Skip to content
  • Home Home Home, current page.
  • Moments Moments Moments, current page.

Saved searches

  • Remove
  • In this conversation
    Verified accountProtected Tweets @
Suggested users
  • Verified accountProtected Tweets @
  • Verified accountProtected Tweets @
  • Language: English
    • Bahasa Indonesia
    • Bahasa Melayu
    • Català
    • Čeština
    • Dansk
    • Deutsch
    • English UK
    • Español
    • Filipino
    • Français
    • Hrvatski
    • Italiano
    • Magyar
    • Nederlands
    • Norsk
    • Polski
    • Português
    • Română
    • Slovenčina
    • Suomi
    • Svenska
    • Tiếng Việt
    • Türkçe
    • Ελληνικά
    • Български език
    • Русский
    • Српски
    • Українська мова
    • עִבְרִית
    • العربية
    • فارسی
    • मराठी
    • हिन्दी
    • বাংলা
    • ગુજરાતી
    • தமிழ்
    • ಕನ್ನಡ
    • ภาษาไทย
    • 한국어
    • 日本語
    • 简体中文
    • 繁體中文
  • Have an account? Log in
    Have an account?
    · Forgot password?

    New to Twitter?
    Sign up
RichFelker's profile
Rich Felker
Rich Felker
Rich Felker
@RichFelker

Tweets

Rich Felker

@RichFelker

Yeah, I do @musllibc, FOSS & infosec stuff. But now is not the time for a mostly-/only-tech Twitter feed.

musl-libc.org
Joined March 2014

Tweets

  • © 2018 Twitter
  • About
  • Help Center
  • Terms
  • Privacy policy
  • Cookies
  • Ads info
Dismiss
Previous
Next

Go to a person's profile

Saved searches

  • Remove
  • In this conversation
    Verified accountProtected Tweets @
Suggested users
  • Verified accountProtected Tweets @
  • Verified accountProtected Tweets @

Promote this Tweet

Block

  • Tweet with a location

    You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more

    Your lists

    Create a new list


    Under 100 characters, optional

    Privacy

    Copy link to Tweet

    Embed this Tweet

    Embed this Video

    Add this Tweet to your website by copying the code below. Learn more

    Add this video to your website by copying the code below. Learn more

    Hmm, there was a problem reaching the server.

    By embedding Twitter content in your website or app, you are agreeing to the Twitter Developer Agreement and Developer Policy.

    Preview

    Why you're seeing this ad

    Log in to Twitter

    · Forgot password?
    Don't have an account? Sign up »

    Sign up for Twitter

    Not on Twitter? Sign up, tune into the things you care about, and get updates as they happen.

    Sign up
    Have an account? Log in »

    Two-way (sending and receiving) short codes:

    Country Code For customers of
    United States 40404 (any)
    Canada 21212 (any)
    United Kingdom 86444 Vodafone, Orange, 3, O2
    Brazil 40404 Nextel, TIM
    Haiti 40404 Digicel, Voila
    Ireland 51210 Vodafone, O2
    India 53000 Bharti Airtel, Videocon, Reliance
    Indonesia 89887 AXIS, 3, Telkomsel, Indosat, XL Axiata
    Italy 4880804 Wind
    3424486444 Vodafone
    » See SMS short codes for other countries

    Confirmation

     

    Welcome home!

    This timeline is where you’ll spend most of your time, getting instant updates about what matters to you.

    Tweets not working for you?

    Hover over the profile pic and click the Following button to unfollow any account.

    Say a lot with a little

    When you see a Tweet you love, tap the heart — it lets the person who wrote it know you shared the love.

    Spread the word

    The fastest way to share someone else’s Tweet with your followers is with a Retweet. Tap the icon to send it instantly.

    Join the conversation

    Add your thoughts about any Tweet with a Reply. Find a topic you’re passionate about, and jump right in.

    Learn the latest

    Get instant insight into what people are talking about now.

    Get more of what you love

    Follow more accounts to get instant updates about topics you care about.

    Find what's happening

    See the latest conversations about any topic instantly.

    Never miss a Moment

    Catch up instantly on the best stories happening as they unfold.

    1. mcc‏ @mcclure111 Apr 23

      mcc Retweeted Daniel Lemire

      I read this and am really confused why Intel would design the CPU in such a way that all this is even a concernhttps://twitter.com/lemire/status/987058338125971456 …

      mcc added,

      Daniel Lemire @lemire
      By how much does AVX-512 slow down your CPU? A first experiment. https://lemire.me/blog/2018/04/19/by-how-much-does-avx-512-slow-down-your-cpu-a-first-experiment/ …
      3 replies 0 retweets 16 likes
    2. whitequark‏ @whitequark Apr 23
      Replying to @mcclure111

      physics. AVX-512 functional units give you a *lot* of throughput, but also consume a lot of power and thus dissipate a lot of heat. running AVX-512 units in highest P-states would destroy the die.

      1 reply 1 retweet 14 likes
    3. whitequark‏ @whitequark Apr 23
      Replying to @whitequark @mcclure111

      if you utilize them continuously, then even with AVX offset clock you get higher aggregate throughput, although immediate clock frequency is reduced. if you only use them rarely then your overall throughput is lower because the core has a hard lower limit on the amount of time...

      2 replies 0 retweets 2 likes
    4. whitequark‏ @whitequark Apr 23
      Replying to @whitequark @mcclure111

      required to change its frequency, so after you get kicked back to a slower P-state and resume your non-AVX-512 workload, you can't immediately get back to the higher one.

      1 reply 0 retweets 3 likes
    5. whitequark‏ @whitequark Apr 23
      Replying to @whitequark @mcclure111

      AVX-512 is a weird remnant of Larrabee/Xeon Phi, it should never be used by autovectorizers and the like; only code specifically written to make use of AVX-512 really should use it

      2 replies 0 retweets 5 likes
    6. Steve Canon‏ @stephentyrone Apr 23
      Replying to @whitequark @mcclure111

      That's a little bit weird to say, since it has a bunch of features (left-over from LRB) that make it a much nicer target for auto-vectorizers. They should be conservative about using full-width operations, however, without explicit hinting.

      2 replies 0 retweets 1 like
    7. whitequark‏ @whitequark Apr 23
      Replying to @stephentyrone @mcclure111

      yes, I meant autovectorizers shouldn't use AVX-512 without hinting, not that they should never use it at all

      2 replies 0 retweets 0 likes
    8. Steve Canon‏ @stephentyrone Apr 23
      Replying to @whitequark @mcclure111

      In my ideal world, the front-end would "change gears" between generating 4x128b uOps, 2x256b uOps, and 1x512b based on the dynamic power state and AVX-512 instruction density, since the core has this information and (static) software rarely does.

      4 replies 0 retweets 3 likes
      Rich Felker‏ @RichFelker Apr 23
      Replying to @stephentyrone @whitequark @mcclure111

      AVX-512 also makes context switch time ridiculous. It really just shouldn't be used, or even enabled by the OS.

      9:34 AM - 23 Apr 2018
      • 1 Like
      • Axel Gneiting
      1 reply 0 retweets 1 like
        1. New conversation
        2. whitequark‏ @whitequark Apr 23
          Replying to @RichFelker @stephentyrone @mcclure111

          what if my application is scientific computing and I have exactly one process pinned to every core

          2 replies 0 retweets 2 likes
        3. Rich Felker‏ @RichFelker Apr 23
          Replying to @whitequark @stephentyrone @mcclure111

          Maybe then. But can it even perform significantly better once you take into account the P-states?

          1 reply 0 retweets 0 likes
        4. Steve Canon‏ @stephentyrone Apr 23
          Replying to @RichFelker @whitequark @mcclure111

          Yes. Hugely better.

          1 reply 0 retweets 3 likes
        5. Steve Canon‏ @stephentyrone Apr 23
          Replying to @stephentyrone @RichFelker and

          Keep in mind that 256b vector ops *also* move you down the turbo schedule, so you’re not comparing against max freq. Dip from 256-512 varies with stepping and #cores active, but nowhere near 1/2 speed.

          1 reply 0 retweets 0 likes
        6. mcc‏ @mcclure111 Apr 23
          Replying to @stephentyrone @RichFelker @whitequark

          this sucks and i am so glad i don't write compiler backends

          0 replies 0 retweets 3 likes
        7. End of conversation

      Loading seems to be taking a while.

      Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

        Promoted Tweet

        false

        • © 2018 Twitter
        • About
        • Help Center
        • Terms
        • Privacy policy
        • Cookies
        • Ads info