François Guthmann 🚄 GPC 26 (@frguthmann) header
François Guthmann 🚄 GPC 26 (@frguthmann) profile picture

Your friendly neighborhood Guthmann | Computer graphics enthusiast | GPU things at AMD. Opinions are my own. https://nitter.cf/t.co/LyF1JrfsMg

Paris, France
Joined December 2012
Photo by François Guthmann 🚄 GPC 26 (@frguthmann)Photo by François Guthmann 🚄 GPC 26 (@frguthmann)Photo by François Guthmann 🚄 GPC 26 (@frguthmann)Photo by François Guthmann 🚄 GPC 26 (@frguthmann)Photo by François Guthmann 🚄 GPC 26 (@frguthmann)
The recordings from GPC24 are up, including mine ! This contains much more information than the slides on their own. Enjoy ! youtube.com/watch?v=sHFb5Xfw… gpuopen.com/presentations/20…
1
18
1
65
8,098
François Guthmann 🚄 GPC 26 retweeted
Honestly, I was so blown away by the quality of the talks this year! If you want to catch up on the awesome slides and demos, the full WebGPU Paris Meetup 2026 presentations are up on YouTube: 👉 youtube.com/playlist?list=PL…
The WebGPU Paris Meetup is NEXT WEEK! 🇫🇷 The lineup is wild: Unity's new renderer, Hugging Face web kernels, WebLLM, William Candillon, and Hexagon's geospatial tech! 100% technical, pure engineering love. Seats are limited, so secure yours now! 👇 🎟️ rsvp.withgoogle.com/events/w…
4
10
978
The recording of my talk about debugging WebGPU with PIX at the Paris meetup is now available online! I walk through debugging an issue I encountered while using the experimental bindless extension :) youtu.be/KQmOrbBihxE
1
2
24
1,037
That's awesome! Ikea also serves GLBs of most of their products on their website. Pretty cool when modeling one's future flat for instance 😁.
Amazon made 3D models of 7,953 real products it sells I packed all 165 GB into about 12 GB of GLB for three.js, WebGPU and web games, free on Hugging Face
7
266
François Guthmann 🚄 GPC 26 retweeted
🎮 AMD FSR Upscaling 4.1.2 comes to XBOX Series X|S! Our ML-based upscaler is now in the October 2026 GDK. 📦 Full and new Lite INT8 models. 🔗 Same FSR SDK API as on PC, plus a new GDK sample to get started. Read the blog ⬇️ gpuopen.com/learn/amd-fsr-up…
91
396
43
3,386
185,506
François Guthmann 🚄 GPC 26 retweeted
The WebGPU Paris Meetup is NEXT WEEK! 🇫🇷 The lineup is wild: Unity's new renderer, Hugging Face web kernels, WebLLM, William Candillon, and Hexagon's geospatial tech! 100% technical, pure engineering love. Seats are limited, so secure yours now! 👇 🎟️ rsvp.withgoogle.com/events/w…
🚨 The agenda for the WebGPU Meetup in Paris (Nov 5th) is live! We have an incredible lineup of talks covering agentic optimization, native debugging, next-gen rendering, and cross-platform mobile graphics. Check out the sessions below and rsvp.withgoogle.com/events/w… 👇
4
1
11
2,802
Wow! That is a PACKED release. Can't wait to play around with some of those new features. The cherry on top, been looking forward to this fix for ages! "Fix stepping into non-inlined helper functions and inspecting their local variables."
156
François Guthmann 🚄 GPC 26 retweeted
✨ AMD FSR "Redstone" SDK 2.4 is now available! ✨ The big news: FSR Upscaling 4.1.2 for AMD Radeon RX 6000 Series GPUs! 🟥 Plus select Ryzen processors, incl. Ryzen AI Max. 🟥 New lightweight model for laptops & handhelds. Read the blog ⬇️ gpuopen.com/learn/amd-fsr-sd…
10
64
4
541
29,419
And experimental bindless support! I almost missed it, had poor man's bindless support in my engine for ages. I can finally get rid of those texture arrays 🥳
What's New in WebGPU (Chrome 155-156) is up! ⚡ State less/greater on builtin(frag_depth) to keep Early-Z optimizations 📦 Create compressed textures with unaligned block sizes 🎨 Render to srgb-linear and display-p3-linear color spaces Check developer.chrome.com/blog/ne… for more!
2
11
695
Everybody chill, the docs are available here: gpuopen.com/amd-gpu-architec…
AMD continues it's enshittification era with requiring a sign in to read isa guides.
1
4
28
2,458
François Guthmann 🚄 GPC 26 retweeted
The AMD Radeon Developer Panel (RDP) is now open source, joining others in our Developer Tool Suite on GitHub! 🔍 Greater transparency into how it works. 🛠️ Contribute improvements & integrate into custom workflows. Explore the code ⬇️ github.com/GPUOpen-Tools/rad… #opensource
1
30
2
130
5,820
A picture really is worth a thousand words I guess 🥲
✅Found a wild token hack that only works properly on Fable 5. Claude charges image tokens by pixel size, not by how much text is in the image. So someone built a proxy that renders your context into PNGs before sending it to the API. Before: 92,000 tokens for a dense tool result After: 4,761 tokens for the same content as a PNG End-to-end bill savings: 59-70%. SWE-bench tasks: same completion rate, half the cost. Link in reply👇
3
728
Oops, the fix (a launcher which attaches the shim with Detours) didn't work with Nsight because it needs to launch the process itself to attach to it. Fine, I made an injector which attaches after the fact instead. Instructions updated ! I promise I tested Nsight this time 🥲
4
410
FSR 4.1 coming to RDNA3 3 7000 series in July 🥳.
As a lifelong gamer, I spend a lot of time thinking about how to push gaming experiences forward across CPUs, GPUs, software, and games. My team and I have been working hard to evolve @AMD FSR 4 and bring it to more cards. We power over 1 billion gaming devices worldwide. It’s a responsibility we care deeply about. 🎮 This July, RDNA 3 players will experience FSR Upscaling 4.1, delivering sharper visuals and smoother gameplay than ever before. I’m grateful to our fans. Your enthusiasm and ideas inspire us to keep pushing gaming forward. FSR Upscaling 4.1 on RDNA 3 will be ready out of the box for Radeon 7000 Series players in over 300 supported games at launch. And for our RDNA 2 players, we have something exciting coming in early 2027. FSR Upscaling 4.1 will be coming to your cards as well, bringing sharper visuals and smoother gameplay to even more gamers. We cannot wait to show you what is next. Stay tuned 🚀
2
249
"Batched Asynchronous Command List APIs"
3
232
François Guthmann 🚄 GPC 26 retweeted
Some time ago, I had the idea to port NVIDIA Physical AI stack to AMD. The motivation was to improve hardware diversity and enable world models and VLAs to run beyond a single ecosystem. We started with NVIDIA Cosmos Predict 2.5-2B. Porting wasn’t trivial: these models are deeply optimized for NVIDIA’s stack. We used this as an opportunity to apply our ROCm kernels. The results were surprising: Both encode and diffusion run faster on AMD Instinct MI300X vs. NVIDIA H200 (FA3) and we still saw significant headroom for further optimization. Quality is unchanged across modalities (validated with WorldJen) To be clear, this is no luck. We have deep experience with diffusion models and AMD GPUs. But this just gives us a good opportunity to get closer to a true hardware-to-hardware comparison, as we work with less software abstractions than usual. Just to give an example, on AMD, memory instructions are async with a hardware queue of ordered pending instructions, enabling concurrent load/store with compute without warp specialization. Bottom line: there are real architectural advantages on AMD, if you take the time to work with the hardware. Note, we did tradeoff ~20% higher memory usage, That being said, AMD has more to give to begin with :) in the coming weeks: AMD versions of Cosmos Transfer and GR00T, an even faster version of Cosmos Predict, and open-sourcing an attention kernel faster than AITER v3 (which is closed-source for some reason? cc: @AnushElangovan )
11
33
4
218
36,683
François Guthmann 🚄 GPC 26 retweeted
You can get into some really fun stuff with the microarchitecture model, start with just the timing of VGPR add instructions to different registers. seb-v.github.io/optimization…
1
2
43
8,519