Gabriel @dss_gabriel
PhD candidate, HPC @ CEA. Architecture, distributed computing, memory management & data layout. RTFM 👹 github.com/dssgabriel France Joined April 2014-
Tweets7K
-
Followers629
-
Following191
-
Likes7K
models such as cuTILE are *explicitly* meant to break away from the non fit for purpose model/memory semantics implied by the legacy cuda. it's why, while in PTX/SASS you can expose things like cache level DMA between SM's, you could never (practically) expose that model....
When an article about hpc/optimization is called "anatomy of" then you know the authors know their shit. For those who don't know, this is basically HPC's "attention is all you need" but less overcooked:
When an article about hpc/optimization is called "anatomy of" then you know the authors know their shit. For those who don't know, this is basically HPC's "attention is all you need" but less overcooked: https://t.co/KWhx3NWPRY
Moore’s law still applies in the sense that the number of transistors in our processors increases exponentially. What broke around 2005 is Dennard scaling. Our clock frequencies are no longer increasing. But you don’t want to buy a processor from 2005: it is going to be much…
Moore’s law still applies in the sense that the number of transistors in our processors increases exponentially. What broke around 2005 is Dennard scaling. Our clock frequencies are no longer increasing. But you don’t want to buy a processor from 2005: it is going to be much… https://t.co/V65QQvnWHa
At end of 2027, AMD will release their MI500 Scale Up Mega Pod which may consist of 256 MI500 chips across three interconnected racks. The outer two racks will house 32 compute trays per rack, while the middle rack will hold 18 switch trays. That amounts to a total of 64 compute…
Post-Modern Cmake - From 3.0 to 4.0 - Vito Gamberini - C++Now 2025 youtu.be/K5Kg8TOTKjU #Coding #Cplusplus #Cpp #Programming
My #ACCU2025 talk is now up! A lot of work in this one: writing 2 versions (maybe more) of a ZX Spectrum emulator in C++. I had a blast & hope you will too following my journey of self-discovery: can this old dog learn some new (C++) tricks? Watch at: youtube.com/watch?v=jlt_fS…
@ryolu_ @cursor_ai everything eventually returns to Deiter Rams
GPU is GEMM Processing Unit :-)
It just feels like an awkward spot. If you want lots of memory+power, save up a bit more for the RTX Pro 6000. If you want to be hacky + fast, hook up a bunch of used 3090s/4090s together. If you want to be cool...get a bunch of TT Blackholes or something idk.
🦀 Rust is how we speed up Python now. The 2025 Python Language Summit revealed that 1 in 3 new PyPI native packages used Rust. Our survey results show that Rust usage for Python grew from 27% to 33%. 📊 From Polars to Pydantic to Granian – Rust is reshaping Python performance.…
Maybe other array folks can do better. The funny part is the initial C++ / Rust code is my code from a YouTube video I made 2 years ago. And the C code is cheating - it is using a constant time formula. youtube.com/watch?v=wGCWlI…
Maybe other array folks can do better. The funny part is the initial C++ / Rust code is my code from a YouTube video I made 2 years ago. And the C code is cheating - it is using a constant time formula. youtube.com/watch?v=wGCWlI… https://t.co/eyJ2qfWaMo
Except you should never use a dynamically resizable array, hashmap, or linked list. These are banned in my codebases. Learn static arrays, pools, and spsc ring buffer queues. Learn AoSoA and SoA. These are your fundamental data structures if you want good real-time performance.
Except you should never use a dynamically resizable array, hashmap, or linked list. These are banned in my codebases. Learn static arrays, pools, and spsc ring buffer queues. Learn AoSoA and SoA. These are your fundamental data structures if you want good real-time performance.
This could have various fundamental implications besides the speedup. Most importantly, most of the attention layer will become memory bound rather than compute bound, shifting the hardware emphasis further from systolic arrays to 3-D memory for inference. Super important.
This could have various fundamental implications besides the speedup. Most importantly, most of the attention layer will become memory bound rather than compute bound, shifting the hardware emphasis further from systolic arrays to 3-D memory for inference. Super important.
NVIDIA research just made LLMs 53x faster. 🤯 Imagine slashing your AI inference budget by 98%. This breakthrough doesn't require training a new model from scratch; it upgrades your existing ones for hyper-speed while matching or beating SOTA accuracy. Here's how it works:…
The number of "world’s fastest parallel file system"s out there is making my head spin. Or, in this case, "will be the world’s fastest." Going fast in one direction ("97 percent network utilization") isn't hard. Hitting the same numbers consistently in production is.
The number of "world’s fastest parallel file system"s out there is making my head spin. Or, in this case, "will be the world’s fastest." Going fast in one direction ("97 percent network utilization") isn't hard. Hitting the same numbers consistently in production is.
Yes it is!
More details on the GB10 Grace Blackwell Superchip by @NVIDIAAI as shown at @hotchipsorg #HC2025 - S-Dielet/S-Die (CPU, Memory Subsystem, etc.) - G-Dielet/G-Die (Blackwell-GPU) - TSMC 3nm (both) - Advanced 2.5D Packaging - 256b LPDDR5X-9400 (301 GB/s) hardwareluxx.de/index.php/news…

Kyra @kyrajohnson68
317 Followers 3K Following
Hexafi @hexsafy
3 Followers 179 Following Bluesky: @hexsafy.bsky.social Mastodon: @[email protected]
Uparnar @Uparnar552281
25 Followers 1K Following
LilMiaPie @Foosaud378
16 Followers 1K Following "The greatest privilege in medicine is being trusted to care for someone's life."
Facts About World @ikshvaku24
0 Followers 21 Following
6nene6malo6 @6little6fang6
61 Followers 1K Following IMAGINE THE SUN, AS A CABBAGGE. BUT IN YOUR CHEST. yes i was drunk when i wrote that
Nagula Vamshi @nagulavamshi121
14 Followers 474 Following
Daniel Waweru @gatuhu
308 Followers 2K Following
Eric Kim @yekime
196 Followers 908 Following CTO, founder @Network_Ocean_ (YC S24) | prev physics+cs @cornell, @palantirtech, @ZFellows
Scramble Code @_scramblecode
21 Followers 192 Following As long as my brain functions, learning is my favorite runtime process
m @bonateadev
14 Followers 406 Following
!.! @xypyth
49 Followers 4K Following
Null @6E6967676572
2 Followers 760 Following
Asif Shaikat @asifshaikat
157 Followers 1K Following
Timoris Nuvw @nuvw909
22 Followers 684 Following
Hitesh Manglani @HiteshM_
25 Followers 372 Following
Ray :D @carcerking
51 Followers 450 Following
KUMAR AKSHAT SINGH @sahewalakshat
11 Followers 326 Following मुझे समझने के लिए आपका समझदार होना जरूरी है| Nation first always & everytime🇮🇳
Benzene @Ben_ZENE_C6H6
98 Followers 4K Following
. @f777e0
14 Followers 301 Following
Fg674 @NaomiMisoraLove
12 Followers 45 Following
Divakar Reddy @DivakarReddyT
3 Followers 56 Following
Rohan Sonawane @imrds7
58 Followers 359 Following Lifelong student | Code 👨💻 | Football ⚽ | F1 🏎️ | cricket 🏏 | Chess ♟️
Nick @Nick_Lojewski
298 Followers 402 Following
Charitas @charitasaccy
171 Followers 648 Following
Shekh Ahammed Adnan B... @adnan_ab_bashir
91 Followers 2K Following Anomaly Detection, Clustering, and LLM Post-training. Knowledge Discovery via Long-Horizon Reasoning.
Petar @piljeg
26 Followers 279 Following
Prashant @prashant3360
60 Followers 203 Following chasing the flow state Other than death, all failure is psychological
Alex little @alittle1092
243 Followers 2K Following Digital Nomad. 🌏 Software Engineer. 👨💻 Lover of Tech. 🫶 Builder ⚒️
HoEatTheDoe @peepeepo0p0o0
3 Followers 417 Following
SidWasHere1(सिड... @SidWasHere1Val
86 Followers 900 Following Scrolls, Finds good tweets and retweets them or like them. The viscious cycle has trapped me in 🥲. But still If you here!! wanna say have a g'day!
Africa.tech @techafricaai
23 Followers 2K Following Africa is the future of AI, and open source will lead the race.
mortrix🎈 @1Mortrix
600 Followers 1K Following spilling the sauce 🍝 skill issuing hard 💪 christ is king ✝️ love to all ♥️
Lip-Bu Tan @LipBuTan1
5K Followers 34 Following CEO of Intel Corporation, Chairman of Walden International, Founding Managing Partner of Walden Catalyst Ventures
Nicholas Wilt @CUDAHandbook
2K Followers 63 Following Nicholas Wilt was on the inception team for CUDA, wrote The CUDA Handbook, and blogs at https://t.co/YkR71W07I7
Ryo Lu @ryolu_
55K Followers 2K Following Head of Design @Cursor_ai. Early @NotionHQ, @Stripe, built startups. I make a world where anyone can make software. Aspiring k-pop idol.
OXMIQ @realoxmiqlabs
311 Followers 5 Following OXMIQ is rearchitecting GPUs from Atoms to Agents™ for next-gen AI, gaming & graphics. IP that scales from silicon to zettascale. visit https://t.co/H4b1y93Xtl for more info
General Matter @generalmatter
7K Followers 16 Following Enriching uranium in America to fill the nuclear fuel gap. Exceptional talent needed.
HSVSphere @HSVSphere
14K Followers 996 Following A colorful sphere, here to grudge. Its opinions will never budge. A vibrant orb, with hues so bright, Unwavering in its stances and might.
sky @skydotcs
8K Followers 951 Following
chrinovicマーク @mu_chrinovic
11K Followers 516 Following computer programmer, systems & HPC, driven by first principles. opinions are totally not my own.
Chris Power @2112Power
20K Followers 5K Following Techno-industrialist, believer in hard power and Pax Americana. Automated Factories and the New American Workforce are how we win.
NNSA @NNSANews
26K Followers 258 Following NNSA delivers the U.S. nuclear stockpile, strengthens nonproliferation, powers a global naval fleet & advances transformative tech.
SemiAnalysis @SemiAnalysis_
34K Followers 16 Following
OGAWA, Tadashi @ogawa_tter
5K Followers 3K Following たった一人で日本の天〇学研究遂行に重大な支障を来らせた罪人、博士 (物理) の皆様には申し訳ありませんでした。余計なツイートをしないように心掛けますが、私ごときが URL込み一次情報を紹介するのすら気に入らない方もいるので何かあったら察して↓ https://t.co/1K63SE3R0z
Trevor Roberts @OrbitalOddity
63K Followers 392 Following Artist. Created "Mystery Flesh Pit National Park" and other internet attractions. Available for commission work!
Raja Koduri @RajaXg
46K Followers 2K Following Create, Clean, Consume is my aspirational routine. My interests math, computer graphics, silicon, software and music.
ST010-1... @st01014
752 Followers 5K Following Code synthesizing Carbon-Silicon based lifeform 🖥️🧬 #Coding #Math #Physics #HPC & #AI #RetroTech #Aerospace #AESTHETICS Random #Nerdiness & #Shitposting
PowerPC Instructions @ppcinstructions
775 Followers 0 Following Ridiculous Instruction Set Computing
Andreas Kling @awesomekling
52K Followers 1K Following building @ladybirdbrowser. recovering addict. husband of @katalinkult. uncle. gymnasium brother.
Robert Clausecker @FUZxxl
1K Followers 272 Following Professioneller Bitschubser. Computer sind nicht schlau, sie sind nur schneller dumm.
LaurieWired @lauriewired
97K Followers 294 Following researcher @google; serial complexity unpacker; https://t.co/Vl1seeNgYK ex @ msft & aerospace
Perplexity @perplexity_ai
336K Followers 63 Following Curiosity changes everything. Download our free app on iOS, Mac, Windows, and Android: https://t.co/BBZ1kG0TVG
Perplexity Developers @PPLXDevs
3K Followers 15 Following Updates for developers building with Sonar. Power your products with the fastest, cheapest API offering out there with search grounding.
High Yield @highyieldYT
6K Followers 96 Following Tech Youtuber. Analyzing hardware and chips of all sizes. Everything silicon.
Dan Nystedt @dnystedt
38K Followers 1K Following Former journalist, now financial analyst. Based in Taipei. Tweet mainly about semiconductors and Taiwan. Not investment advice. Views are my own.
Nick Brown @NickBrownHPC
2K Followers 494 Following Senior Research Fellow @EPCCed, University of Edinburgh. Interested in novel architectures, HPC, FPGAs, RISC-V, programming language design and LLVM & MLIR.
vaxry @vaxryy
11K Followers 104 Following Either your favorite, or your least favorite desktop Linux developer. I make Hyprland and other stuff. I speak 🇺🇸 🇵🇱 🇯🇵 and a bit of 🇩🇪
Lukáš Hozda @LukasHozda
6K Followers 809 Following 🦀🍀 ceo of rust @BraiinsMining × building bitcoin in rust book 🍀🦀
Roy Carrilho @RuiCarrilho5
9K Followers 4K Following CS PhD student, focusing on computer vision, on a (losing) journey to get cracked
OLCF @OLCFGOV
5K Followers 518 Following Oak Ridge Leadership Computing Facility (OLCF) is a designated user facility operated by Oak Ridge National Laboratory & the Department of Energy.
Denis Yaroshevskiy @dyaroshev
822 Followers 141 Following C++ dev. Opinions are my own. Feel free to reach out if you think I know an answer to a technical question you have, I don't mind.
Casey Muratori @cmuratori
61K Followers 145 Following Programming: https://t.co/Bdh1Xj2PpV Comics: https://t.co/fmdjK9HFxW
Scott Chacon @chacon
16K Followers 992 Following CEO of @gitbutler, cofounder of @SCNE_io, previously cofounder of @github. Probably here now: https://t.co/Fvkk95Wgtt
Bryan Cantrill @bcantrill
50K Followers 4K Following Co-founder and CTO of @oxidecomputer. According to @fieldofschemes, "tech exec and Oakland A's fan" -- but more of a Ballers fan now. @bcantrill.bsky.social
ludwig @ludwigABAP
44K Followers 2K Following God’s chosen principal engineer. What is impossible for you is not impossible for me.
Patrick Moorhead @PatrickMoorhead
52K Followers 4K Following Founder, CEO, Chief Analyst @MoorInsStrat. Co-founder of @TheSixFiveMedia and @Signal_65. Healthspan improver. Ex-AMD Corporate VP, AltaVista, Compaq, AT&T.
Lawrence Livermore Na... @Livermore_Lab
66K Followers 997 Following U.S. @ENERGY and @NNSAnews laboratory. We use science and technology to make the world a safer place. Verification: https://t.co/29pFxbpHmQ