fixing real problems'

    Linux Torvalds, the creator of Linux, offered up some interesting thoughts on Intel's Advanced Vector Extensions 512 (AVX-512) instruction set, calling it a "power virus" that was only created to make the company's CPU hardware perform well in benchmarks. He also admitted to being "biased" and "grumpy" in his assessment.

    His comments came in a mailing list (via Phoronix) discussing an article suggesting AVX-512 might not be part of Intel's upcoming Alder Lake architecture. If that comes to pass, it will be just fine by Torvalds.

    "I hope AVX512 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. 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," Torvalds said.

    Intel introduced AVX-512 in 2013, initially as part of its Xeon Phi x200 and Skylake-X processor lines. It has also found its way into more current CPU architectures, including Ice Lake.

Linus Torvalds: I hope Intel's AVX-512 'dies a painful death'

  • Linus Torvalds: I hope Intel's AVX-512 'dies a painful death'

    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 matter 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.

Intel slammed by Linux founder for using 'power virus' tech

  • Intel slammed by Linux founder for using 'power virus' tech

    Intel has been called out by Linux founder Linus Torvalds over the power usage of one of its most central technologies.

    Intel’s 512-bit AVX-512 SIMD extensions for x86 instruction set architecture are used for various compute-intensive workloads on workstations and servers, but AVX-512 hardware execution units are power hungry and that causes some headaches for developers.

    This led Linux creator Linus Torvalds to recently condemn AVX-512 and call on the company to develop a better solution for complex HPC problems. The software icon went as far as calling AVX-512 a ‘power virus’ and wished it to ‘die a painful death.’

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

More in Tux Machines

Today in Techrights

Android Leftovers

LibreOffice 6.4.5 finally for Slackware 14.2

The Document Foundation recently released version 7.0.0 of their Libre Office suite of applications. The packages for Slackware-current can be found in my repository. But the situation for Slackware 14.2 used to be different – I got stuck after LibreOffice 6.2 because the newer source releases (6.3 and onwards) require versions of system software that our stable Slackware 14.2 platform does not offer. From time to time during the last year, when there was time and the build box was not compiling packages, I messed around with the libreoffice.SlackBuild script in futile attempts to compile recent versions of LibreOffice on Slackware 14.2. I failed all the time. Until last week. After I had uploaded the new KDE Plasma5 packages to ‘ktown‘, I had an epiphany and decided to use a new approach. What I did was: question all the historic stuff in the SlackBuild script that got added whenever I needed to work around compilation failures; and accept that the compilation needs newer versions of software than Slackware 14.2 offers. The first statement meant that I disabled patches and variable declarations that messed with compiler and linker; and for the second statement I stuck to a single guideline: the end product, if I were able to compile a package successfully, has to run out of the box on Slackware 14.2 without the need to update any of the core Slackware packages. Read more

Web Browsers: New Tor RC, Firefox/Mozilla Trouble, and Web Browsers Need to Stop

  • New release candidate: 0.4.4.4-rc

    There's a new alpha release available for download. If you build Tor from source, you can download the source code for 0.4.4.4-rc from the download page. Packages should be available over the coming weeks, with a new alpha Tor Browser release likely in the coming weeks.

    Remember, this is a release candidate, not a a stable release: you should only run this if you'd like to find and report more bugs than usual.

  • Mozilla is dead

    If Mozilla wants to survive, the management will be fired with unearned compensation, the most important departments will be strengthened, products that nobody ordered will be discontinued and the organization will be limited to its core competence. Browser, email, security, adaptability and the fight for a free Internet. And they work with all their might to ensure that the products will become an integral part of everyday life and all operating systems.

    Three months. That’s all the time they have for a clear signal. After that, users have to make a decision. Unfortunately, it will probably only be something with chromium.

    Poor Internet.

  • Web browsers need to stop

    I call for an immediate and indefinite suspension of the addition of new developer-facing APIs to web browsers. Browser vendors need to start thinking about reducing scope and cutting features. WebUSB, WebBluetooth, WebXR, WebDRM WebMPAA WebBootlicking replacing User-Agent with Vendor-Agent cause let’s be honest with ourselves at this point “Encrypted Media Extensions” — this crap all needs to go. At some point you need to stop adding scope and start focusing on performance, efficiency, reliability, and security5 at the scope you already have.