--- title: NVIDIA Posts Linux Scheduler Patches To Further Boost... description: NVIDIA engineer Andrea Righi submitted a Linux kernel scheduler patch series to the LKML that adds preferred SMT sibling support for NVIDIA's upcoming Vera... canonical: https://daily.dev/posts/nvidia-posts-linux-scheduler-patches-to-further-boost-smt-performance-on-nvidia-vera-asvofq3qw twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: NVIDIA Posts Linux Scheduler Patches To Further Boost SMT Performance On NVIDIA Vera | daily.dev og:description: NVIDIA engineer Andrea Righi submitted a Linux kernel scheduler patch series to the LKML that adds preferred SMT sibling support for NVIDIA's upcoming Vera... og:url: https://daily.dev/posts/nvidia-posts-linux-scheduler-patches-to-further-boost-smt-performance-on-nvidia-vera-asvofq3qw og:image: https://api.daily.dev/og/posts/aSVOFq3qW.png og:image:alt: NVIDIA Posts Linux Scheduler Patches To Further Boost SMT Performance On NVIDIA Vera og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # NVIDIA Posts Linux Scheduler Patches To Further Boost SMT Performance On NVIDIA Vera **[Phoronix](https://daily.dev/sources/phoronix)** · 3 min read · 0 upvotes · 0 comments ## Summary NVIDIA engineer Andrea Righi submitted a Linux kernel scheduler patch series to the LKML that adds preferred SMT sibling support for NVIDIA's upcoming Vera platform with Olympus cores. Olympus SMT implements two processing elements (PEs) per core, and Olympus is unusually sensitive to brief sibling activation since returning to single-thread performance is not immediate. The new patches make PE0 the preferred sibling using SD_ASYM_PACKING and teach the fair scheduler's idle-selection paths to respect asymmetric SMT priority. Testing on a two-socket Vera system with an 88-thread GEMM workload showed throughput improve from about 9.4 TFLOP/s to about 10.1 TFLOP/s, with more consistent results as the workload settled onto PE0. This builds on an earlier merged change (for a prior kernel release) that already boosted the same benchmark from 6.2 to 9.2 TFLOP/s by preferring fully idle cores for NOHZ balancing. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What performance improvement do NVIDIA's new Linux scheduler patches provide for Vera SMT cores? NVIDIA's patch series for preferred SMT siblings on Vera Olympus cores improved GEMM throughput from approximately 9.4 TFLOP/s to approximately 10.1 TFLOP/s in testing on a two-socket, 88-physical-core Vera system. The patches use SD_ASYM_PACKING to make PE0 the preferred sibling and teach the fair scheduler's idle-selection paths to respect asymmetric SMT priority, making repeated runs more predictable as workloads consistently settled on PE0. _daily.dev surfaces kernel scheduler patch discussions for engineers tracking SMT performance work._ ### Why is NVIDIA Olympus SMT particularly sensitive to brief sibling wakeups? NVIDIA Olympus implements SMT with two symmetric processing elements per core, and when a sibling briefly activates and then goes idle again, full single-thread performance is not restored immediately. After the idle load balancer finishes and the CPU enters WFI, single-thread performance returns only after the sibling has stayed idle for a qualification interval measured at roughly 10 Ki cycles on tested Vera hardware. _Engineers debugging SMT-related performance quirks follow low-level scheduler discussions like this on daily.dev._ ## Similar posts on daily.dev - [Linux Patches Posted To Fix ~2x Performance Drop For CPU Workloads On NVIDIA Vera Rubin](https://daily.dev/posts/linux-patches-posted-to-fix-2x-performance-drop-for-cpu-workloads-on-nvidia-vera-rubin-kdfueg0xp) · Phoronix · 0 upvotes · 0 comments - [LLVM 22 Lands NVIDIA Olympus CPU Scheduling Model](https://daily.dev/posts/llvm-22-lands-nvidia-olympus-cpu-scheduling-model-vcbkejch6) · Phoronix · 0 upvotes · 0 comments - [Linux's sched\_ext Will Prioritize Idle SMT Siblings For Better Performance](https://daily.dev/posts/linux-s-sched-ext-will-prioritize-idle-smt-siblings-for-better-performance-tiery1ycf) · Phoronix · 0 upvotes · 0 comments --- Tags: [#linux](https://daily.dev/tags/linux), [#performance](https://daily.dev/tags/performance), [#nvidia](https://daily.dev/tags/nvidia) [View this post on daily.dev](https://daily.dev/posts/nvidia-posts-linux-scheduler-patches-to-further-boost-smt-performance-on-nvidia-vera-asvofq3qw) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"NVIDIA Posts Linux Scheduler Patches To Further Boost SMT Performance On NVIDIA Vera","url":"https://daily.dev/posts/nvidia-posts-linux-scheduler-patches-to-further-boost-smt-performance-on-nvidia-vera-asvofq3qw","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/nvidia-posts-linux-scheduler-patches-to-further-boost-smt-performance-on-nvidia-vera-asvofq3qw"},"datePublished":"2026-09-01T00:35:15.893Z","dateModified":"2026-09-04T22:30:40.894Z","description":"NVIDIA engineer Andrea Righi submitted a Linux kernel scheduler patch series to the LKML that adds preferred SMT sibling support for NVIDIA's upcoming Vera...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/2fd07e184e3e7f000817157fef661582?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/2fd07e184e3e7f000817157fef661582?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Phoronix","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Phoronix","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/6a26ba379dd64932830aa0c9803dd7ee","url":"https://daily.dev/sources/phoronix"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/nvidia-posts-linux-scheduler-patches-to-further-boost-smt-performance-on-nvidia-vera-asvofq3qw","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"linux,performance,nvidia","timeRequired":"PT3M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Phoronix","item":"https://daily.dev/sources/phoronix"},{"@type":"ListItem","position":3,"name":"NVIDIA Posts Linux Scheduler Patches To Further Boost SMT Performance On NVIDIA Vera"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/nvidia-posts-linux-scheduler-patches-to-further-boost-smt-performance-on-nvidia-vera-asvofq3qw#faq","mainEntity":[{"@type":"Question","name":"What performance improvement do NVIDIA's new Linux scheduler patches provide for Vera SMT cores?","acceptedAnswer":{"@type":"Answer","text":"NVIDIA's patch series for preferred SMT siblings on Vera Olympus cores improved GEMM throughput from approximately 9.4 TFLOP/s to approximately 10.1 TFLOP/s in testing on a two-socket, 88-physical-core Vera system. The patches use SD_ASYM_PACKING to make PE0 the preferred sibling and teach the fair scheduler's idle-selection paths to respect asymmetric SMT priority, making repeated runs more predictable as workloads consistently settled on PE0. daily.dev surfaces kernel scheduler patch discussions for engineers tracking SMT performance work."}},{"@type":"Question","name":"Why is NVIDIA Olympus SMT particularly sensitive to brief sibling wakeups?","acceptedAnswer":{"@type":"Answer","text":"NVIDIA Olympus implements SMT with two symmetric processing elements per core, and when a sibling briefly activates and then goes idle again, full single-thread performance is not restored immediately. After the idle load balancer finishes and the CPU enters WFI, single-thread performance returns only after the sibling has stayed idle for a qualification interval measured at roughly 10 Ki cycles on tested Vera hardware. Engineers debugging SMT-related performance quirks follow low-level scheduler discussions like this on daily.dev."}}]} ```