A while ago I was working on some CUDA kernels for n-body physics simulations. It wasn’t too complicated and the end result was generative art. The problem was that it was quite slow and I didn’t know why. Well the core of the application was written in Clojure so I wrote a simple macro to wrap every function in a ns with a span and then ship all the data to jaeger. This ended up being exactly what I needed - I found out that the two slowest functions were data transfer between the GPU memory and writing out a frame (image) to my disk.
In many other places I see the usefulness of this approach but OTel is too often too geared towards HTTP services. Even simple async/queue processing is not as simple. Though, there have been improvements (like span links and trace links).
Fun fact, there are multiple pieces of software that will real-time correct your eye tracking. Useful for virtual all hands or talks to fix teleprompter eyes.
It's basically like ShaderToy but for WebGPU instead of WebGL. I started it as I have been doing some Rust + wgpu development for art projects and I need a easy way to play around with shaders.
It's very early in development - you can go and just use it right now. But soon I want to support creating an account, saving / sharing shaders, and eventually go beyond the featureset of ShaderToy by allowing for custom input images / textures.
I am in my 20s and I use a typewriter somewhat regularly to journal. I was raised on computers, getting the jumble from my brain onto paper is faster with a keyboard than a pen/pencil and paper. And a typewriter is nice and analog - no screen, no lights, no battery. I'm disconnected, focused, and performant.
To me, its more about the style than the use of an AI. But I agree.
I enjoyed this writeup by Michael Lynch on finding an illustrator [1], for their blog. In doing some of my own writing, I've really found it enlightening how much secondary work goes into publishing your own work. I often think its so nice to be able to _just_ plug in what I want on a site and get a (more or less) free illustration. But as someone selling their own work / time, it feels wrong. I'd rather pay a real human and build a relationship and have something more quality. On the other hand, though, it can be expensive, time consuming, and I've been screwed over. Often it seems like a bigger risk than its worth.
So idk, you're trading some hardship and risk for an ethical dilemma but ease of use.
If you are interested in a more rigorous collection of blending and compositing operations, the W3C "Compositing and Blending Level 1" document is a wonderful resource>
A while ago I was working on some CUDA kernels for n-body physics simulations. It wasn’t too complicated and the end result was generative art. The problem was that it was quite slow and I didn’t know why. Well the core of the application was written in Clojure so I wrote a simple macro to wrap every function in a ns with a span and then ship all the data to jaeger. This ended up being exactly what I needed - I found out that the two slowest functions were data transfer between the GPU memory and writing out a frame (image) to my disk.
In many other places I see the usefulness of this approach but OTel is too often too geared towards HTTP services. Even simple async/queue processing is not as simple. Though, there have been improvements (like span links and trace links).