Nathan Hoobler
@nathanhoobler.bsky.social
Pixel Pusher and Digital Illusionist
@ Nvidia
All opinions my own
@ Nvidia
All opinions my own
Reposted by Nathan Hoobler
You don't need to understand the problem or the proofs to get at the interesting part: mathematicians working in communities and using LLMs mostly to generate "workhorse" proofs of key elements of the problem (augmentation), rather than just dumping the problem in and letting the LLM do everything
The best description you'll read anywhere of how cutting-edge quick-turnaround mathematical research looks like today. It's
1. Not what the Altmans of the world are selling.
2. Not how it was done even a few years ago.
3. Reasonably mind-blowing. 1/
From/partially by @teorth.bsky.social
1. Not what the Altmans of the world are selling.
2. Not how it was done even a few years ago.
3. Reasonably mind-blowing. 1/
From/partially by @teorth.bsky.social
The story of Erdős problem #1026
https://terrytao.wordpress.com/2025/12/08/the-story-of-erdos-problem-126/
#math #mathematics
https://terrytao.wordpress.com/2025/12/08/the-story-of-erdos-problem-126/
#math #mathematics
December 11, 2025 at 1:25 AM
You don't need to understand the problem or the proofs to get at the interesting part: mathematicians working in communities and using LLMs mostly to generate "workhorse" proofs of key elements of the problem (augmentation), rather than just dumping the problem in and letting the LLM do everything
Reposted by Nathan Hoobler
After nine years of development, meshoptimizer has reached its first major version, 1.0!
This release focuses on improvements in clusterization and simplification as well as stabilization. Here's a release announcement with more details on past, present and future; please RT!
meshoptimizer.org/v1
This release focuses on improvements in clusterization and simplification as well as stabilization. Here's a release announcement with more details on past, present and future; please RT!
meshoptimizer.org/v1
🐇 meshoptimizer v1.0
Mesh optimization library that makes meshes smaller and faster to render
meshoptimizer.org
December 8, 2025 at 4:56 PM
After nine years of development, meshoptimizer has reached its first major version, 1.0!
This release focuses on improvements in clusterization and simplification as well as stabilization. Here's a release announcement with more details on past, present and future; please RT!
meshoptimizer.org/v1
This release focuses on improvements in clusterization and simplification as well as stabilization. Here's a release announcement with more details on past, present and future; please RT!
meshoptimizer.org/v1
Reposted by Nathan Hoobler
ImPlot 0.17 + ImPlot3D 0.30 released:
- github.com/epezent/impl...
- github.com/brenocq/impl...
brenocq (Breno Cunha Queiroz) main author of ImPlot3D took over maintenance of ImPlot, you can sponsor him at: github.com/sponsors/bre...
- github.com/epezent/impl...
- github.com/brenocq/impl...
brenocq (Breno Cunha Queiroz) main author of ImPlot3D took over maintenance of ImPlot, you can sponsor him at: github.com/sponsors/bre...
December 1, 2025 at 11:03 AM
ImPlot 0.17 + ImPlot3D 0.30 released:
- github.com/epezent/impl...
- github.com/brenocq/impl...
brenocq (Breno Cunha Queiroz) main author of ImPlot3D took over maintenance of ImPlot, you can sponsor him at: github.com/sponsors/bre...
- github.com/epezent/impl...
- github.com/brenocq/impl...
brenocq (Breno Cunha Queiroz) main author of ImPlot3D took over maintenance of ImPlot, you can sponsor him at: github.com/sponsors/bre...
Reposted by Nathan Hoobler
Ah, nice, yet another "wrote a huge chunk of the bloody engine" credit gets marked up as "additional programming." As a contractor/consultant you have much better financial control over your future but don't ever expect to be part of the team. If you want legitimate credit: put it in your contract.
November 24, 2025 at 11:00 AM
Ah, nice, yet another "wrote a huge chunk of the bloody engine" credit gets marked up as "additional programming." As a contractor/consultant you have much better financial control over your future but don't ever expect to be part of the team. If you want legitimate credit: put it in your contract.
Reposted by Nathan Hoobler
the economic success of the U.S. is significantly built on the land grant universities and in particular their excellent agricultural science tradition.
Really important to stress that the Crown Jewels of the US higher education system were never the Ivies or elite SLACs (other countries have equivalents of these) but the well-funded, large, cheap, and excellently staffed public state university systems bringing high quality education to the masses.
November 23, 2025 at 5:26 PM
the economic success of the U.S. is significantly built on the land grant universities and in particular their excellent agricultural science tradition.
Reposted by Nathan Hoobler
If you use GMail, AI (Gemini) was turned on yesterday by default and now scans all of your content for machine learning. To turn off, go to Settings>General and scroll down. Uncheck the box for "Smart features."
There's other "Smart" add-ons as well, but that's the one that reads your content.
There's other "Smart" add-ons as well, but that's the one that reads your content.
November 20, 2025 at 5:32 PM
If you use GMail, AI (Gemini) was turned on yesterday by default and now scans all of your content for machine learning. To turn off, go to Settings>General and scroll down. Uncheck the box for "Smart features."
There's other "Smart" add-ons as well, but that's the one that reads your content.
There's other "Smart" add-ons as well, but that's the one that reads your content.
Reposted by Nathan Hoobler
other people's intuitions about the composition of thought are very different than my own. this paragraph treats as absolutely ridiculous something i would say is self-evidently true.
November 18, 2025 at 12:16 AM
other people's intuitions about the composition of thought are very different than my own. this paragraph treats as absolutely ridiculous something i would say is self-evidently true.
Reposted by Nathan Hoobler
thanks to a colleague I just learned about isaac rosenberg's haunting "break of day in the trenches" and I think you should learn about it, too, on this remembrance day www.poetryfoundation.org/poetrymagazi...
November 11, 2025 at 1:27 PM
thanks to a colleague I just learned about isaac rosenberg's haunting "break of day in the trenches" and I think you should learn about it, too, on this remembrance day www.poetryfoundation.org/poetrymagazi...
It's kind of fascinating to me that two of the prime obsessions of the 21st century science are "meat without mind" and "mind without meat"
If people want to do this, we need to learn how to make a bunch of enzymes at scale and it may actually not be possible for cheaper than real meat for a long time
I think what is going to happen is that lab-grown meat will gradually catch up with real animals and when that happens we will all quietly admit that the vegans were right about factory farming
November 11, 2025 at 6:19 AM
It's kind of fascinating to me that two of the prime obsessions of the 21st century science are "meat without mind" and "mind without meat"
Reposted by Nathan Hoobler
Impressive as that sounds, thermal engineering has lagged behind transistor density improvements.
Thermal envelopes significantly cap current chip performance.
Any "breakthrough" in cooling technology (think microfluidics) will have disproportionally large effects on human progress.
Thermal envelopes significantly cap current chip performance.
Any "breakthrough" in cooling technology (think microfluidics) will have disproportionally large effects on human progress.
October 27, 2025 at 7:12 PM
Impressive as that sounds, thermal engineering has lagged behind transistor density improvements.
Thermal envelopes significantly cap current chip performance.
Any "breakthrough" in cooling technology (think microfluidics) will have disproportionally large effects on human progress.
Thermal envelopes significantly cap current chip performance.
Any "breakthrough" in cooling technology (think microfluidics) will have disproportionally large effects on human progress.
Reposted by Nathan Hoobler
From a crewmember on yesterday's Teal 74 mission into now-Category 5 Hurricane #Melissa. As clear of an eye as you will see in the Atlantic basin.
October 27, 2025 at 4:11 PM
From a crewmember on yesterday's Teal 74 mission into now-Category 5 Hurricane #Melissa. As clear of an eye as you will see in the Atlantic basin.
Reposted by Nathan Hoobler
Couldn't be prouder of my intern @bingxxu.bsky.social
for leading this work! 🚀
It took immense persistence to turn the initial hunch, that an attention mechanism could model long-range interactions in light transport, into a robust, working system. Huge congrats to the entire team! 🎉
for leading this work! 🚀
It took immense persistence to turn the initial hunch, that an attention mechanism could model long-range interactions in light transport, into a robust, working system. Huge congrats to the entire team! 🎉
🎄 Introducing our paper A Generalizable Light Transport 3D Embedding for Global Illumination lnkd.in/gQUMSAyV .
🙈 Just as Transformers learn long-range relationships between words or pixels, our new paper shows they can also learn how light interacts and bounces around a 3D scene.
🙈 Just as Transformers learn long-range relationships between words or pixels, our new paper shows they can also learn how light interacts and bounces around a 3D scene.
October 23, 2025 at 3:40 AM
Couldn't be prouder of my intern @bingxxu.bsky.social
for leading this work! 🚀
It took immense persistence to turn the initial hunch, that an attention mechanism could model long-range interactions in light transport, into a robust, working system. Huge congrats to the entire team! 🎉
for leading this work! 🚀
It took immense persistence to turn the initial hunch, that an attention mechanism could model long-range interactions in light transport, into a robust, working system. Huge congrats to the entire team! 🎉
Reposted by Nathan Hoobler
This feels very true. With copilot I kinda want a sub-PR mode (that it doesn't have). I want it to create a tiny change, show it to me as a difference, chat with me about it, iterate. Then at some point I'd accept it, and commit.
This granularity doesn't exist, sadly. Either tab-complete, or a PR
This granularity doesn't exist, sadly. Either tab-complete, or a PR
Andrej Karpathy’s take on AI coding agents feels grounded. The industry’s chasing full autonomy when models still hallucinate too much.
Agents that churn out a thousand lines of code leave you either blindly trusting them or slogging through reviews. These tools should embrace their fallibility.
Agents that churn out a thousand lines of code leave you either blindly trusting them or slogging through reviews. These tools should embrace their fallibility.
October 19, 2025 at 8:21 PM
This feels very true. With copilot I kinda want a sub-PR mode (that it doesn't have). I want it to create a tiny change, show it to me as a difference, chat with me about it, iterate. Then at some point I'd accept it, and commit.
This granularity doesn't exist, sadly. Either tab-complete, or a PR
This granularity doesn't exist, sadly. Either tab-complete, or a PR
Reposted by Nathan Hoobler
Proud papa moment: my son J published a letter in the @nytimes.com today (in print tomorrow) on the need to think our definitions of autism.
October 18, 2025 at 8:06 PM
Proud papa moment: my son J published a letter in the @nytimes.com today (in print tomorrow) on the need to think our definitions of autism.
Reposted by Nathan Hoobler
Relatedly I found this graph on some tech progress for computers, just to show how things don't always just go up forever necessarily.
October 14, 2025 at 4:08 PM
Relatedly I found this graph on some tech progress for computers, just to show how things don't always just go up forever necessarily.
Reposted by Nathan Hoobler
Reposted by Nathan Hoobler
"Inside NVIDIA GPUs: Anatomy of high performance matmul kernels", includes a great intro to GPU architecture and PTX/SASS: www.aleksagordic.com/blog/matmul
Inside NVIDIA GPUs: Anatomy of high performance matmul kernels - Aleksa Gordić
From GPU architecture and PTX/SASS to warp-tiling and deep asynchronous tensor core pipelines.
www.aleksagordic.com
October 1, 2025 at 7:36 PM
"Inside NVIDIA GPUs: Anatomy of high performance matmul kernels", includes a great intro to GPU architecture and PTX/SASS: www.aleksagordic.com/blog/matmul
Reposted by Nathan Hoobler
My absolutely favorite paper in this space I recommend to anyone working on reconstruction or inverse tasks (or even any machine learning): arxiv.org/abs/1711.06077 "Perception-Distortion Tradeoff". Goes beyond quantifying the error/distortion of the reconstruction statistically, 2/
September 17, 2025 at 1:06 AM
My absolutely favorite paper in this space I recommend to anyone working on reconstruction or inverse tasks (or even any machine learning): arxiv.org/abs/1711.06077 "Perception-Distortion Tradeoff". Goes beyond quantifying the error/distortion of the reconstruction statistically, 2/
Reposted by Nathan Hoobler
September 12, 2025 at 11:03 AM
Reposted by Nathan Hoobler
I got tired of mashing together tools to write long threads with 𝐫𝐢𝐜𝐡 𝑓𝑜𝑟𝑚𝑎𝑡𝑡𝑖𝑛𝑔 and ℳα†ℏ—so I wrote La𝑇𝑤𝑒𝑒𝑡!
It converts Markdown and LaTeX to Unicode that can be used in “tweets”, and automatically splits long threads. Try it out!
keenancrane.github.io/LaTweet/
It converts Markdown and LaTeX to Unicode that can be used in “tweets”, and automatically splits long threads. Try it out!
keenancrane.github.io/LaTweet/
September 11, 2025 at 1:28 PM
I got tired of mashing together tools to write long threads with 𝐫𝐢𝐜𝐡 𝑓𝑜𝑟𝑚𝑎𝑡𝑡𝑖𝑛𝑔 and ℳα†ℏ—so I wrote La𝑇𝑤𝑒𝑒𝑡!
It converts Markdown and LaTeX to Unicode that can be used in “tweets”, and automatically splits long threads. Try it out!
keenancrane.github.io/LaTweet/
It converts Markdown and LaTeX to Unicode that can be used in “tweets”, and automatically splits long threads. Try it out!
keenancrane.github.io/LaTweet/
Reposted by Nathan Hoobler
I've got a new blog post for all of you fine folks! It runs through the additions to D3D12 since it was released, and finishes up with some of the things that have changed for me personally in my code.
(And yes it's really been 10 years 👴).
therealmjp.github.io/posts/ten-ye...
(And yes it's really been 10 years 👴).
therealmjp.github.io/posts/ten-ye...
Ten Years of D3D12
For those of us that have been using it from the start, it can be hard to believe that Direct3D 12 has been around for nearly ten years now. Windows 10 was released on July 29th 2015, and D3D12 has be...
therealmjp.github.io
September 8, 2025 at 12:49 AM
I've got a new blog post for all of you fine folks! It runs through the additions to D3D12 since it was released, and finishes up with some of the things that have changed for me personally in my code.
(And yes it's really been 10 years 👴).
therealmjp.github.io/posts/ten-ye...
(And yes it's really been 10 years 👴).
therealmjp.github.io/posts/ten-ye...
Reposted by Nathan Hoobler
I am boycotting #SIGGRAPHAsia 2026 because it will be in Malaysia, where there are serious safety concerns for the LGBTQ+ members of our community, thereby excluding them. I will not attend, submit my work to, or perform any reviews for SIGGRAPH Asia 2026.
Yet, I'm excited about #SIGGRAPHAsia 2025!
Yet, I'm excited about #SIGGRAPHAsia 2025!
September 5, 2025 at 11:37 PM
I am boycotting #SIGGRAPHAsia 2026 because it will be in Malaysia, where there are serious safety concerns for the LGBTQ+ members of our community, thereby excluding them. I will not attend, submit my work to, or perform any reviews for SIGGRAPH Asia 2026.
Yet, I'm excited about #SIGGRAPHAsia 2025!
Yet, I'm excited about #SIGGRAPHAsia 2025!
Reposted by Nathan Hoobler
been a couple years since I wrote a dumb pathtracer-in-a-shader; felt that itch again and decided to brush off the cobwebs by 'covering'/'blatantly ripping off' noby's brilliant 10yo 4k www.youtube.com/watch?v=9r8p...
September 4, 2025 at 5:08 PM
been a couple years since I wrote a dumb pathtracer-in-a-shader; felt that itch again and decided to brush off the cobwebs by 'covering'/'blatantly ripping off' noby's brilliant 10yo 4k www.youtube.com/watch?v=9r8p...
Reposted by Nathan Hoobler
The Octopus thought experiment sucks. If your instinct is like "LLMs can't learn anything about the actual world from mere text" AND you are intellectually curious, here is your reading list.
1/4
1/4
September 3, 2025 at 8:24 PM
The Octopus thought experiment sucks. If your instinct is like "LLMs can't learn anything about the actual world from mere text" AND you are intellectually curious, here is your reading list.
1/4
1/4
Reposted by Nathan Hoobler
I can't* fathom why the top picture, and not the bottom picture, is the standard diagram for an autoencoder.
The whole idea of an autoencoder is that you complete a round trip and seek cycle consistency—why lay out the network linearly?
The whole idea of an autoencoder is that you complete a round trip and seek cycle consistency—why lay out the network linearly?
August 29, 2025 at 10:46 PM
I can't* fathom why the top picture, and not the bottom picture, is the standard diagram for an autoencoder.
The whole idea of an autoencoder is that you complete a round trip and seek cycle consistency—why lay out the network linearly?
The whole idea of an autoencoder is that you complete a round trip and seek cycle consistency—why lay out the network linearly?