rawhide kobayashi
df952ea2d0
All checks were successful
blog.neet.works deployment / deploy_mamizou (push) Successful in 4s
41 lines
73 KiB
HTML
41 lines
73 KiB
HTML
<!doctype html><html lang=en dir=ltr class=scroll-smooth data-default-appearance=dark data-auto-appearance=false><head><meta charset=utf-8><meta http-equiv=content-language content="en"><meta name=viewport content="width=device-width,initial-scale=1"><meta http-equiv=X-UA-Compatible content="ie=edge"><title>Benchmarking AVX-512 Video Encoding On Zen 5 · N.E.E.T. Works</title>
|
||
<meta name=title content="Benchmarking AVX-512 Video Encoding On Zen 5 · N.E.E.T. Works"><meta name=description content="Comparing the performance of x265 and SVT-AV1 with AVX-512 On/Off on a Ryzen 9 9950X, as well as against an Intel i7-14700F."><meta name=keywords content="AMD,AVX-512,intel,x265,svt-av1,"><link rel=canonical href=https://blog.neet.works/posts/benchmarking-avx-512-video-encoding-zen-5/><link type=text/css rel=stylesheet href=/css/main.bundle.min.40e00100ae5ec4637cecd46ba4ae5f8db39bb83da61b5efe4900676fc0f4c10ac17e7da11e467fb8f1f39dfa5a5070c05bba06b4c05c629c2d389f4263d4c633.css integrity="sha512-QOABAK5exGN87NRrpK5fjbObuD2mG17+SQBnb8D0wQrBfn2hHkZ/uPHznfpaUHDAW7oGtMBcYpwtOJ9CY9TGMw=="><script type=text/javascript src=/js/appearance.min.516a16745bea5a9bd011138d254cc0fd3973cd55ce6e15f3dec763e7c7c2c7448f8fe7b54cca811cb821b0c7e12cd161caace1dd794ac3d34d40937cbcc9ee12.js integrity="sha512-UWoWdFvqWpvQERONJUzA/TlzzVXObhXz3sdj58fCx0SPj+e1TMqBHLghsMfhLNFhyqzh3XlKw9NNQJN8vMnuEg=="></script><script defer type=text/javascript id=script-bundle src=/js/main.bundle.min.a2d78d78672e549fbfc972ece871725b5478ba0b65708dda20cb97ab80a865eae6d247e1b05a4aec6ebbf78647ec3233bad8b2609ed98eee53cd58aa17128bc7.js integrity="sha512-oteNeGcuVJ+/yXLs6HFyW1R4ugtlcI3aIMuXq4CoZerm0kfhsFpK7G6794ZH7DIzutiyYJ7Zju5TzViqFxKLxw==" data-copy data-copied></script><link rel=apple-touch-icon sizes=180x180 href=/apple-touch-icon.png><link rel=icon type=image/png sizes=32x32 href=/favicon-32x32.png><link rel=icon type=image/png sizes=16x16 href=/favicon-16x16.png><link rel=manifest href=/site.webmanifest><meta property="og:url" content="https://blog.neet.works/posts/benchmarking-avx-512-video-encoding-zen-5/"><meta property="og:site_name" content="N.E.E.T. Works"><meta property="og:title" content="Benchmarking AVX-512 Video Encoding On Zen 5"><meta property="og:description" content="Comparing the performance of x265 and SVT-AV1 with AVX-512 On/Off on a Ryzen 9 9950X, as well as against an Intel i7-14700F."><meta property="og:locale" content="en"><meta property="og:type" content="article"><meta property="article:section" content="posts"><meta property="article:published_time" content="2025-02-10T23:21:38+00:00"><meta property="article:modified_time" content="2025-02-10T23:21:38+00:00"><meta property="article:tag" content="AMD"><meta property="article:tag" content="AVX-512"><meta property="article:tag" content="Intel"><meta property="article:tag" content="X265"><meta property="article:tag" content="Svt-Av1"><meta property="og:image" content="https://blog.neet.works/posts/benchmarking-avx-512-video-encoding-zen-5/featured.png"><meta name=twitter:card content="summary_large_image"><meta name=twitter:image content="https://blog.neet.works/posts/benchmarking-avx-512-video-encoding-zen-5/featured.png"><meta name=twitter:title content="Benchmarking AVX-512 Video Encoding On Zen 5"><meta name=twitter:description content="Comparing the performance of x265 and SVT-AV1 with AVX-512 On/Off on a Ryzen 9 9950X, as well as against an Intel i7-14700F."><script type=application/ld+json>[{"@context":"https://schema.org","@type":"Article","articleSection":"Posts","name":"Benchmarking AVX-512 Video Encoding On Zen 5","headline":"Benchmarking AVX-512 Video Encoding On Zen 5","description":"Comparing the performance of x265 and SVT-AV1 with AVX-512 On\/Off on a Ryzen 9 9950X, as well as against an Intel i7-14700F.","inLanguage":"en","url":"https:\/\/blog.neet.works\/posts\/benchmarking-avx-512-video-encoding-zen-5\/","author":{"@type":"Person","name":"rawhide kobayashi"},"copyrightYear":"2025","dateCreated":"2025-02-10T23:21:38\u002b00:00","datePublished":"2025-02-10T23:21:38\u002b00:00","dateModified":"2025-02-10T23:21:38\u002b00:00","keywords":["amd 9950x video encoding benchmark","amd svt-av1 benchmark","AVX-512 performance benchmarking","Best CPU for x265 and AV1 encoding","SVT-AV1 AVX-512 efficiency","SVT-AV1 encoding performance Zen 5","video encoding benchmark zen 5","x265 Zen 5 AVX-512 speedup","zen 5 avx 512 benchmark","Zen 5 AVX-512 impact","zen 5 avx512 benchmark","AMD Zen 5 video encoding benchmarks"],"mainEntityOfPage":"true","wordCount":"3420"}]</script><meta name=author content="rawhide kobayashi"><link href=mailto:rawhide@neet.works rel=me><link href=https://git.neet.works/rawhide_k rel=me><link href=https://github.com/rawhide-kobayashi rel=me><script src=/lib/jquery/jquery.slim.min.b0dca576e87d7eaa5850ae4e61759c065786cdb6489d68fcc82240539eebd5da522bdb4fda085ffd245808c8fe2acb2516408eb774ef26b5f6015fc6737c0ea8.js integrity="sha512-sNylduh9fqpYUK5OYXWcBleGzbZInWj8yCJAU57r1dpSK9tP2ghf/SRYCMj+KsslFkCOt3TvJrX2AV/Gc3wOqA=="></script><script defer src=/lib/packery/packery.pkgd.min.js integrity></script><script type=text/javascript src=/js/shortcodes/gallery.min.5fddb0ac85fed097f7b3630b6616e67caf1eea7ca23a42a6f7fc645f561306e7dabaef0a4eda37c74fb627ae3c97e11cb39b8895a4c839816632940e4a482508.js integrity="sha512-X92wrIX+0Jf3s2MLZhbmfK8e6nyiOkKm9/xkX1YTBufauu8KTto3x0+2J648l+Ecs5uIlaTIOYFmMpQOSkglCA=="></script><meta name=theme-color><script type=text/javascript src=/js/panzoom.min.b48202c3408ba8ef63ef9a6cc583e9d7ddb6574828cf79a4aaefe3c3964d0098ad77da46fdd5dddb66314ca975f6b9562093fa227ef51db7eba6d640facf448c.js integrity="sha512-tIICw0CLqO9j75psxYPp1922V0goz3mkqu/jw5ZNAJitd9pG/dXd22YxTKl19rlWIJP6In71HbfrptZA+s9EjA=="></script><script type=text/javascript src=/js/panzoom-util.687cf1af7ca1d075a5ce49ea7cef103a94c400423cc5b30927bf4db06c6a8ae0b8ee6f2b2bf501be099039b2f4ac5f4b348ed4af0429559ebb0743c54f051d5a.js integrity="sha512-aHzxr3yh0HWlzknqfO8QOpTEAEI8xbMJJ79NsGxqiuC47m8rK/UBvgmQObL0rF9LNI7UrwQpVZ67B0PFTwUdWg=="></script></head><body class="flex flex-col h-screen px-6 m-auto text-lg leading-7 max-w-7xl bg-neutral text-neutral-900 dark:bg-neutral-800 dark:text-neutral sm:px-14 md:px-24 lg:px-32 scrollbar-thin scrollbar-track-neutral-200 scrollbar-thumb-neutral-400 dark:scrollbar-track-neutral-800 dark:scrollbar-thumb-neutral-600"><div id=the-top class="absolute flex self-center"><a class="px-3 py-1 text-sm -translate-y-8 rounded-b-lg bg-primary-200 focus:translate-y-0 dark:bg-neutral-600" href=#main-content><span class="font-bold text-primary-600 ltr:pr-2 rtl:pl-2 dark:text-primary-400">↓</span>Skip to main content</a></div><div style=padding-left:0;padding-right:0;padding-top:2px;padding-bottom:3px class="main-menu flex items-center justify-between px-4 py-6 sm:px-6 md:justify-start space-x-3"><div class="flex flex-1 items-center justify-between"><nav class="flex space-x-3"><a href=/ class="text-base font-medium text-gray-500 hover:text-gray-900">N.E.E.T. Works</a></nav><nav class="hidden md:flex items-center space-x-5 md:ml-12 h-12"><a href=/posts/ class="flex items-center text-gray-500 hover:text-primary-600 dark:hover:text-primary-400"><p class="text-base font-medium" title=Posts>Blog</p></a><a href=/categories/ class="flex items-center text-gray-500 hover:text-primary-600 dark:hover:text-primary-400"><p class="text-base font-medium" title=Categories>Categories</p></a><a href=/tags/ class="flex items-center text-gray-500 hover:text-primary-600 dark:hover:text-primary-400"><p class="text-base font-medium" title=Tags>Tags</p></a><button id=search-button aria-label=Search class="text-base hover:text-primary-600 dark:hover:text-primary-400" title>
|
||
<span class="relative block icon"><svg aria-hidden="true" focusable="false" data-prefix="fas" data-icon="search" class="svg-inline--fa fa-search fa-w-16" role="img" viewBox="0 0 512 512"><path fill="currentcolor" d="M505 442.7 405.3 343c-4.5-4.5-10.6-7-17-7H372c27.6-35.3 44-79.7 44-128C416 93.1 322.9.0 208 0S0 93.1.0 208s93.1 208 208 208c48.3.0 92.7-16.4 128-44v16.3c0 6.4 2.5 12.5 7 17l99.7 99.7c9.4 9.4 24.6 9.4 33.9.0l28.3-28.3c9.4-9.4 9.4-24.6.1-34zM208 336c-70.7.0-128-57.2-128-128 0-70.7 57.2-128 128-128 70.7.0 128 57.2 128 128 0 70.7-57.2 128-128 128z"/></svg></span></button></nav><div class="flex md:hidden items-center space-x-5 md:ml-12 h-12"><span></span>
|
||
<button id=search-button-mobile aria-label=Search class="text-base hover:text-primary-600 dark:hover:text-primary-400" title>
|
||
<span class="relative block icon"><svg aria-hidden="true" focusable="false" data-prefix="fas" data-icon="search" class="svg-inline--fa fa-search fa-w-16" role="img" viewBox="0 0 512 512"><path fill="currentcolor" d="M505 442.7 405.3 343c-4.5-4.5-10.6-7-17-7H372c27.6-35.3 44-79.7 44-128C416 93.1 322.9.0 208 0S0 93.1.0 208s93.1 208 208 208c48.3.0 92.7-16.4 128-44v16.3c0 6.4 2.5 12.5 7 17l99.7 99.7c9.4 9.4 24.6 9.4 33.9.0l28.3-28.3c9.4-9.4 9.4-24.6.1-34zM208 336c-70.7.0-128-57.2-128-128 0-70.7 57.2-128 128-128 70.7.0 128 57.2 128 128 0 70.7-57.2 128-128 128z"/></svg></span></button></div></div><div class="-my-2 -mr-2 md:hidden"><label id=menu-button class=block><div class="cursor-pointer hover:text-primary-600 dark:hover:text-primary-400"><span class="relative block icon"><svg viewBox="0 0 448 512"><path fill="currentcolor" d="M0 96C0 78.33 14.33 64 32 64H416c17.7.0 32 14.33 32 32 0 17.7-14.3 32-32 32H32C14.33 128 0 113.7.0 96zM0 256c0-17.7 14.33-32 32-32H416c17.7.0 32 14.3 32 32s-14.3 32-32 32H32c-17.67.0-32-14.3-32-32zM416 448H32c-17.67.0-32-14.3-32-32s14.33-32 32-32H416c17.7.0 32 14.3 32 32s-14.3 32-32 32z"/></svg></span></div><div id=menu-wrapper style=padding-top:5px class="fixed inset-0 z-30 invisible w-screen h-screen m-0 overflow-auto transition-opacity opacity-0 cursor-default bg-neutral-100/50 backdrop-blur-sm dark:bg-neutral-900/50"><ul class="flex space-y-2 mt-3 flex-col items-end w-full px-6 py-6 mx-auto overflow-visible list-none ltr:text-right rtl:text-left max-w-7xl"><li id=menu-close-button><span class="cursor-pointer inline-block align-text-bottom hover:text-primary-600 dark:hover:text-primary-400"><span class="relative block icon"><svg viewBox="0 0 320 512"><path fill="currentcolor" d="M310.6 361.4c12.5 12.5 12.5 32.75.0 45.25C304.4 412.9 296.2 416 288 416s-16.38-3.125-22.62-9.375L160 301.3 54.63 406.6C48.38 412.9 40.19 416 32 416S15.63 412.9 9.375 406.6c-12.5-12.5-12.5-32.75.0-45.25l105.4-105.4L9.375 150.6c-12.5-12.5-12.5-32.75.0-45.25s32.75-12.5 45.25.0L160 210.8l105.4-105.4c12.5-12.5 32.75-12.5 45.25.0s12.5 32.75.0 45.25l-105.4 105.4L310.6 361.4z"/></svg></span></span></li><li class=mt-1><a href=/posts/ class="flex items-center text-gray-500 hover:text-primary-600 dark:hover:text-primary-400"><p class="text-bg font-bg" title=Posts>Blog</p></a></li><li class=mt-1><a href=/categories/ class="flex items-center text-gray-500 hover:text-primary-600 dark:hover:text-primary-400"><p class="text-bg font-bg" title=Categories>Categories</p></a></li><li class=mt-1><a href=/tags/ class="flex items-center text-gray-500 hover:text-primary-600 dark:hover:text-primary-400"><p class="text-bg font-bg" title=Tags>Tags</p></a></li></ul></div></label></div></div><div class="relative flex flex-col grow"><main id=main-content class=grow><article><div class="fixed inset-x-0 top-0 h-[800px] single_hero_background nozoom" style=background-image:url(/posts/benchmarking-avx-512-video-encoding-zen-5/featured_hu_a35b1e60e83f45b9.png)><div class="absolute inset-0 bg-gradient-to-t from-neutral dark:from-neutral-800 to-transparent mix-blend-normal"></div><div class="absolute inset-0 opacity-60 bg-gradient-to-t from-neutral dark:from-neutral-800 to-neutral-100 dark:to-neutral-800 mix-blend-normal"></div></div><section class="flex flex-col max-w-full mt-0 prose dark:prose-invert lg:flex-row"><div class="order-first lg:ml-auto px-0 lg:order-last ltr:lg:pl-8 rtl:lg:pr-8"><div class="toc ltr:pl-5 rtl:pr-5 print:hidden lg:sticky lg:top-10"><details open id=TOCView class="toc-right mt-0 overflow-y-scroll overscroll-contain scrollbar-thin scrollbar-track-neutral-200 scrollbar-thumb-neutral-400 dark:scrollbar-track-neutral-800 dark:scrollbar-thumb-neutral-600 rounded-lg ltr:-ml-5 ltr:pl-5 rtl:-mr-5 rtl:pr-5 hidden lg:block"><summary class="block py-1 text-lg font-semibold cursor-pointer bg-neutral-100 text-neutral-800 ltr:-ml-5 ltr:pl-5 rtl:-mr-5 rtl:pr-5 dark:bg-neutral-700 dark:text-neutral-100 lg:hidden">Table of Contents</summary><div class="min-w-[220px] py-2 border-dotted ltr:-ml-5 ltr:border-l ltr:pl-5 rtl:-mr-5 rtl:border-r rtl:pr-5 dark:border-neutral-600"><nav id=TableOfContents><ul><li><a href=#some-background>Some Background</a></li><li><a href=#but-actually-avx-512>But, actually, AVX-512</a></li><li><a href=#test-goals--parameters>Test Goals / Parameters</a></li><li><a href=#systems-setup>Systems Setup</a></li><li><a href=#results>Results!</a><ul><li><a href=#global-geomean>Global Geomean</a></li><li><a href=#9950x-vs-9950x-x265>9950X vs 9950X, x265</a></li><li><a href=#9950x-vs-9950x-svt-av1>9950X vs 9950X, SVT-AV1</a></li><li><a href=#i7-14700f-vs-9950x-x265>i7-14700F vs 9950X, x265</a></li><li><a href=#i7-14700f-vs-9950x-svt-av1>i7-14700F vs 9950X, SVT-AV1</a></li></ul></li><li><a href=#overall-conclusion>Overall Conclusion?</a></li></ul></nav></div></details><details class="toc-inside mt-0 overflow-hidden rounded-lg ltr:-ml-5 ltr:pl-5 rtl:-mr-5 rtl:pr-5 lg:hidden"><summary class="py-1 text-lg font-semibold cursor-pointer bg-neutral-100 text-neutral-800 ltr:-ml-5 ltr:pl-5 rtl:-mr-5 rtl:pr-5 dark:bg-neutral-700 dark:text-neutral-100 lg:hidden">Table of Contents</summary><div class="py-2 border-dotted border-neutral-300 ltr:-ml-5 ltr:border-l ltr:pl-5 rtl:-mr-5 rtl:border-r rtl:pr-5 dark:border-neutral-600"><nav id=TableOfContents><ul><li><a href=#some-background>Some Background</a></li><li><a href=#but-actually-avx-512>But, actually, AVX-512</a></li><li><a href=#test-goals--parameters>Test Goals / Parameters</a></li><li><a href=#systems-setup>Systems Setup</a></li><li><a href=#results>Results!</a><ul><li><a href=#global-geomean>Global Geomean</a></li><li><a href=#9950x-vs-9950x-x265>9950X vs 9950X, x265</a></li><li><a href=#9950x-vs-9950x-svt-av1>9950X vs 9950X, SVT-AV1</a></li><li><a href=#i7-14700f-vs-9950x-x265>i7-14700F vs 9950X, x265</a></li><li><a href=#i7-14700f-vs-9950x-svt-av1>i7-14700F vs 9950X, SVT-AV1</a></li></ul></li><li><a href=#overall-conclusion>Overall Conclusion?</a></li></ul></nav></div></details><script>var margin=200,marginError=50;(function(){var t=$(window),e=$("#TOCView"),s=e.height();function n(){var n=t.height()-margin;s>=n?(e.css("overflow-y","scroll"),e.css("max-height",n+marginError+"px")):(e.css("overflow-y","hidden"),e.css("max-height","9999999px"))}t.on("resize",n),$(document).ready(n)})()</script></div></div><div class="min-w-0 min-h-0 max-w-fit"><header id=single_header class="mt-5 max-w-prose"><h1 class="mt-0 text-4xl font-extrabold cool-title">Benchmarking AVX-512 Video Encoding On Zen 5</h1><div class=vari-flex><img class="!mt-0 !mb-0 h-24 w-24 rounded-full ltr:mr-4 rtl:ml-4" width=96 height=96 alt="rawhide kobayashi" src=/img/blowfish_logo_hu_97645b937385f547.webp><div class=place-self-center><div class="text-[0.6rem] uppercase leading-3 text-neutral-500 dark:text-neutral-400">Author</div><div class="font-semibold leading-6 text-neutral-800 dark:text-neutral-300">rawhide kobayashi</div><div class="text-sm text-neutral-700 dark:text-neutral-400">ᗜ⩊ᗜ</div><div class="text-2xl sm:text-lg"><div class="flex flex-wrap text-neutral-400 dark:text-neutral-500"><a class="px-1 hover:text-primary-700 dark:hover:text-primary-400" href=mailto:rawhide@neet.works target=_blank aria-label=Email rel="me noopener noreferrer"><span class="inline-block align-text-bottom"><span class="relative block icon"><svg viewBox="0 0 512 512"><path fill="currentcolor" d="M207.8 20.73c-93.45 18.32-168.7 93.66-187 187.1-27.64 140.9 68.65 266.2 199.1 285.1 19.01 2.888 36.17-12.26 36.17-31.49l1e-4-.6631c0-15.74-11.44-28.88-26.84-31.24-84.35-12.98-149.2-86.13-149.2-174.2.0-102.9 88.61-185.5 193.4-175.4 91.54 8.869 158.6 91.25 158.6 183.2v16.16c0 22.09-17.94 40.05-40 40.05s-40.01-17.96-40.01-40.05v-120.1c0-8.847-7.161-16.02-16.01-16.02l-31.98.0036c-7.299.0-13.2 4.992-15.12 11.68-24.85-12.15-54.24-16.38-86.06-5.106-38.75 13.73-68.12 48.91-73.72 89.64-9.483 69.01 43.81 128 110.9 128 26.44.0 50.43-9.544 69.59-24.88 24 31.3 65.23 48.69 109.4 37.49C465.2 369.3 496 324.1 495.1 277.2V256.3c0-149.2-133.9-265.632-287.3-235.57zM239.1 304.3c-26.47.0-48-21.56-48-48.05s21.53-48.05 48-48.05 48 21.56 48 48.05-20.6 48.05-48 48.05z"/></svg>
|
||
</span></span></a><a class="px-1 hover:text-primary-700 dark:hover:text-primary-400" href=https://git.neet.works/rawhide_k target=_blank aria-label=Gitea rel="me noopener noreferrer"><span class="inline-block align-text-bottom"><span class="relative block icon"><svg viewBox="5.67 143.05 628.65 387.55"><path fill="currentcolor" d="M115.912 143.075c-6.462.0-13.762.525-22.012 2.325-8.7 1.8-33.5 7.4-53.8 26.9C-4.9 212.4 6.6 276.2 8 285.8c1.7 11.7 6.9 44.2 31.7 72.5 45.8 56.1 144.4 54.8 144.4 54.8s12.1 28.9 30.6 55.5c25 33.1 50.7 58.9 75.7 62 63 0 188.9-.1 188.9-.1s12 .1 28.3-10.3c14-8.5 26.5-23.4 26.5-23.4S547 483 565 451.5c5.5-9.7 10.1-19.1 14.1-28 0 0 55.2-117.1 55.2-231.1-1.1-34.5-9.6-40.6-11.6-42.6-4.1-4.1-9.6-4-9.6-4s-117.2 6.6-177.9 8c-13.267.3-26.433.597-39.5.697l.1 117.002s57.4 24.202 83.1 40.102c3.7 2.3 10.2 6.798 12.9 14.398 2.1 6.1 2 13.101-1 19.301l-61 126.9c-6.2 12.7-21.4 18.1-33.9 12l-126.9-61c-12.5-6-17.9-21.2-11.8-33.8l61-126.9c6-12.5 21.2-17.9 33.8-11.8a5290.322 5290.322.0 0027 12.954c0-36.449-.1-109.053-.1-109.053-29 .4-89.2-2.201-89.2-2.201s-141.4-7.1-156.8-8.5c-4.9-.3-10.525-.825-16.988-.825zm12.188 48.026s7.1 59.399 15.7 94.199c7.2 29.2 24.8 77.7 24.8 77.7s-26.1-3.1-43-9.1c-25.9-8.5-36.9-18.7-36.9-18.7S69.6 321.8 60 295.4c-16.5-44.2-1.4-71.2-1.4-71.2s8.4-22.5 38.5-30c13.8-3.7 31-3.1 31-3.1z"/><path fill="currentcolor" d="M326.8 380.1c-8.2.1-15.4 5.8-17.3 13.8-1.9 8 2 16.3 9.1 20 7.7 4 17.5 1.8 22.7-5.4 5.1-7.1 4.3-16.9-1.8-23.1l24-49.1c1.5.1 3.7.2 6.2-.5 4.1-.9 7.1-3.6 7.1-3.6 4.2 1.8 8.6 3.8 13.2 6.1 4.8 2.4 9.3 4.9 13.4 7.3.9.5 1.8 1.1 2.8 1.9 1.6 1.3 3.4 3.1 4.7 5.5 1.9 5.5-1.9 14.9-1.9 14.9-2.3 7.6-18.4 40.6-18.4 40.6-8.1-.2-15.3 5-17.7 12.5-2.6 8.1 1.1 17.3 8.9 21.3 7.8 4 17.4 1.7 22.5-5.3 5-6.8 4.6-16.3-1.1-22.6 1.9-3.7 3.7-7.4 5.6-11.3 5-10.4 13.5-30.4 13.5-30.4.9-1.7 5.7-10.3 2.7-21.3-2.5-11.4-12.6-16.7-12.6-16.7-12.2-7.9-29.2-15.2-29.2-15.2s0-4.1-1.1-7.1c-1.1-3.1-2.8-5.1-3.9-6.3 4.7-9.7 9.4-19.3 14.1-29-4.1-2-8.1-4-12.2-6.1-4.8 9.8-9.7 19.7-14.5 29.5-6.7-.1-12.9 3.5-16.1 9.4-3.4 6.3-2.7 14.1 1.9 19.8l-24.6 50.4z"/></svg>
|
||
</span></span></a><a class="px-1 hover:text-primary-700 dark:hover:text-primary-400" href=https://github.com/rawhide-kobayashi target=_blank aria-label=Github rel="me noopener noreferrer"><span class="inline-block align-text-bottom"><span class="relative block icon"><svg viewBox="0 0 496 512"><path fill="currentcolor" d="M165.9 397.4c0 2-2.3 3.6-5.2 3.6-3.3.3-5.6-1.3-5.6-3.6.0-2 2.3-3.6 5.2-3.6 3-.3 5.6 1.3 5.6 3.6zm-31.1-4.5c-.7 2 1.3 4.3 4.3 4.9 2.6 1 5.6.0 6.2-2s-1.3-4.3-4.3-5.2c-2.6-.7-5.5.3-6.2 2.3zm44.2-1.7c-2.9.7-4.9 2.6-4.6 4.9.3 2 2.9 3.3 5.9 2.6 2.9-.7 4.9-2.6 4.6-4.6-.3-1.9-3-3.2-5.9-2.9zM244.8 8C106.1 8 0 113.3.0 252c0 110.9 69.8 205.8 169.5 239.2 12.8 2.3 17.3-5.6 17.3-12.1.0-6.2-.3-40.4-.3-61.4.0.0-70 15-84.7-29.8.0.0-11.4-29.1-27.8-36.6.0.0-22.9-15.7 1.6-15.4.0.0 24.9 2 38.6 25.8 21.9 38.6 58.6 27.5 72.9 20.9 2.3-16 8.8-27.1 16-33.7-55.9-6.2-112.3-14.3-112.3-110.5.0-27.5 7.6-41.3 23.6-58.9-2.6-6.5-11.1-33.3 2.6-67.9 20.9-6.5 69 27 69 27 20-5.6 41.5-8.5 62.8-8.5s42.8 2.9 62.8 8.5c0 0 48.1-33.6 69-27 13.7 34.7 5.2 61.4 2.6 67.9 16 17.7 25.8 31.5 25.8 58.9.0 96.5-58.9 104.2-114.8 110.5 9.2 7.9 17 22.9 17 46.4.0 33.7-.3 75.4-.3 83.6.0 6.5 4.6 14.4 17.3 12.1C428.2 457.8 496 362.9 496 252 496 113.3 383.5 8 244.8 8zM97.2 352.9c-1.3 1-1 3.3.7 5.2 1.6 1.6 3.9 2.3 5.2 1 1.3-1 1-3.3-.7-5.2-1.6-1.6-3.9-2.3-5.2-1zm-10.8-8.1c-.7 1.3.3 2.9 2.3 3.9 1.6 1 3.6.7 4.3-.7.7-1.3-.3-2.9-2.3-3.9-2-.6-3.6-.3-4.3.7zm32.4 35.6c-1.6 1.3-1 4.3 1.3 6.2 2.3 2.3 5.2 2.6 6.5 1 1.3-1.3.7-4.3-1.3-6.2-2.2-2.3-5.2-2.6-6.5-1zm-11.4-14.7c-1.6 1-1.6 3.6.0 5.9 1.6 2.3 4.3 3.3 5.6 2.3 1.6-1.3 1.6-3.9.0-6.2-1.4-2.3-4-3.3-5.6-2z"/></svg></span></span></a></div></div></div><div style=padding-left:8px class="mt-1 mb-6 text-base text-neutral-500 dark:text-neutral-400 print:hidden max-w-fit"><div class="flex flex-row flex-wrap items-center"><time datetime=2025-02-10T23:21:38+00:00>February 10 2025</time><span class="px-2 text-primary-500">·</span><span>3420 words</span></div><hr><div class="flex flex-row flex-wrap items-center"><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/categories/benchmarking/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Benchmarking
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/categories/video-encoding/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Video Encoding
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/amd/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">AMD
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/avx-512/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">AVX-512
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/intel/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Intel
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/x265/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">X265
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/svt-av1/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Svt-Av1</span></span></span></div></div></div><hr></header><div class="article-content max-w-prose mb-20"><h2 class="relative group">Some Background<div id=some-background class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#some-background aria-label=Anchor>#</a></span></h2><p>The Zen 5 launch was largely considered, by Gamers™️, to be a disaster. There were accusations of AMD intentionally creating entirely false graphs, accusations of green-type greed for failing to cut >$100 off of the MSRP of brand-new SKUs to compete on value with older SKUs, and all other sorts of nonsense because Gamers™️ didn’t get another ++30% generational performance uplift like between Zen 3 and Zen 4. AMD’s marketing department should actually be hung out to dry for their multiple, high-profile, embarrassingly inaccurate and/or misleading graphs over the years, but that’s neither here nor there. Shortly afterwards, Arrow Lake launched to raucus apathy, Zen 5 prices dropped to market value, and the 9800x3D became the best gaming CPU in the world. AMD are now Certified Good Guys once again.</p><p>The most outlandish claim AMD made at the time was the performance improvement in HandBrake.</p><hr><div id=gallery-77db58ae913c530fff4be939b4c3b3e8 class=gallery><div class="panzoom-container grid-w50"><img class="my-0 rounded-md zoomable" src=images/compressed/AMD%202024_Tech%20Day_David%20McAfee-05.webp alt="Charts showing claimed performance improvements of the 9900x compared to the i9-14900k." data-src=images/compressed/AMD%202024_Tech%20Day_David%20McAfee-05.webp></div><div class="panzoom-container grid-w50"><img class="my-0 rounded-md zoomable" src=images/compressed/AMD%202024_Tech%20Day_David%20McAfee-06.webp alt="Charts showing claimed performance improvements of the 9800x compared to the i7-14700k." data-src=images/compressed/AMD%202024_Tech%20Day_David%20McAfee-06.webp></div><div class="panzoom-container grid-w100"><img class="my-0 rounded-md zoomable" src=images/compressed/AMD%202024_Tech%20Day_David%20McAfee-07.webp alt="Charts showing claimed performance improvements of the 9600x compared to the i5-14600k." data-src=images/compressed/AMD%202024_Tech%20Day_David%20McAfee-07.webp></div></div><figcaption class=caption>Various charts claiming various performance improvements of Zen 5 CPUs compared to Raptor Lake Intel CPUs.</figcaption><hr><p>Quite impressive, especially when comparing against SKUs with double-or-more core counts thanks to E-cores. What contrived circumstances did they pull to get a result like that? Six cores beating 6P+8E? It seemed ridiculous. The referenced end note card had these completely useless statements.</p><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/david-mcafee-21.webp srcset="/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/david-mcafee-21_hu_7e3758c710ac5fbd.webp 330w,
|
||
/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/david-mcafee-21_hu_c56cc54ecacefddf.webp 660w,
|
||
/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/david-mcafee-21_hu_c8195fd0d251e380.webp 1320w,
|
||
/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/david-mcafee-21.webp 2x" alt="Endnote card detailing the benchmark configurations from the above slides." data-src=images/compressed/david-mcafee-21.webp></div><figcaption class=caption>Endnote card detailing the benchmark configurations from the above slides.</figcaption><hr><p>No information about the video resolution, bit depth, encoding settings, encoder, or… Literally anything you might want to detail to prevent people from thinking you just made up a number. Just, “HandBrake.” Thank you AMD, very cool! Some people accused them of accidentally using <em>hardware acceleration</em> while performing these benchmarks. Unfortunately, in this instance, this only serves to prove the accuser’s own ignorance, as hardware accelerated encoding has <em>significantly</em> higher performance gains than 41-94%. Except…</p><h2 class="relative group">But, actually, AVX-512<div id=but-actually-avx-512 class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#but-actually-avx-512 aria-label=Anchor>#</a></span></h2><p>Zen 5 <em>does</em> have an advantage that was ignored by the “muh value” glazers. From a certain point of view, you could actually call it hardware acceleration, but it’s not in the way that the detractors were claiming at the time. Zen 5 brought full-fat, true 512-bit AVX-512 execution units and data paths. Four of them per core!<sup id=fnref:1><a href=#fn:1 class=footnote-ref role=doc-noteref>1</a></sup></p><p>AVX-512 is the 512-bit extension of the Advanced Vector Extensions instruction set, which is in the SIMD - Single Instruction, Multiple Data - family. It is an instruction designed to faciliate the simultaneous performance of mathematical operations on sets of numbers. Up to 512 bits’ worth. You combine smaller sets of numbers - the <strong>Vector</strong> in Advanced Vector Extensions - and the CPU can then perform an operation on each number in that set, simultaneously, in a single clock cycle. With the four full-width AVX-512 execution units in every Zen 5 core, every CPU can (theoretically) do 2048 bits’ worth of calculations, per core, per clock cycle. This is spherical cow in a vacuum territory, especially with the painful limitation of dual-channel memory, but on the 9950X, which has 16 cores, that means you could do operations on up to 1024<sup id=fnref:2><a href=#fn:2 class=footnote-ref role=doc-noteref>2</a></sup> 32-bit numbers simultaneously, on <em>every single clock cycle</em>.</p><p>For those who have not been keeping track of such things, AVX-512 has been a staple in Intel server SKUs since Skylake, but their consumer SKUs lacked it until Rocket Lake. Alder Lake introduced the asymmetrical P/E core paradigm on desktop SKUs, and it <em>launched</em> with AVX-512 - but only on the P-cores, which caused problems with then-current schedulers. They did not know how to deal with heterogenous instruction sets, and as a result, AVX-512 applications would crash when a thread moved to an E-core. Access to the instruction was eventually removed via microcode/BIOS updates, and then fused off physically on newer production runs. Intel has, as of this writing, yet to reintroduce AVX-512 to the consumer market, though it continues to bring meaningful performance benefits to their server platforms.</p><p>AMD lacked AVX-512 support for all their products until Zen 4, but when they did get there, they implemented it across their entire hardware stack. For Zen 4, rather than full 512-bit hardware, they had “double pumped” 256-bit hardware that took two clock cycles to complete an instruction rather than one. This has a number of beneficial implications in terms of power consumption, silicon area, and the ability to re-use existing 256-bit silicon for the AVX-512 execution units. It holds back the maximum possible performance, but when you’re competing against Intel’s <em>utter lack</em> of AVX-512 in consumer chips, that’s an ♾️% advantage!</p><p>AVX instructions have a history of derision among consumers, particularly gamers and overclockers. AVX execution units need large areas of silicon, because they work on large amounts of data. Logically, per clock cycle, it will use more power, generate more heat, and do more work, because there are more transistors involved in executing the instructions. To make up for that, you will have to run fewer clock cycles per second to avoid overheating and excessive power draw that may cause voltge droop. There’s a balancing act between clock speed and the accelerated overall speed of computation that AVX enables. Intel has historically maintained a very poor balance between those factors, leading to unstable overclocks, complaints about power consumption and thermals, and sometimes objectively reduced overall performance. It doesn’t help that consumers are highly uneducated about thermal management, either. 100*C is <em>fine</em>, but a gamer won’t accept it.</p><p>As a result of Intel’s inelegant handling of these requirements, AVX has garnered a bad rap in the general populace. The vast majority of applications are still compiled without AVX2, let alone AVX-512. Zen 4/5 desktop CPUs may have been bestsellers for a while now, but relative to the global population, very few people own CPUs with AVX-512 support. Some people are still out there using Core 2 Duos or other very, very low-spec Pentium/Atom chips that lack support for AVX2! In addition, most software isn’t written in a hyper-optimized fashion that has inline assembly or AVX intrinsics, and compilers are not very good at auto-vectorizing code that was not written in a format that is intended to be vectorized. Consumers don’t tend to process vast amounts of data that would benefit from the capabilities of AVX instructions anyway, right? So who really cares?!</p><p>Well, AMD does, and I’m glad that they do. There is one area where (relatively) average consumers need heavy compute, and that’s video encoding. Hardware acceleration has gotten pretty good for casual use, especially for livestreaming. The energy efficiency and quality at low-latency is impossible for a software encoder to beat. However, if software is given the chance to stretch its legs latency/processing time-wise, the quality-per-bit, or compression efficiency, just can’t be beat. Plus, hardware encoders, as fixed-function silicon, don’t really get to keep up with new innovations in software design. AV1 may be fast if you use NVENC, but will it match the latest release of SVT-AV1-PSY at any given bitrate? Absolutely not. Besides, there are a variety of other operations in video processing that can benefit from (non-disastrously-downclocking) AVX-512 (provided the software was written correctly!) that are unrelated to the final encoding task.</p><p>Video encoding software like x265 and SVT-AV1, which are what I will be using for my test here, contain large quantities of hand-written assembly optimized for various SIMD instruction sets, including AVX-512. They exist regardless of the compiler flags used to build the software. Every build should be capable of using AVX-512 acceleration, and everyone should be able to reap the benefits without seeking out special builds of these pieces of software.</p><p>There are a litany of asterisks to go along with all that information, and I’m not an expert in assembly or CPU design. There’s a much more detailed breakdown of Zen 5’s AVX-512 over on <a href=http://www.numberworld.org/blogs/2024_8_7_zen5_avx512_teardown/ target=_blank>numberworld.org.</a> Go ahead and give it a read if you’re interested! Phoronix also has a <a href=https://www.phoronix.com/review/amd-epyc-9755-avx512 target=_blank>benchmark</a> comparing the performance of a Zen 5 Turin server SKU with AVX-512 off, in double-pumped 256-bit mode, and full 512-bit mode in a variety of applications.</p><h2 class="relative group">Test Goals / Parameters<div id=test-goals--parameters class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#test-goals--parameters aria-label=Anchor>#</a></span></h2><p>While AMD, for some indeterminable reason, did not make this obvious in their presentation, the gains in HandBrake could be attributed to the AVX-512 improvements present in Zen 5, and the continued efforts from video encoders to provide optimized AVX-512 code. HandBrake ships x265 as its HEVC encoder, and SVT-AV1 as its AV1 encoder, so those are what I will be testing here.</p><p>I have two goals with this test:</p><ul><li>Quantify the performance differences between a Zen 5 CPU with AVX-512 off, and on</li><li>Quantify the performance differences between a Zen 5 CPU with AVX-512 on, and a Raptor Lake CPU</li></ul><p>The first goal will implicate exactly how much the presence of AVX-512 could theoretically improve performance on Intel, should they adopt it. The second goal will determine to what degree AMD may have engaged in selective benchmarking tomfoolery in the earlier slides.</p><p>Most reviewers don’t have any knowledge of encoders beyond extremely surface-level use of HandBrake or export options in video editing software. They just toss a file into HandBrake, pick something - hopefully it’s consistent between tested products! - and hit go. Maybe they use the Phoronix Test Suite if they’re a Linux shop, but that doesn’t adequately cover the bases regarding A-B testing the impact of AVX-512 on performance on a single SKU. I waited six months and found zero reviews examining this specific topic to my satisfaction. Now, incidentally, I had reason to purchase a Zen 5 CPU, and decided to bench it for myself.</p><p>The tests are as simple as possible. There are far, far too many possible combinations of command line options to test within any kind of reasonable amount of time, and I don’t think it’s useful to test in that way. Most people just choose a built-in preset, maybe a <code>-tune</code> parameter, but that’s the extent of their customization. I chose to perform a simple, like-for-like comparison of x265 and SVT-AV1. I used the three most common video resolutions - 720p, 1080p, and 4k - swept through every stock preset in x265 and SVT-AV1, under three different hardware configurations: Raptor Lake, Zen 5 AVX-512 Off, and Zen 5 AVX-512 On. I ran each configuration five times, and took the average of the combined wall-to-wall run times as my measurement. Then, I created graphs exhibiting the execution time improvements compared to the presumed slower configuration. They are available below, but there is more relevant information to get to before that.</p><p>By default, x265 does not enable AVX-512, even on supported systems, even if you do build it with relevant microarchitecture features enabled. You have to pass the parameter <code>asm=avx512</code> to enable it. HandBrake does <em>not</em> pass this parameter by default, either. You have to do it manually in the “Advanced Options” section. SVT-AV1 <em>does</em> enable AVX-512 by default, and for this test, I had to <em>limit</em> the featureset with the <code>asm=9</code>. This restricts SVT-AV1 to AVX2 and older features.</p><p>In addition to the run time, I also catalogged some other details like the average MHz and power draw as reported by turbostat, and reported average die temperature by sensors, but I didn’t find the results very interesting (or accurate, in some cases) so they’ve been omitted from the below analysis. If you’d like to look at my test scripts and the raw data, you can do so at the repository below.</p><a id=gitea-938f96d620681c1efd16d5decceab89e target=_blank href=https://git.neet.works/rawhide_k/zen-5-avx-512-encoding-benchmark class=cursor-pointer><div class="w-full md:w-auto pt-3 p-5 border border-neutral-200 dark:border-neutral-700 border rounded-md shadow-2xl"><div class="flex items-center"><span class="text-2xl text-neutral-800 dark:text-neutral" style=margin-right:10px><span class="relative block icon"><svg viewBox="5.67 143.05 628.65 387.55"><path fill="currentcolor" d="M115.912 143.075c-6.462.0-13.762.525-22.012 2.325-8.7 1.8-33.5 7.4-53.8 26.9C-4.9 212.4 6.6 276.2 8 285.8c1.7 11.7 6.9 44.2 31.7 72.5 45.8 56.1 144.4 54.8 144.4 54.8s12.1 28.9 30.6 55.5c25 33.1 50.7 58.9 75.7 62 63 0 188.9-.1 188.9-.1s12 .1 28.3-10.3c14-8.5 26.5-23.4 26.5-23.4S547 483 565 451.5c5.5-9.7 10.1-19.1 14.1-28 0 0 55.2-117.1 55.2-231.1-1.1-34.5-9.6-40.6-11.6-42.6-4.1-4.1-9.6-4-9.6-4s-117.2 6.6-177.9 8c-13.267.3-26.433.597-39.5.697l.1 117.002s57.4 24.202 83.1 40.102c3.7 2.3 10.2 6.798 12.9 14.398 2.1 6.1 2 13.101-1 19.301l-61 126.9c-6.2 12.7-21.4 18.1-33.9 12l-126.9-61c-12.5-6-17.9-21.2-11.8-33.8l61-126.9c6-12.5 21.2-17.9 33.8-11.8a5290.322 5290.322.0 0027 12.954c0-36.449-.1-109.053-.1-109.053-29 .4-89.2-2.201-89.2-2.201s-141.4-7.1-156.8-8.5c-4.9-.3-10.525-.825-16.988-.825zm12.188 48.026s7.1 59.399 15.7 94.199c7.2 29.2 24.8 77.7 24.8 77.7s-26.1-3.1-43-9.1c-25.9-8.5-36.9-18.7-36.9-18.7S69.6 321.8 60 295.4c-16.5-44.2-1.4-71.2-1.4-71.2s8.4-22.5 38.5-30c13.8-3.7 31-3.1 31-3.1z"/><path fill="currentcolor" d="M326.8 380.1c-8.2.1-15.4 5.8-17.3 13.8-1.9 8 2 16.3 9.1 20 7.7 4 17.5 1.8 22.7-5.4 5.1-7.1 4.3-16.9-1.8-23.1l24-49.1c1.5.1 3.7.2 6.2-.5 4.1-.9 7.1-3.6 7.1-3.6 4.2 1.8 8.6 3.8 13.2 6.1 4.8 2.4 9.3 4.9 13.4 7.3.9.5 1.8 1.1 2.8 1.9 1.6 1.3 3.4 3.1 4.7 5.5 1.9 5.5-1.9 14.9-1.9 14.9-2.3 7.6-18.4 40.6-18.4 40.6-8.1-.2-15.3 5-17.7 12.5-2.6 8.1 1.1 17.3 8.9 21.3 7.8 4 17.4 1.7 22.5-5.3 5-6.8 4.6-16.3-1.1-22.6 1.9-3.7 3.7-7.4 5.6-11.3 5-10.4 13.5-30.4 13.5-30.4.9-1.7 5.7-10.3 2.7-21.3-2.5-11.4-12.6-16.7-12.6-16.7-12.2-7.9-29.2-15.2-29.2-15.2s0-4.1-1.1-7.1c-1.1-3.1-2.8-5.1-3.9-6.3 4.7-9.7 9.4-19.3 14.1-29-4.1-2-8.1-4-12.2-6.1-4.8 9.8-9.7 19.7-14.5 29.5-6.7-.1-12.9 3.5-16.1 9.4-3.4 6.3-2.7 14.1 1.9 19.8l-24.6 50.4z"/></svg></span></span><div id=gitea-938f96d620681c1efd16d5decceab89e-full_name class="m-0 font-bold text-xl text-neutral-800 decoration-primary-500 hover:underline hover:underline-offset-2 dark:text-neutral">rawhide_k/zen-5-avx-512-encoding-benchmark</div></div><p id=gitea-938f96d620681c1efd16d5decceab89e-description class="m-0 mt-2 text-md text-neutral-800 dark:text-neutral"></p><div class="m-0 mt-2 flex items-center"><span class="mr-1 inline-block h-3 w-3 rounded-full" style=background-color:#3572a5></span><div class="m-0 mr-5 text-md text-neutral-800 dark:text-neutral">Python</div><span class="text-md mr-1 text-neutral-800 dark:text-neutral"><span class="relative block icon"><svg viewBox="0 0 576 512"><path fill="currentcolor" d="M287.9.0C297.1.0 305.5 5.25 309.5 13.52L378.1 154.8l153.3 22.7C540.4 178.8 547.8 185.1 550.7 193.7 553.5 202.4 551.2 211.9 544.8 218.2L433.6 328.4l26.3 155.5C461.4 492.9 457.7 502.1 450.2 507.4 442.8 512.7 432.1 513.4 424.9 509.1L287.9 435.9 150.1 509.1C142.9 513.4 133.1 512.7 125.6 507.4 118.2 502.1 114.5 492.9 115.1 483.9l27.1-155.5L31.11 218.2C24.65 211.9 22.36 202.4 25.2 193.7 28.03 185.1 35.5 178.8 44.49 177.5L197.7 154.8 266.3 13.52C270.4 5.249 278.7.0 287.9.0zm0 78.95L235.4 187.2C231.9 194.3 225.1 199.3 217.3 200.5L98.98 217.9 184.9 303C190.4 308.5 192.9 316.4 191.6 324.1L171.4 443.7l105.2-56.2C283.7 383.7 292.2 383.7 299.2 387.5l105.2 56.2-20.2-119.6C382.9 316.4 385.5 308.5 391 303l85.9-85.1-118.3-17.4C350.7 199.3 343.9 194.3 340.5 187.2L287.9 78.95z"/></svg></span></span><div id=gitea-938f96d620681c1efd16d5decceab89e-stars_count class="m-0 mr-5 text-md text-neutral-800 dark:text-neutral">0</div><span class="text-md mr-1 text-neutral-800 dark:text-neutral"><span class="relative block icon"><svg viewBox="0 0 448 512"><path fill="currentcolor" d="M80 104c13.3.0 24-10.7 24-24S93.3 56 80 56 56 66.7 56 80s10.7 24 24 24zm80-24c0 32.8-19.7 61-48 73.3V192c0 17.7 14.3 32 32 32H304c17.7.0 32-14.3 32-32V153.3C307.7 141 288 112.8 288 80c0-44.2 35.8-80 80-80s80 35.8 80 80c0 32.8-19.7 61-48 73.3V192c0 53-43 96-96 96H256v70.7c28.3 12.3 48 40.5 48 73.3.0 44.2-35.8 80-80 80s-80-35.8-80-80c0-32.8 19.7-61 48-73.3V288H144c-53 0-96-43-96-96V153.3C19.7 141 0 112.8.0 80 0 35.8 35.8.0 80 0s80 35.8 80 80zm208 24c13.3.0 24-10.7 24-24s-10.7-24-24-24-24 10.7-24 24 10.7 24 24 24zM248 432c0-13.3-10.7-24-24-24s-24 10.7-24 24 10.7 24 24 24 24-10.7 24-24z"/></svg></span></span><div id=gitea-938f96d620681c1efd16d5decceab89e-forks_count class="m-0 mr-5 text-md text-neutral-800 dark:text-neutral">0</div></div></div><script>fetch("https://git.neet.works/api/v1/repos/rawhide_k/zen-5-avx-512-encoding-benchmark",{headers:new Headers({"User-agent":"Mozilla/4.0 Custom User Agent"})}).then(e=>e.json()).then(e=>{document.getElementById("gitea-938f96d620681c1efd16d5decceab89e-full_name").innerHTML=e.full_name,document.getElementById("gitea-938f96d620681c1efd16d5decceab89e-description").innerHTML=e.description,document.getElementById("gitea-938f96d620681c1efd16d5decceab89e-stars_count").innerHTML=e.stars_count,document.getElementById("gitea-938f96d620681c1efd16d5decceab89e-forks_count").innerHTML=e.forks_count}).catch(e=>console.error(e))</script></a><h2 class="relative group">Systems Setup<div id=systems-setup class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#systems-setup aria-label=Anchor>#</a></span></h2><p>For this test, I had two systems. One was based on the Ryzen 9 9950X, and the other on the Intel i7-14700F. Both systems have 2x32GB memory kits running at 6000MHz, though they aren’t identical. Don’t you worry, the slight variations on the timings are utterly irrelevant. Both systems ran identical software configurations - as identical as possible, at least, considering the architecture differences. They both ran Arch Linux with CachyOS optimized repositories - x86-64-v3 for the i7-14700F, and znver4 for the 9950X. The kernel version was <code>6.12.10-2-cachyos-lts</code>. Other relevant package versions were <code>ffmpeg 2:7.1-6.1</code>, <code>x265 4.0-1.1</code>, and <code>svt-av1 2.3.0-2</code>.</p><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/9950x.webp srcset="/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/9950x_hu_1f194e8e65bbf185.webp 330w,
|
||
/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/9950x_hu_58b51abc7faf8b99.webp 660w,
|
||
/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/9950x_hu_2fcc323ba560c0d7.webp 1320w,
|
||
/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/9950x.webp 2x" alt="AMD Ryzen 9 9950X based test bench." data-src=images/compressed/9950x.webp></div><figcaption class=caption>AMD Ryzen 9 9950X based test bench. The motherboard is an ASRock B650E PG Riptide Wi-Fi</figcaption><hr><p>The 9950X had the socket power limit set to the stock 200w, and the current limit to the stock 160A. This is just about exactly what an NH-D15 can dissipate, when using a graphite thermal pad, as I did with this setup. Not great, not terrible, but it’s what I had on hand. It didn’t thermal throttle at stock settings, so that’s good enough for me, for this test. The only options I changed regarding performance are memory related, enabling XMP and dropping vSOC to 1.1v. No PBO, no undervolting, stock fmax. 2000MHz fCLK and 3000MHz uCLK, as typical of 6000MHz memory.</p><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/14700f.webp srcset="/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/14700f_hu_7c591360c604f05b.webp 330w,
|
||
/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/14700f_hu_ceacc3c4be6c83f8.webp 660w,
|
||
/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/14700f_hu_3c43105490638076.webp 1320w,
|
||
/posts/benchmarking-avx-512-video-encoding-zen-5/images/compressed/14700f.webp 2x" alt="Intel i7-14700F based test bench." data-src=images/compressed/14700f.webp></div><figcaption class=caption>Intel i7-14700F based test bench.</figcaption><hr><p>The i7-14700F is a non-K SKU, so you can’t overclock it. On some motherboards you can undervolt non-K SKUs, but not the one that I have. It’s some kind of stripped down model ASUS uses for prebuilts. You can change the vdroop, which I have adjusted to whichever setting gave me optimal performance, but the differences were extremely minor, and that’s really all you can do with it. I have the power/current limits technically uncapped, but the board has a hard current limit somewhere around 220-280w power draw, load type dependent. With a 240mm liquid cooler, temperature is not a concern. The only hardware-based limit on the performance is the motherboard’s current limit.</p><p>Now you might say, “But Mr. Blogger! None of the slides earlier in the deck had a 9950X or an i7-14700F! This comparison is not fair, and you are a hack fraudster!” To which I would say… Yes, absolutely. It’s not a fair comparison, and I’m not going to prove any of the aforementioned slides right or wrong. However, if you look at <a href=https://www.phoronix.com/review/intel-core-ultra-9-285k-linux/14 target=_blank>this Phoronix benchmark</a>, you can observe that there’s not a huge difference between the 9900X and 9950X in SVT-AV1 and x265, nor between a selection of Raptor Lake chips. Scaling suffers greatly beyond twelve cores, even at 4k, unless you specifically invoke paralellism-enhancing commands that cause the compression efficiency to suffer. Chunking up a video and running multiple encode jobs to get the absolute maximum possible performance out of a given CPU with a given encoder with the best possible compression efficiency is a whole ’nother topic. By the way, the Phoronix Test Suite does <em>not</em> enable AVX-512 in x265. Their numbers would be much further apart between AMD and Intel if they did. Please feel free to email me if you have a desire to send me free hardware to conduct additional testing!</p><h2 class="relative group">Results!<div id=results class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#results aria-label=Anchor>#</a></span></h2><p>For those with a short attention span, here’s the gist.</p><ul><li>The 9950X demolishes the i7-14700F, as you would hope, with double the “P”-core count and AVX-512 present.</li><li>AVX-512 gains are not significant with presets faster than “slow” with x265, or faster than “4” with SVT-AV1.</li><li>SVT-AV1 benefits less from AVX-512 than I expected overall, given the fact it’s newer and sees more consistent development.</li><li>4k brings out the biggest differences both in AVX-512 and between the processors in general, as expected.</li><li>Faster presets and lower resolutions are more dependent on single core performance, with even worse scaling under the default conditions that SVT-AV1 and x265 operate under.</li></ul><h3 class="relative group">Global Geomean<div id=global-geomean class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#global-geomean aria-label=Anchor>#</a></span></h3><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=images/compressed/Geomean%20performance%20difference%20at%20major%20resolutions,%20across%20all%20presets.webp alt data-src=images/compressed/Geomean%20performance%20difference%20at%20major%20resolutions,%20across%20all%20presets.webp></div><hr><p>Geomean kinda sucks with the spread of values here. The superduperfast presets really bring the averages down, especially with x265. This graph is almost entirely useless. Read on for information on your specific preset and encoder of interest.</p><h3 class="relative group">9950X vs 9950X, x265<div id=9950x-vs-9950x-x265 class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#9950x-vs-9950x-x265 aria-label=Anchor>#</a></span></h3><hr><div id=gallery-91ee588228b9c611b00d20aac5022b3f class=gallery><div class="panzoom-container grid-w100"><img class="my-0 rounded-md zoomable" src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%20720p,%20x265.webp alt data-src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%20720p,%20x265.webp></div><div class="panzoom-container grid-w100"><img class="my-0 rounded-md zoomable" src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%201080p,%20x265.webp alt data-src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%201080p,%20x265.webp></div><div class="panzoom-container grid-w100"><img class="my-0 rounded-md zoomable" src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%204k,%20x265.webp alt data-src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%204k,%20x265.webp></div></div><figcaption class=caption>Various charts detailing the uplift from enabling AVX-512 on the 9950X on the x265 encoder, from 720p to 4k.</figcaption><hr><p>We see significant performance gains here thanks to AVX-512, on the slow-placebo presets, across every tested resolution. The x265 documentation has not commented on AVX-512 since the version 2.8 release in May 2018… Where it said, “For 4K main10 high-quality encoding, we are seeing good gains; for other resolutions and presets, we don’t recommend using this setting for now.”<sup id=fnref:3><a href=#fn:3 class=footnote-ref role=doc-noteref>3</a></sup> However, it seems that there are slight gains universally, increasing at 4k, but increasing greatly at every resolution as long as you use slow-placebo presets. I’d like to see the default behavior changed to enable AVX-512 by default, with a toggle to turn it off, in case you’re running mixed workloads on older Intel servers with less well-behaved AVX downclocking.</p><h3 class="relative group">9950X vs 9950X, SVT-AV1<div id=9950x-vs-9950x-svt-av1 class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#9950x-vs-9950x-svt-av1 aria-label=Anchor>#</a></span></h3><hr><div id=gallery-9d03451354d0cc21a56805dff1c5a0fd class=gallery><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%20720p,%20SVT-AV1.webp alt data-src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%20720p,%20SVT-AV1.webp></div><hr><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%201080p,%20SVT-AV1.webp alt data-src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%201080p,%20SVT-AV1.webp></div><hr><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%204k,%20SVT-AV1.webp alt data-src=images/compressed/9950X%20AVX-512%20Off%20vs%209950X%20AVX-512%20On,%204k,%20SVT-AV1.webp></div><hr></div><figcaption class=caption>Various charts detailing the uplift from enabling AVX-512 on the 9950X on the SVT-AV1 encoder, from 720p to 4k.</figcaption><hr><p>SVT-AV1 has AVX-512 enabled by default, and the documentation makes no special note of it. These gains aren’t all that great. I would have expected SVT-AV1 to have greater uplifts than x265, given it’s newer, and under more active development - but, that could also be exactly <em>why</em> the gains are fewer - functions are still being actively worked on, and have not been finalized in a way that makes anyone want to commit to writing a fully-optimized assembly version of them. It could also be possible that SVT-AV1 is already approaching memory starvation without AVX-512, and you need more bandwidth to get additional gains. Either way, I’m not particularly fond of AV1 in general, and I’m not interested in going down any rabbit holes related to this result, unless someone feels like donating a Sapphire Rapids or Genoa (or newer) server. Or a Zen 4 threadripper system.</p><h3 class="relative group">i7-14700F vs 9950X, x265<div id=i7-14700f-vs-9950x-x265 class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#i7-14700f-vs-9950x-x265 aria-label=Anchor>#</a></span></h3><hr><div id=gallery-60e6c86aaffb910f81de9fed8ad8714e class=gallery><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%20720p,%20x265.webp alt data-src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%20720p,%20x265.webp></div><hr><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%201080p,%20x265.webp alt data-src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%201080p,%20x265.webp></div><hr><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%204k,%20x265.webp alt data-src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%204k,%20x265.webp></div><hr></div><figcaption class=caption>Various charts detailing the uplift between the i7-14700F and the 9950X on the x265 encoder, from 720p to 4k.</figcaption><hr><h3 class="relative group">i7-14700F vs 9950X, SVT-AV1<div id=i7-14700f-vs-9950x-svt-av1 class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#i7-14700f-vs-9950x-svt-av1 aria-label=Anchor>#</a></span></h3><p>These are nice results. The AVX-512 gains from x265 really let the 9950X mog the poor i7-14700F (and by extension, i9-14900K, as the extra E-cores make almost zero difference). Keep in mind the i7-14700F is using ~260w throughout these tests, while the 9950X is capped to 200w. Pure performance aside - which is obviously significant - the efficiency improvement is also a huge win. Up to twice as fast, while using less power. What’s not to like?</p><hr><div id=gallery-733829b669da4639631f390972364c3f class=gallery><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%20720p,%20SVT-AV1.webp alt data-src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%20720p,%20SVT-AV1.webp></div><hr><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%201080p,%20SVT-AV1.webp alt data-src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%201080p,%20SVT-AV1.webp></div><hr><hr><div class=panzoom-container><img class="my-0 rounded-md zoomable" src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%204k,%20SVT-AV1.webp alt data-src=images/compressed/i7-14700F%20vs%209950X%20AVX-512%20On,%204k,%20SVT-AV1.webp></div><hr></div><figcaption class=caption>Various charts detailing the uplift between the i7-14700F and the 9950X on the SVT-AV1 encoder, from 720p to 4k.</figcaption><hr><p>Given the previous lack of significant uplift with SVT-AV1 due to AVX-512, it makes sense that these figures are much less impressive than x265. It’s still a nice uplift, to be sure, but not as outstanding as 2x!</p><h2 class="relative group">Overall Conclusion?<div id=overall-conclusion class=anchor></div><span class="absolute top-0 w-6 transition-opacity opacity-0 ltr:-left-6 rtl:-right-6 not-prose group-hover:opacity-100"><a class="group-hover:text-primary-300 dark:group-hover:text-neutral-700" style=text-decoration-line:none!important href=#overall-conclusion aria-label=Anchor>#</a></span></h2><p>Zen 5 carries a clear advantage over Raptor Lake, core for core. Gains attributed to the presence of AVX-512 can be great, or insignificant, depending on the encoder, the resolution, and the preset. For those interested in finding out more about the specific settings each preset contains that might be accelerated by AVX-512, you can find what features are enabled by specific presets in the <a href=https://x265.readthedocs.io/en/master/presets.html target=_blank>x265 documentation</a> and on the <a href=https://gitlab.com/AOMediaCodec/SVT-AV1/-/blob/master/Docs/CommonQuestions.md#what-presets-do target=_blank>SVT-AV1 GitLab.</a></p><p>Did AMD lie? Maybe. I did two additional synthetic tests, not graphed here. I can very, very closely emulate the 9700X test, by just limiting the 9950X to a single chiplet, and using the i7-14700F as-is. Under the most favorable circumstances, the speedup was only 1.2x. Then, I emulated the i5-14600K/9600X graph, by limiting i7-14700F to 6P+8E, as well as limiting the 9950X to 6 cores on one chiplet. In that scenario, the speedup was 1.24x. A far cry from 41-94%. However, there are a number of differences between my setup and AMD’s, apart from the inexact hardware.</p><ul><li>AMD tested on Windows. I tested on Linux. Windows’ scheduler is famously terrible with assymetrical architectures. Could that have nerfed Intel enough to make up the difference?</li><li>I only tested SVT-AV1 and x265. HandBrake also offers x264 for AVC, and libvpx for VP9. They could have tested with those, and gotten more disparate results.</li><li>I’m using more recent versions of the encoders than HandBrake would have shipped when those graphs were made. There could be other improvements in the interim that have closed the gap.</li></ul><p>I was not intending to do a strict DEBOONKING or affirmation of AMD’s graphs in any case. They’re just here as comparison, and they prompted my interest in examining AVX-512’s presence in video encoders. I do not care enough to investigate the differences in encoders that I’m not interested in, so this story ends here, for now, at least.</p><div class=footnotes role=doc-endnotes><hr><ol><li id=fn:1><p>Zen 5 mobile and Zen 5c continue to use double-pumped/otherwise hybrid AVX-512 implementations. These claims are only accurate to the Granite Ridge chiplets used in desktop SKUs and some server SKUs. Presumably upcoming HX-type mobile SKUs as well, since they’re just unpackaged desktop SKUs. <a href=#fnref:1 class=footnote-backref role=doc-backlink>↩︎</a></p></li><li id=fn:2><p>512 / 32 * 4 * 16 <a href=#fnref:2 class=footnote-backref role=doc-backlink>↩︎</a></p></li><li id=fn:3><p><a href=https://x265.readthedocs.io/en/master/releasenotes.html#version-2-8 target=_blank>https://x265.readthedocs.io/en/master/releasenotes.html#version-2-8</a> <a href=#fnref:3 class=footnote-backref role=doc-backlink>↩︎</a></p></li></ol></div><strong class="block mt-8"><a target=_blank class="m-1 rounded bg-neutral-300 p-1.5 text-neutral-700 hover:bg-primary-500 hover:text-neutral dark:bg-neutral-700 dark:text-neutral-300 dark:hover:bg-primary-400 dark:hover:text-neutral-800" href="mailto:rawhide@neet.works?subject=Reply%20to%20Benchmarking%20AVX-512%20Video%20Encoding%20On%20Zen%205">Reply by Email</a></strong></div><h2 class="mt-8 text-2xl font-extrabold mb-10">Related</h2><section class="w-full grid gap-4 sm:grid-cols-2 md:grid-cols-3"><a href=/posts/watercooling-homelab/ class=min-w-full><div class="min-h-full border border-neutral-200 dark:border-neutral-700 border-2 rounded overflow-hidden shadow-2xl relative"><div class="w-full thumbnail_card_related nozoom" style=background-image:url(/posts/watercooling-homelab/featured_hu_9bec814a39f70286.webp)></div><div class="px-6 py-4"><div class="font-bold text-xl text-neutral-800 decoration-primary-500 hover:underline hover:underline-offset-2 dark:text-neutral" href=/posts/watercooling-homelab/>Watercooling My Homelab</div><div class="text-sm text-neutral-500 dark:text-neutral-400"><div class="flex flex-row flex-wrap items-center"><time datetime=2025-01-16T05:56:41+00:00>January 16 2025</time><span class="px-2 text-primary-500">·</span><span>5240 words</span></div><hr><div class="flex flex-row flex-wrap items-center">Watercooling for my homelab with a custom, leak-resistant controller and monitoring!</div><hr><div class="flex flex-row flex-wrap items-center"><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/categories/homelab/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Homelab
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/categories/servers/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Servers
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/categories/watercooling/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Watercooling
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/categories/diy/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Diy
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/alphacool/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Alphacool
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/arduino/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Arduino
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/intel/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Intel
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/nvidia/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Nvidia
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/supermicro/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Supermicro
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/aqua-computer/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Aqua Computer</span></span></span></div></div></div><div class="px-6 pt-4 pb-2"></div></div></a><a href=/posts/maxcloudon-pcie-bifurcation-riser-review/ class=min-w-full><div class="min-h-full border border-neutral-200 dark:border-neutral-700 border-2 rounded overflow-hidden shadow-2xl relative"><div class="w-full thumbnail_card_related nozoom" style=background-image:url(/posts/maxcloudon-pcie-bifurcation-riser-review/featured_hu_381f1d47da1b5abe.webp)></div><div class="px-6 py-4"><div class="font-bold text-xl text-neutral-800 decoration-primary-500 hover:underline hover:underline-offset-2 dark:text-neutral" href=/posts/maxcloudon-pcie-bifurcation-riser-review/>MaxCloudON PCIe Bifurcation Riser Review</div><div class="text-sm text-neutral-500 dark:text-neutral-400"><div class="flex flex-row flex-wrap items-center"><time datetime=2025-01-19T07:32:31+00:00>January 19 2025</time><span class="px-2 text-primary-500">·</span><span>1936 words</span></div><hr><div class="flex flex-row flex-wrap items-center">My experience with the MaxCloudON PCIe Bifurcation Riser and a look at some alternatives.</div><hr><div class="flex flex-row flex-wrap items-center"><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/categories/homelab/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Homelab
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/categories/servers/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Servers
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/categories/reviews/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Reviews
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/nvidia/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Nvidia
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/supermicro/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Supermicro
|
||
</span></span></span><span style=margin-top:.5rem class=mr-2 onclick='return window.open("/tags/pcie-bifurcation/","_self"),!1'><span class=flex style=cursor:pointer><span class="rounded-md border border-primary-400 px-1 py-[1px] text-xs font-normal">Pcie Bifurcation</span></span></span></div></div></div><div class="px-6 pt-4 pb-2"></div></div></a></section></div><script>var oid="views_posts/benchmarking-avx-512-video-encoding-zen-5/index.md",oid_likes="likes_posts/benchmarking-avx-512-video-encoding-zen-5/index.md"</script><script type=text/javascript src=/js/page.min.0860cf4e04fa2d72cc33ddba263083464d48f67de06114529043cb4623319efed4f484fd7f1730df5abea0e2da6f3538855634081d02f2d6e920b956f063e823.js integrity="sha512-CGDPTgT6LXLMM926JjCDRk1I9n3gYRRSkEPLRiMxnv7U9IT9fxcw31q+oOLabzU4hVY0CB0C8tbpILlW8GPoIw=="></script></section><footer class="pt-8 max-w-prose print:hidden"><div class=pt-8><hr class="border-dotted border-neutral-300 dark:border-neutral-600"><div class="flex justify-between pt-3"><span><a class="flex group mr-3" href=/posts/maxcloudon-pcie-bifurcation-riser-review/><span class="mr-3 text-neutral-700 group-hover:text-primary-600 ltr:inline rtl:hidden dark:text-neutral dark:group-hover:text-primary-400">←</span>
|
||
<span class="ml-3 text-neutral-700 group-hover:text-primary-600 ltr:hidden rtl:inline dark:text-neutral dark:group-hover:text-primary-400">→</span>
|
||
<span class="flex flex-col"><span class="mt-[0.1rem] leading-6 group-hover:underline group-hover:decoration-primary-500">MaxCloudON PCIe Bifurcation Riser Review</span>
|
||
<span class="mt-[0.1rem] text-xs text-neutral-500 dark:text-neutral-400"><time datetime=2025-01-19T07:32:31+00:00>January 19 2025</time>
|
||
</span></span></a></span><span></span></div></div></footer></article><div id=top-scroller class="pointer-events-none absolute top-[110vh] bottom-0 w-12 ltr:right-0 rtl:left-0"><a href=#the-top class="pointer-events-auto sticky top-[calc(100vh-5.5rem)] flex h-12 w-12 mb-16 items-center justify-center rounded-full bg-neutral/50 text-xl text-neutral-700 hover:text-primary-600 dark:bg-neutral-800/50 dark:text-neutral dark:hover:text-primary-400" aria-label="Scroll to top" title="Scroll to top">↑</a></div></main><footer id=site-footer class="py-10 print:hidden"><div class="flex items-center justify-between"><p class="text-sm text-neutral-500 dark:text-neutral-400"><em>© 2025 rawhide kobayashi | <strong>Original content</strong> licensed under <strong><a href=https://creativecommons.org/licenses/by-sa/4.0/ target=_blank>CC BY-SA 4.0</a></strong> unless otherwise noted.</em></p><p class="text-xs text-neutral-500 dark:text-neutral-400">Powered by <a class="hover:underline hover:decoration-primary-400 hover:text-primary-500" href=https://gohugo.io/ target=_blank rel="noopener noreferrer">Hugo</a> & <a class="hover:underline hover:decoration-primary-400 hover:text-primary-500" href=https://blowfish.page/ target=_blank rel="noopener noreferrer">Blowfish</a></p></div><script></script><script type=text/javascript src=/js/process.min.ee03488f19c93c2efb199e2e3014ea5f3cb2ce7d45154adb3399a158cac27ca52831db249ede5bb602700ef87eb02434139de0858af1818ab0fb4182472204a4.js integrity="sha512-7gNIjxnJPC77GZ4uMBTqXzyyzn1FFUrbM5mhWMrCfKUoMdsknt5btgJwDvh+sCQ0E53ghYrxgYqw+0GCRyIEpA=="></script></footer><div id=search-wrapper class="invisible fixed inset-0 flex h-screen w-screen cursor-default flex-col bg-neutral-500/50 p-4 backdrop-blur-sm dark:bg-neutral-900/50 sm:p-6 md:p-[10vh] lg:p-[12vh]" data-url=https://blog.neet.works/ style=z-index:500><div id=search-modal class="flex flex-col w-full max-w-3xl min-h-0 mx-auto border rounded-md shadow-lg top-20 border-neutral-200 bg-neutral dark:border-neutral-700 dark:bg-neutral-800"><header class="relative z-10 flex items-center justify-between flex-none px-2"><form class="flex items-center flex-auto min-w-0"><div class="flex items-center justify-center w-8 h-8 text-neutral-400"><span class="relative block icon"><svg aria-hidden="true" focusable="false" data-prefix="fas" data-icon="search" class="svg-inline--fa fa-search fa-w-16" role="img" viewBox="0 0 512 512"><path fill="currentcolor" d="M505 442.7 405.3 343c-4.5-4.5-10.6-7-17-7H372c27.6-35.3 44-79.7 44-128C416 93.1 322.9.0 208 0S0 93.1.0 208s93.1 208 208 208c48.3.0 92.7-16.4 128-44v16.3c0 6.4 2.5 12.5 7 17l99.7 99.7c9.4 9.4 24.6 9.4 33.9.0l28.3-28.3c9.4-9.4 9.4-24.6.1-34zM208 336c-70.7.0-128-57.2-128-128 0-70.7 57.2-128 128-128 70.7.0 128 57.2 128 128 0 70.7-57.2 128-128 128z"/></svg></span></div><input type=search id=search-query class="flex flex-auto h-12 mx-1 bg-transparent appearance-none focus:outline-dotted focus:outline-2 focus:outline-transparent" placeholder=Search tabindex=0></form><button id=close-search-button class="flex items-center justify-center w-8 h-8 text-neutral-700 hover:text-primary-600 dark:text-neutral dark:hover:text-primary-400" title="Close (Esc)">
|
||
<span class="relative block icon"><svg viewBox="0 0 320 512"><path fill="currentcolor" d="M310.6 361.4c12.5 12.5 12.5 32.75.0 45.25C304.4 412.9 296.2 416 288 416s-16.38-3.125-22.62-9.375L160 301.3 54.63 406.6C48.38 412.9 40.19 416 32 416S15.63 412.9 9.375 406.6c-12.5-12.5-12.5-32.75.0-45.25l105.4-105.4L9.375 150.6c-12.5-12.5-12.5-32.75.0-45.25s32.75-12.5 45.25.0L160 210.8l105.4-105.4c12.5-12.5 32.75-12.5 45.25.0s12.5 32.75.0 45.25l-105.4 105.4L310.6 361.4z"/></svg></span></button></header><section class="flex-auto px-2 overflow-auto"><ul id=search-results></ul></section></div></div></div></body></html> |