Biovf

gfxstrand, 11 days ago to random

This week's project: Reworking NVK cbuf support. We've had a lot of issues with too much internal stalling and I think a lot of them come down to the fact that we're re-binding cbufs every draw call.

My plan for root constants, is to do inline updates with the LOAD_CONSTANT_BUFFER command. I don't know how much of a difference there is but I strongly suspect this pipelines much better.

For bound cbufs, I'm planning to just make our dirty tracking way more competent.

We'll see how it goes!

aras, 14 days ago to demoscene

"Everybody Wants to Crank the World", a :demoscene: demo for :playdate:
@playdate. Took 4th place at @outline! #demoscene #playdate #outlinedemoparty

Youtube: https://www.youtube.com/watch?v=QjAKiwQxrQI
Playdate binary: https://aras-p.info/files/demos/2024/Nesnausk_CrankTheWorld-20240421.zip
Source code: https://github.com/aras-p/demo-pd-cranktheworld

Maybe I should write a short blog post about development for playdate, or somesuch. It is quite fun!

image/png
image/png
image/png

aras, 19 days ago to Playdate

Alright, demo submission done. Now we wait! :playdate: :demoscene: #playdate #demoscene

castano, 1 month ago to random

With the discounted price on the Quest 2 I thought it'd be worth adding it to the library.

The demo app works great out of the box, without the need for any tweaks.

However, the benchmark appears to be much more unstable than on cell phones. I imagine that's due to a combination of thermals and background processes interrupting the tests.

Are there any best practices for benchmarking on Quest?

castano, 2 months ago (edited 2 months ago) to random

Huawei has been kind enough to benchmark the Spark codecs on the Mate 60 and the performance is breathtaking. I haven't seen numbers like that on any other phone.

Can wait to learn more about the GPU architecture or get my hands on one of those devices!

aras, 2 months ago to Playdate

Oh no #playdate #blender

NOTimothyLottes, 3 months ago to random

Better energy conservation in sub-pixel motion is a big reason why pre-post graphics with SGSSAA driver overrides looked so good in comparison with what happened after.

Imagine a high frequency signal {16,0,16,0}.
Now imagine a half pixel shift {8,8,8,8}.
Note 'tonemap(8) != tonemap(16)/2 + tonemap(0)/2', so energy is not conserved!
Human will see perceptual flicker as the high frequency signal is in sub-pixel motion.
Instant uncanny valley.

castano, 3 months ago to random

I'm excited to announce that Spark 1.0 is finally out!

https://ludicon.com/spark/

Spark is the culmination of over a year of dedicated research and development, built upon a solid foundation of years of experience in GPU compression technology.

What sets Spark apart?

castano, 4 months ago to random

This is starting to look better 😊

castano, 5 months ago to random

I'm having fun writing holiday greeting letters to my clients and playing with the ignition themed terms that I'm using for my products. I hope it's not too cheesy.

aras, 5 months ago to blender

Someone, uhh, accidentally convinced me to look at #blender video sequencer.

I know absolutely nothing about video, sequencers, timelines, composition or these parts of Blender codebase. Which is to say, perfect! :)

Just landed for Blender 4.1, make drawing UI of complex timelines 4x faster https://projects.blender.org/blender/blender/pulls/115311

But also, I started to look into related parts, and oof, now I want to speed up like 300 other parts. Which one to start with first?