my internship is about improving the rendering infrastructure here at Adobe along with Mike. We’ve been using his SPEAR framework and the flexibility is a huge thing, there’s not much you can’t do directly from python you could do in C++
TIL: Khronos is removing HLSL/the most useful frontend in glslang, at the time when HLSL is getting standardized by Ecma.
"The HLSL front-end will be removed at the next major version of glslang, with at least 18 months of notice from this announcement."
https://t.co/300Qts9ADW
Dev story to celebrate the good times at idSoftware [this one in the era before MS ownership] - For Doom 2016, both NVIDIA and AMD had offered onsite devtech support to help in OpenGL to Vulkan porting effort for idTech. So AMD shipped me out to id's Richardson/TX office ...
Before AI, coding was 80% writing code and 20% figuring out the right code to write.
The mentally taxing part was the 20%. As long as you used a familiar language, you could get into a relaxing flow state with the other 80%. You'd need that relaxing time to recharge for when you have to deal with the 20% again.
Now you have to deal with the hard 20% all the time, and it's decision fatigue all day.
Admittedly a bit of a champagne problem, but still a problem.
In this video, we’re looking at the CPU inside the EVO III Pro.
We always pick the best-performing game-optimized CPU available on the market that delivers the best single-threaded performance.
Watch the video for more on the story of meticulously selecting the CPU!
#AFCON25
2M TV leveraged the full power of Reality 5 platform, turning the Africa Cup of Nations coverage into a cutting-edge visual experience and the pinnacle of storytelling.
More details available here: https://t.co/BE9FJJ39Gt
New blog post! "Measuring acceleration structures", in which we will compare BVH costs on various GPU architectures and drivers and attempt to understand the details enough on AMD hardware to make sense of the numbers!
Retweets appreciated :) Link in reply because algorithm.
Steel Swarm: Apocalypse Demo Çıktı!🚀
Hızlı tempolu savaşlar, stratejik derinlik ve birbirinden farklı XBT'ler ile dolu bir dünyaya adım atma zamanı
Yeni trailer:https://t.co/khE84J9SaT
Steam linki video açıklamasında
#SteelSwarmApocalypse#MOBA#DemoRelease#GameDev#Steam
Thread divergence during constant buffer reads can have a big perf impact, this presentation discusses how to detect this on NVidia GPUs using GPU Trace and Shader profiler. Access to ISA is very useful for understanding what the shader does under the hood https://t.co/p7x4iT20xu
@matiasgoldberg How did you manage it without re-creating the process? I wanted our system to recover as well (non-game, multi-threaded system) but VkCreateInstance fails even though validation layer can't find a resource non-destroyed. So I only made crashes go away and our vulkan backend stops
@BattleAxeVR You are absolutely right!
ATM the problem is: Windows' new API (MediaFoundation) converts MJPG to NV12 but there is no decompressor filter for NV12. So if i want my virtual cam to use physical Logi cam's output, i have to write a converter again 😭 Lost a day for same conclusion
It's interesting that most of the GPUs doesn't support RGB24 format.
The API+driver I use for creating virtual camera (Softcam) wants RGB24 for textures which means I have to write texture->buffer converter for every format we have.