Sabitlenmiş Tweet
All my socials are here: omershapira.com/about-me/
Tangentially: what the actual fuck
English
SIMD Crawford 🟣
38.1K posts
SIMD Crawford 🟣
@omershapira
High-functioning millennial ■ AI 4 XR @ Meta ■ Prev: NVIDIA Research, XR, Omniverse ■ VFX⎯Mathematics⎯Electronic Elsewheres ■ Opinions mine, you can't have them
/dev/brooklyn Katılım Temmuz 2009
1.6K Takip Edilen5.3K Takipçiler
Most of the furniture I have ever built started with me thrashing around in a ralph loop on mcmaster.com until I had a functional CAD of what I'd thought of in the shower just then. There must be a better way and no, it does not involve me reading
English
I love McMaster-Carr's site as much as the next girl, but if there's *one* site that could use a VLM-backed chatbot, it's def the one you go in to find out if there's a ding-dong strong enough to connect to that whachamacallit under whatever-the-fuck kilonewtons lateral force
English
SIMD Crawford 🟣 retweetledi
@MainzOnX Don't ever yield to the demands of the public. You're here by your actual name and this is real life, no need to change for your audience
(HPC arch)
English
No idea why 2500+ of you follow me but I very much appreciate it 🙈
Why do you follow me? I want to make sure I’m giving you more of whatever it is!
English
SIMD Crawford 🟣 retweetledi
Did you know that Gaussian splatting always assumed that none of the Gaussians overlap, which is definitely not a correct assumption? Our CVPR 2026 (findings) effort Softmax-GS improves over the non-overlapping assumption in Gaussian splatting, and make multiple Gaussians compete in regions where they overlap using a softmax function, which greatly improves rendering performance when less Gaussians are used, as well as rendering consistency from different viewpoints. Especially, check out our experiments fitting multiple geometric shapes with only 4 Gaussians!
arthurhero.github.io/projects/smgs/
Zhenjun Zhao@zhenjun_zhao
Softmax-GS: Generalized Gaussians Learning When to Blend or Bound @chenziwee, @totoro97_, @HaoTan5, @zexiangxu, @fuxinli2 tl;dr: softmax-based color-merging mechanism for overlapping Gaussians with controllable competition strength arxiv.org/abs/2604.27437
English
I really wonder what the gameplan here is.
Implementing the divergent map of graphics APIs is painful even when the market is on your side.
Are they hoping to win this market on a telco-style subsidy? What happens if they succeed?
Chubby♨️@kimmonismus
NVIDIA has completely halted the release of graphics cards for the gaming market for 2026. Full Focus AI: a gap that China is now also attempting to fill. Not only are high-quality, affordable ev-cars now coming out of China, but - presumably soon - affordable consumer GPUs as well. Curious for benchmarks. If they compete with - let’s say - a RTX 3070 for 1/5 its price this would be a huge win.
English
Pick a region of Mars and find a reason to tweet something racist at it
Fermat's Library@fermatslibrary
Here's a map of Mars if, like Earth, it were covered by water on 71% of its surface.
English
SIMD Crawford 🟣 retweetledi
unfortunately openai didn't publish the unsummarized chain of thought, but the summary is 125 pages!
the model reaches the crucial idea (which it describes as 'frightening,' i would love to read the unabridged chain of thought here...) on page 39
OpenAI@OpenAI
Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in 1946. For nearly 80 years, mathematicians believed the best possible solutions looked roughly like square grids. An OpenAI model has now disproved that belief, discovering an entirely new family of constructions that performs better. This marks the first time AI has autonomously solved a prominent open problem central to a field of mathematics.
English
SIMD Crawford 🟣 retweetledi
It’s a really special time to be alive…some thoughts from training this model 🧵
OpenAI@OpenAI
Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in 1946. For nearly 80 years, mathematicians believed the best possible solutions looked roughly like square grids. An OpenAI model has now disproved that belief, discovering an entirely new family of constructions that performs better. This marks the first time AI has autonomously solved a prominent open problem central to a field of mathematics.
English
@docmilanfar You're saying "Approximately All" editors have resigned?
English
Almost the entire editorial board of the Journal of Approximation Theory just resigned after a dispute with the publisher.
Almost.
You can’t make this stuff up.
English
SIMD Crawford 🟣 retweetledi
Maybe I am simplifying a bit, but in the 1990s onwards numerical optimization was not considered an exciting area. There were many breakthroughs (say, Karmarkar, and in hindsight esp. Nesterov/Nemirovski) but not much funding. Junior faculty in the area were denied tenure. The problems and the algorithms seemed to suffice. Convex problems, tens of thousands of variables. Who needs more?
So let’s spend a minute in silence contemplating the ONE TRILLION spend (this year) devoted to efficiently solving a distributed, non-convex numerical optimization problem in ~1E12 variables. Did we ever get to spend that much in simulations and PDEs for nuclear devices? Not even close; less than $10B/yr. What about NOAA (weather prediction) ? Also lss than $10B/yr.
So this is the biggest ever application of numerical methods ever, by a mile. Also, the most consequential optimization problem, ever.
Many departments withering, or closing. It turns out we were just not looking far enough. I guess there is a lesson about research in this.
Jessica Lessin@Jessicalessin
Good lord. Half-ish of the cloud backlog at Microsoft, Oracle, Google and Amazon is OpenAI and Anthropic????
English
ceremoniously changing my name from SIMD Crawford to Crossbarbra Streisand
English
Joke's on him, no C++ code can be validated
Haider.@haider1
Creator of C++, Bjarne Stroustrup: AI-generated code isn't ready — it generates more bugs, more bloat, more security holes, and is nearly impossible to validate "senior developers are already retiring rather than deal with it" The problem is that even a small prompt change can shift the entire codebase in unpredictable ways
English
SIMD Crawford 🟣 retweetledi
New poll in Lebanon (April 28–May 5) on a potential peace agreement with Israel:
- 84.2% Druze support.
- 76.5% Maronites support.
- 72.2% Orthodox Christians support.
- 51.6% Sunnis support vs. 44.1% oppose.
- 92.1% Shiites oppose.
English
Eurovision Post.
I’ve been a patreon supporter of LOOK MUM NO COMPUTER since ~2019, and absolutely love his attitude, so I can’t remotely be objective. But if being silly to show a divided world that positivity gets you somewhere is suddenly bad, what are we doing here?
English
SIMD Crawford 🟣 retweetledi
Terence Tao says AI is forcing mathematics to rethink graduate education
Many problems used to train young mathematicians now overlap with what AI can plausibly solve, which creates a cultural challenge
"it's hard to define the shift clearly, but we need to address it soon"
English