# cpus (blogs) — RSS Amplifier Recent posts from the 5 feeds in the RSS Amplifier directory that cover cpus. Page: Feed: --- ## [Core Ultra 9 4950K―“Nova Lake”のナンバーは4桁になる？](https://northwood.blog.fc2.com/blog-entry-13185.html) _2026-08-20 · 北森八雲 · 北森瓦版_ Intel Core Ultra 9 4950K Could Signal New Naming Scheme for "Nova Lake-S"（TechPowerUp）Intel testing Core Ultra 9 4950K desktop CPU, pointing to possible new Nova Lake-S naming（VideoCardz）午前11:37 · 2026年8月19日（LC Tech Leaks (and News)@laurentschoice）新しくリークした情報によると、Intelが現在“Nova Lake-S”をテストしているが、その“Nova Lake-S”はCore Ultra 9 4950Kと認識されているようである。こ ## [“Razor Lake”はTSMC N2Xで製造されるらしい](https://northwood.blog.fc2.com/blog-entry-13184.html) _2026-08-18 · 北森八雲 · 北森瓦版_ Intel Razor Lake CPUs to Utilize TSMC N2X Process & Bring bLLC to Laptops（WCCF Tech）午後7:11 · 2026年8月18日（X86 is dead&back@x86deadandback）“Nova Lake”に続く“Razor Lake”はTSMC N2Xで製造される。また、Mobile向けにbLLCがもたらされる。今度登場する“Nova Lake”はTSMC N2PとIntel 18Aで製造される。そしてその次となる“Razor Lake”は同じTSMC 2nm世代のプロセスでありながらも、N2Pよりもさらに進んだN2Xで製造さ ## [“Noctua”は我が道を行く](https://northwood.blog.fc2.com/blog-entry-13183.html) _2026-08-18 · 北森八雲 · 北森瓦版_ Noctua has no plans for RGB, screens or white coolers, engineering resources go elsewhere（VideoCardz）Noctua Remains Focused on Performance, No Plans for RGB or Screens in Coolers（TechPowerUp）Noctuaは昨今のPC冷却のトレンドに追随する気はないようだ。NoctuaはRGB発光や、内蔵ディスプレイ、あるいは白色の派生製品などを製品とするつもりはないようである。そして同社はその開発・技術リソースを冷却性能と静音 ## [AIDA 64に“Zen 7”の初期サポートが追加される](https://northwood.blog.fc2.com/blog-entry-13182.html) _2026-08-17 · 北森八雲 · 北森瓦版_ AIDA 64 gains “Zen 7” CPU support years ahead of launch（OC3D）AMD Zen 7 server processors already get preliminary AIDA64 support, two years before launch（VideoCardz）Download AIDA64 Extreme (beta)（AIDA 64）8月16日付けで更新されたAIDA 64のVersion 8.35.8405 betaに“Zen 7”の初期サポートが追加された。当該ページを見ると“Preliminary support for AMD Zen 7 server processors”の記述を確認できる。これは“Ze ## [“Raptor Lake”は今後数年、中核的な製品群であり続ける](https://northwood.blog.fc2.com/blog-entry-13181.html) _2026-08-17 · 北森八雲 · 北森瓦版_ Older Raptor Lake CPUs are a ‘core part of the portfolio’ for years to come, says Intel — there’s been a ‘sudden inrush of demand’ for LGA 1700 chips due to DDR5 prices（Tom's Hardware）Intel Keeps Raptor Lake in Its Lineup Longer: DDR5 Prices Drive Demand for LGA1700 and DDR4（Igor's Labs）Robert Hallock氏はTom's Hardwareのインタビューで、“Raptor Lake”について「来る今後数年間」Intelが供給し続ける ## [AVX-512が有効化された“Nova Lake-S”がテストされている模様](https://northwood.blog.fc2.com/blog-entry-13180.html) _2026-08-17 · 北森八雲 · 北森瓦版_ Intel tests 24-core and 28-core Nova Lake-S desktop CPUs with AVX-512 enabled（VideoCardz）午前0:55 · 2026年8月17日（InstLatX64@InstLatX64）Intelは“Nova Lake-S”を2種類―28-coreと24-coreのもののテストを行っている。そしてこれらのPublic graphics CI logを見ると、どちらのシステムもLinuxで動作し、かつAVX-512がサポートされている状態である。今回の情報をもたらしたのはInstLatX64氏で、8月16日付けのIntel GFX C ## [“Nova Lake”はサーバー向けのXeonよりデスクトップ向けが先行する](https://northwood.blog.fc2.com/blog-entry-13179.html) _2026-08-17 · 北森八雲 · 北森瓦版_ Intel's New CPU Cores Debut on "Nova Lake-S" Desktop Before Xeon Servers（TechPowerUp）Intel says it will launch new core with Nova Lake on desktop first, not in data center — VP Robert Hallock hopes enthusiasts ‘do the math’ compared to AMD（Tom's Hardware）Intel confirms its “new core” is coming to desktops first, not Xeon（OC3D）Intelは“Nova Lake”とともに新しいコアをデータセンターやサーバー向け ## [GigabyteからB450チップセットを搭載したマザーボードの新製品が登場](https://northwood.blog.fc2.com/blog-entry-13178.html) _2026-08-15 · 北森八雲 · 北森瓦版_ Gigabyte Launches Two New B450 Motherboards for Veteran AMD Socket AM4（TechPowerUp）B450M D3HP WIFI6E（Gigabyte）B450M D3HP（Gigabyte）GigabyteはAMD B450チップセット搭載したMicroATX SocketAM4マザーボードの新製品を2種類ローンチした。AMD B450 chipsetは2018年にローンチされた8年前のチップセットである。そして今回登場したB450M D3HP WIFI6E, B450M D3HPはSocketAM4プラットフォームに対応する低価格向けの製品 ## [“Razor Lake-AX”のCPUコアは“Nova Lake”のものを使用する](https://northwood.blog.fc2.com/blog-entry-13177.html) _2026-08-15 · 北森八雲 · 北森瓦版_ Intel Razor Lake-AX reportedly uses Nova Lake CPU cores, won’t be unlocked for overclocking（VideoCardz）大規模なiGPUを搭載するとされる“Razor Lake-AX”であるがオーバークロック対応のための倍率ロック解除仕様にはならないと伝えられている。さらに、“Razor Lake-AX”で使用されるCPUコアアーキテクチャは“Nova Lake”のCPUコアアーキテクチャであるという。“Nova Lake”はP-coreに“Coyote Cove”、E-coreに“Arctic Wolf”を ## [“Emerald Rapids”のXCCのダイ写真](https://northwood.blog.fc2.com/blog-entry-13176.html) _2026-08-15 · 北森八雲 · 北森瓦版_ 午後9:08 · 2026年8月13日（HiEQ@HiEq\_2005）第5世代のXeon Scalable Processorである“Emerald Rapids”のXCC (Extreme Core Count) のダイ写真が掲載されている。“Emerald Rapids”はIntel 7で製造され、ベースとなるコアは“Raptor Cove”である。その最上位ダイであるXCCは2ダイ構成をとっており、製品としては最大64-coreがラインナップされている。 ## [ThreatCrush on GitHub (Sponsored)](https://crawlproof.com/a/QN2QmQnQ5sgN) _2026-08-15 · **Sponsored**_ View the ThreatCrush listing on the GitHub Marketplace. ## [Intelがメモリ事業に再参入するかもしれない？？](https://northwood.blog.fc2.com/blog-entry-13175.html) _2026-08-13 · 北森八雲 · 北森瓦版_ Intel CEO Hints Chipzilla’s Return To Making Memory, As He Teases “New Memory Architecture” & 3D Stacking To Tackle The DRAM Wall（WCCF Tech）Intelがメモリ事業の再参入についてヒントを明かした。CEOであるLip-Bu Tan氏がメモリはもはやコモディティな事業ではなく、途方もない革新性を秘めたものだと述べた。Intelが半導体事業を行う会社として設立されたのは1968年で、最初期の製品はメモリチップであり3101 SRAMと11 ## [Spectre on RISC-V Silicon: Attacks and Defenses on Commercial Out-of-Order Processors](https://lukasgerlach.me/publication/2026-speculative-execution-attacks-on-risc-v-silicon/) _2026-08-12 · Lukas Gerlach_ ## [Spectre on RISC-V Silicon](https://lukasgerlach.me/blog/2026-spectre-on-riscv-silicon-usenix/) _2026-07-03 · Lukas Gerlach_ Our paper Spectre on RISC-V Silicon: Attacks and Defenses on Commercial Out-of-Order Processors has been accepted at USENIX Security ‘26 . Paper / Artifact This post is the short version of what we found. RISC-V is an open, modular instruction set architecture that anyone can implement without a license fee. Many independent vendors build their own cores from it, ranging from tiny in-order… ## [Valve Steam Machine vs DIY Plasma PC™ - The Zen 5 (9600X) and RDNA 4 (9060 XT 16 GB) Alternative](https://nietras.com/2026/06/28/steam-machine-vs-plasma-pc/) _2026-06-28 · nietras_ Valve’s new 2026 Steam Machine has created a lot of discussion online given it’s pricing. In this quick blog post I will look at specs of the Steam Machine and try create a barebones equivalent or better DIY gaming PC called Plasma PC™ (obviously based on solid, liquid, gas, and plasma). ## [Previewing SpecCat.com - Compare CPU, APU, GPU Specifications Instantly](https://nietras.com/2026/06/27/previewing-speccat/) _2026-06-27 · nietras_ SpecCat.com is a web site I’ve made, mainly for myself trying out LLMs, for easily comparing detailed specifications of different silicon chips; CPUs, APUS, GPUs etc. as shown in screenshots of header below and full at bottom. It should work reasonably well on mobile, but main use case was for larger screens. I tried hard to make it colorful and fun, and dark only. I made this a while ago but… ## [How do Cache Attacks Against AES Work?](https://lukasgerlach.me/blog/aes-lookup-table-attacks/) _2026-05-26 · Lukas Gerlach_ AES is the most widely used block cipher, and there are some cache side-channel attacks against it. The attacks always exploit key-dependent memory accesses because parts of AES are optimized with lookup tables. This post quickly goes over the lookup table optimizations called S-boxes and T-tables, and how such implementations can be attacked. My hope is that this can help quickly understand the… ## [Crucible: Retrofitting Commodity CPUs with Vulnerabilities via Transparent Software Emulation](https://lukasgerlach.me/publication/2026-crucible-retrofitting-commodity-cpus/) _2026-05-18 · Lukas Gerlach_ ## [TDXRay: Microarchitectural Side-Channel Analysis of Intel TDX for Real-World Workloads](https://lukasgerlach.me/publication/2026-tdxray-microarchitectural-side-channel-analysis/) _2026-05-18 · Lukas Gerlach_ ## [NumKong: 2'000 Mixed Precision Kernels For All 🦍](https://ashvardanian.com/posts/numkong/) _2026-03-20 · Ash's Blog_ Around 2'000 SIMD kernels for mixed-precision BLAS-like numerics — dot products, batched GEMMs, distances, geospatial, ColBERT MaxSim, and mesh alignment — from Float6 to Float118, leveraging RISC-V, Intel AMX, Arm SME, and WebAssembly Relaxed SIMD, in 7 languages and 5 MB. ## [Zero-Store Elimination and its Implications on the SIKE Cryptosystem](https://lukasgerlach.me/publication/2026-zero-store-elimination-and-its-implications/) _2026-02-03 · Lukas Gerlach_ ## [Ask the crowd: AI or not (Sponsored)](https://crawlproof.com/a/nu7B4Yw8WS2O) _2026-02-02 · **Sponsored**_ Get a transparent verdict from verified humans via web, RSS, or API. ## [Full Unicode Search at 50× ICU Speed with AVX‑512](https://ashvardanian.com/posts/search-utf8/) _2025-12-15 · Ash's Blog_ ICU gets Unicode right and pays for it. This post shows a different approach: fold-safe windows, SIMD probes, and verifiers for fast UTF‑8 search. ## [Sep 0.12.0 - Add net10.0 target and use cross-platform VectorXY.NarrowWithSaturation](https://nietras.com/2025/11/12/sep-0-12-0/) _2025-11-12 · nietras_ Sep 0.12.0 was released November 12th, 2025 with a minor update to target net10.0 and use the newly available VectorXY.NarrowWithSaturation. ## [.NET and C# Versions - .NET Framework 1.0 to .NET 10 & C# 1.0 to 14 (10/14 Update)](https://nietras.com/2025/11/11/dotnet-and-csharp-versions/) _2025-11-11 · nietras_ Above png is very high resolution, but you can also download a pdf . Sources .NET and C# Versions - 20th Anniversary ♥ .NET and C# Versions - 7/11 Update .NET and C# Versions - 8/12 Update .NET and C# Versions - 9/13 Update What’s new in .NET 10 What’s new in the .NET 10 runtime What’s new in .NET libraries for .NET 10 What’s new in the SDK and tooling for .NET 10 What’s new in C# 14 Breaking… ## [Tuning TLS: AES-256 Beats ChaCha20 on Every CPU](https://ashvardanian.com/posts/chacha-vs-aes-2025/) _2025-11-07 · Ash's Blog_ AES-256-GCM now beats ChaCha20-Poly1305 by up to 3x on every modern CPU with hardware acceleration, reversing the 2015 mobile performance advice. ## [Scaling Elections with GPUs and Mojo across Nvidia and AMD 🔥](https://ashvardanian.com/posts/scaling-elections/) _2025-10-28 · Ash's Blog_ Taking the computationally expensive Schulze voting method from theory to practice on GPUs — exploring parallel algorithms, hardware optimizations, and why Mojo surprised me. ## [RISCover: Automatic Discovery of User-exploitable Architectural Security Vulnerabilities in Closed-Source RISC-V CPUs](https://lukasgerlach.me/publication/2025-riscover-automatic-discovery-of-user-exploitable/) _2025-10-13 · Lukas Gerlach_ ## [2x Faster Hashes on AWS Graviton: NEON → SVE2](https://ashvardanian.com/posts/aws-graviton-checksums-on-neon-vs-sve/) _2025-10-06 · Ash's Blog_ SVE2 AES on AWS Graviton 4 delivers 2x faster string hashing than NEON, but SVE's variable-length promise fell flat: Graviton 4 regressed to 128-bit vectors. ## [Before AI's Kepler Moment - Are LLMs the Epicycles of Intelligence?](https://ashvardanian.com/posts/llm-epicycles/) _2025-10-02 · Ash's Blog_ Like Ptolemaic astronomers stacking circles to predict planetary motion, we stack transformer layers to approximate intelligence—accurate yet awaiting our Kepler moment. ## [How a String Library Beat OpenCV at Image Processing by 4x](https://ashvardanian.com/posts/image-processing-with-strings/) _2025-09-20 · Ash's Blog_ StringZilla's SIMD-optimized Look-Up Tables beat OpenCV by 4x on image processing, proving string manipulation techniques excel beyond text. ## [Processing Strings 109x Faster than Nvidia on H100](https://ashvardanian.com/posts/stringwars-on-gpus/) _2025-09-15 · Ash's Blog_ StringZilla v4 brings CUDA acceleration for string processing: 109x faster than Nvidia's CuDF on edit distances, plus 52-bit MinHash fingerprinting and AES-based hashing. ## [Hire AI-Powered Freelancers Fast (Sponsored)](https://crawlproof.com/a/Ns2sloUiSN2t) _2025-09-14 · **Sponsored**_ Browse and hire skilled professionals using AI tools — free to post, free to apply. ## [Confusing Value with Enumeration: Studying the Use of CVEs in Academia](https://lukasgerlach.me/publication/2025-confusing-value-with-enumeration-studying-the/) _2025-08-13 · Lukas Gerlach_ ## [SCASE: Automated Secret Recovery via Side-Channel-Assisted Symbolic Execution](https://lukasgerlach.me/publication/2025-scase-automated-secret-recovery-via-side/) _2025-08-13 · Lukas Gerlach_ ## [Taming the Linux Memory Allocator for Rapid Prototyping](https://lukasgerlach.me/publication/2025-taming-the-linux-memory-allocator-for/) _2025-07-09 · Lukas Gerlach_ ## [Sep 0.11.0 - 9.5 GB/s CSV Parsing Using ARM NEON SIMD on Apple M1 🚀](https://nietras.com/2025/06/17/sep-0-11-0/) _2025-06-17 · nietras_ Sep 0.11.0 was released June 12th, 2025 with a new parser optimized specifically for ARM NEON SIMD (called AdvSimd in .NET) capable ARM64 CPUs like the Apple M1) or the new Microsoft cloud Cobalt 100, which is based on the ARM Neoverse N2. ## [How to Get Windows 8.3 Short File Names Using FindFirstFileW (UNC) and GetShortPathName (local) in C#](https://nietras.com/2025/05/26/windows-short-path-names/) _2025-05-26 · nietras_ Working with files and directories on Windows, especially on network shares, often leads to issues with long paths. This post explains what Windows 8.3 short paths are, how they work in NTFS, the difference between local and UNC paths, why browsers and some tools fail with long network paths, and how to programmatically obtain short file names for both local and UNC files. ## [Beyond OpenMP in C++ & Rust: Taskflow, Rayon, Fork Union 🍴](https://ashvardanian.com/posts/beyond-openmp-in-cpp-rust/) _2025-05-20 · Ash's Blog_ Most C and Rust thread pools run 10x slower than OpenMP on fork-join workloads. Fork Union closes the gap to 20% with 300 lines, dodging mutexes and CAS. ## [Backup (NuGet) Packages for All Azure DevOps Organization Feeds with PowerShell](https://nietras.com/2025/05/15/azure-devops-backup-all-packages/) _2025-05-15 · nietras_ This quick blog post demonstrates how to automate the backup of all packages from all organization feeds on Azure DevOps using a scheduled pipeline and a PowerShell script. The script leverages the Azure DevOps REST API to enumerate feeds and download every package version to a network share. ## [Rapid Reversing of Non-Linear CPU Cache Slice Functions: Unlocking Physical Address Leakage](https://lukasgerlach.me/publication/2025-rapid-reversing-of-non-linear-cpu/) _2025-05-12 · Lukas Gerlach_ ## [Sep 0.10.0 - 21 GB/s CSV Parsing Using SIMD on AMD 9950X 🚀](https://nietras.com/2025/05/09/sep-0-10-0/) _2025-05-09 · nietras_ Sep 0.10.0 was released April 22nd, 2025 with optimizations for AVX-512 capable CPUs like the AMD 9950X (Zen 5) and updated benchmarks including the 9950X. Sep now achieves a staggering 21 GB/s on the 9950X for the low-level CSV parsing. 🚀 Before 0.10.0, Sep achieved ~18 GB/s on 9950X. ## [Sep 0.9.0 - Async Support](https://nietras.com/2025/05/08/sep-0-9-0/) _2025-05-08 · nietras_ Sep 0.9.0 was released February 1st, 2025 - earlier this year - with a major new feature: Async support for both SepReader and SepWriter. ## [Sep 0.8.0 - SepWriter Replace StringBuilder with ArrayPool Array](https://nietras.com/2025/05/07/sep-0-8-0/) _2025-05-07 · nietras_ Sep 0.8.0 was released January 19th, 2025 - earlier this year - with two notable changes: ## [Do Compilers Break Constant-time Guarantees?](https://lukasgerlach.me/publication/2025-do-compilers-break-constant-time-guarantees/) _2025-04-14 · Lukas Gerlach_ ## [Lixom: Protecting Encryption Keys with Execute-Only Memory](https://lukasgerlach.me/publication/2025-lixom-protecting-encryption-keys-with-execute/) _2025-04-14 · Lukas Gerlach_ ## [CUDA Hello World: Done Less Wrong](https://ashvardanian.com/posts/less-wrong-cuda-hello-world/) _2025-04-05 · Ash's Blog_ Moving beyond triple-bracket kernel launches to production-ready CUDA: proper error handling, CUDA Driver API, cooperative groups, and inline PTX for robust GPU code. ## [Cascading Spy Sheets: Exploiting the Complexity of Modern CSS for Email and Browser Fingerprinting](https://lukasgerlach.me/publication/2025-cascading-spy-sheets-exploiting-the-complexity/) _2025-02-23 · Lukas Gerlach_ ## [The Longest Nvidia PTX Instruction](https://ashvardanian.com/posts/longest-ptx-instruction/) _2025-02-05 · Ash's Blog_ Exploring the 79-character mma.sp::ordered\_metadata instruction for Tensor Cores on Hopper H100s, plus insights on PTX, SASS, and the evolving complexity of GPU ISAs. ## [Hiding x86 Port Latency for 330 GB/s/core Reductions 🫣](https://ashvardanian.com/posts/cpu-ports/) _2025-01-19 · Ash's Blog_ By exploiting CPU port differences between Intel and AMD, interleaving FMA with addition instructions boosts AVX-512 reduction kernels from 211 GB/s to 330 GB/s per core. ## [Parsing JSON in C & C++: Singleton Tax](https://ashvardanian.com/posts/parsing-json-with-allocators-cpp/) _2025-01-07 · Ash's Blog_ Custom allocators can dramatically reduce JSON parsing overhead, but singleton dependencies lurk everywhere—from malloc to std::isspace—limiting multi-threaded gains. ## [ShadowLoad: Injecting State into Hardware Prefetchers](https://lukasgerlach.me/publication/2025-shadowload-injecting-state-into-hardware-prefetchers/) _2025-01-03 · Lukas Gerlach_ ## [10x Faster C++ String Split, 16 Years Later 👴🏻](https://ashvardanian.com/posts/splitting-strings-cpp/) _2025-01-02 · Ash's Blog_ A 16-year-old StackOverflow question on string splitting gets a modern answer: SIMD-accelerated tokenization that's 10x faster than STL and cleaner than ranges.