Nicholas Wilt

Nicholas Wilt was on the inception team for CUDA, wrote The CUDA Handbook, and writes at

[Nicholas Wilt @CUDAHandbook]

This tweet is too dense with misrepresentations to dignify with a response other than blocking the poster.

[Nicholas Wilt @CUDAHandbook]

The thing is, and full disclosure I reported directly to Ian for part of the time I worked on CUDA… he is a software guy, and NVIDIA is a hardware company.

[𝐷𝑟. 𝐼𝑎𝑛 𝐶𝑢𝑡𝑟𝑒𝑠𝑠 @IanCutress]

Honestly though, the name that keeps coming up for full time CEO is Ian Buck @ia_buck.

Ian has been at NVIDIA since 2004, co-designed CUDA, and currently runs the datacenter business. A familiar face on stage at conferences and placed in charge while Jensen travels.

But...

[Nicholas Wilt @CUDAHandbook]

@0xCuttlefish @paraschopra Have you tried The Lathe Of Heaven?

[Nicholas Wilt @CUDAHandbook]

@geofflangdale Well. A lot of the mysterious happenings in the book have concrete explanations (big reveals later).

I put up with the literary references by reminding myself that Simmons was an English major

Anyway, thanks for trying lol

https://t.co/FHsGFd2EQg

[👻🚂🔥 @GhostTrainBlaze]

Can we all take a moment to follow Don?

He is one of the only Representatives we have that actually has a backbone and will fight for the people!

Don only has 623 followers; let's amplify him biggly.

[Nicholas Wilt @CUDAHandbook]

Someone has been “liking” each article in the Covert Caches series in turn, at about the interval it would take to read each article, and I’m here for it.

(Posting will resume soon)

[Nicholas Wilt @CUDAHandbook]

@saroshiTV @new_era72 It worked a couple times last year

[Nicholas Wilt @CUDAHandbook]

@kellabyte MMX was integer only, not very wide, aliased FP registers so mutually exclusive with doing any floating point. There has been steady improvement across the decades and AVX-512 is pretty good.

https://t.co/EfyW2MK1VO

[Nicholas Wilt @CUDAHandbook]

It’s more accurate to say you can evade Draconian performance penalties by not violating the store forwarding rules.

https://t.co/IsIhvzxBwU

[Aliaksandr Valialkin @valyala]

TIL: you can increase hot loop performance by 20x by properly exploiting store forwarding property on modern CPUs. And Go better fits than Rust for achieving this performance boost, since it doesn't automatically reorder struct fields.

[Nicholas Wilt @CUDAHandbook]

And every time machine and teleportation portal in science fiction magically accounts for all this motion down to millimeter accuracy

[Philosophy Of Physics @PhilosophyOfPhy]

We were taught that Earth orbits the Sun in a closed circle. In reality, Earth has never returned to the same point in space once in 4.5 billion years.
​Right now, the Sun is carrying Earth and every planet around the center of the Milky Way at 230 km/s (828,000 km/h), more than 18,000 kilometers every 80 seconds.

​Because the Solar System is tilted 60.2° to the galactic plane and bobs up and down through the galactic disc roughly every 30 million years, the planets trace open 3D corkscrew helices across deep space rather than flat loops.

​The last time the Sun stood where it stands today, one Galactic Year (230 million years) ago, the very first dinosaurs were only beginning to appear on Earth. Since forming 4.6 billion years ago, our Sun has completed only 20 laps.

[Nicholas Wilt @CUDAHandbook]

The Jevons Paradox is real, and no profession has embraced it to a greater degree than software engineers.

[Daniel Lemire @lemire]

It is not intuitive at first, but AI will lead to more compute usage all around. And I don't mean GPUs to run AI models. I mean regular boring compute.

It took me an embarrassingly long time to understand.

My AWS bills have gone up. Why? Because it is easier than ever to design and deploy a new service. It is easier than ever to pull up a server and run hours of tests on it. I had servers that were just running idles most days. I can now load them up with endless work.

And this work creates value. More testing, more applications.

The opposing force is that we can better optimize software with AI. It is possible that it may compensate and, in time, stop the growth of hardware usage. But on the short run, our computing needs will grow.

Like it on not, we are going to run many more computers in the future... and we will have to find a way to do so that does not make people angry.

We don't want computing to turn into 'big oil'.

[Nicholas Wilt @CUDAHandbook]

@ID_AA_Carmack There are robots employing lethal force *today*. The military industrial complex has a seamless tech transition from DoD funded toys to local police forces. If the 1033 program can transfer MRAPs and LRADs, it can transfer lethal drones/robot dogs or other dystopic technologies.

[Nicholas Wilt @CUDAHandbook]

@ID_AA_Carmack With 2M pages ftw

[Nicholas Wilt @CUDAHandbook]

@lemire Multiple agents on multiple CMs/servers

You’ll be like a chess grandmaster, context switching from one board to the next

[Nicholas Wilt @CUDAHandbook]

@geofflangdale Are you sure it isn’t vulgar tho

[Nicholas Wilt @CUDAHandbook]

After C (“portable asm”) came Python, Visual BASIC, Java, HTML, DSLs like SQL.. too many layered abstractions to count, enabling new types of software to be built.

[Nicholas Wilt @CUDAHandbook]

This tweet summarizes my experience with AI and gives the transition from assembly to C as a concrete example of software engineers’ decades-long embrace of the Jevons Paradox.

[Tim Sweeney @TimSweeneyEpic]

When I was writing Unreal Engine 1, my time was divided as follows:

- Around 30% went to seriously deep coding on hard problems: assembly language texture mappers, C++ BSP trees, data persistence, object oriented systems design, and networking.

- Around 70% was just typing and refactoring large bodies of code needed for a complete engine feature set. This was not hard, and while I was doing it my mind was mostly on the deeper coding topics, thinking and planning for what's next. It didn't feel much different from vibe coding today, except that it was just me typing code on an IBM Model M keyboard rather than prompting it and reviewing the results.

So, it has always seemed perfectly logical that we would rely on AI for that 70%, once the AI could do it with a similar quality standard as we could, which has never been perfection. It came as no surprise that, after a few years of AI model evolution, it was able to take the reigns on that 70% and free up programmers' time for more valuable tasks.

What's more of a surprise is the emergence of AI as a sort of insta-Carmack able to solve hard technical problems quite cleverly and often cleanly, and that it's able to reason about and work in complex 3D scenes both by operating the tools and by directly understanding the raw data files.

I abandoned assembly language in 1998 when out-of-order CPUs made much simpler C/C++ code faster, and never I missed it. If prompting through architectural documents, concept art, and design plans can transcend typing code and tweaking vertices, that will enable developers to spend their time on higher-level, higher value technical and creative tasks.

What matters is the end product, whether it's a game or an engine. Anything developers can do to try out more ideas and make better software and games faster is a win for everyone.

[Nicholas Wilt @CUDAHandbook]

@bscholl Probably more customer-focused than you realize. Users even of non-life-critical systems often loathe change to familiar user interfaces.

[Nicholas Wilt @CUDAHandbook]

@DonAtkinsonNFL Schneider probably thought it meant something that the entire O line is back from last year. I think they are missing Kubiak and his lieutenants more than anyone.

[Nicholas Wilt @CUDAHandbook]

@mookiealexander Against the 49ers, Holani fumbled in the end zone, enabling a touchdown that made the difference in the game.. bench him first