Hacker News new | past | comments | ask | show | jobs | submit login

"it seems" combined with "I'm learning" was meant to convey ignorance not arrogance. I think I'm just being misread and that's probably my fault. It wouldn't be the first time - so I'm sorry if you found it off-putting

My day job is optimization for logistics and multi-agent systems on edge deployments. We vary between heterogenous compute environment on-platform and occasional connections to cloud-like servers, but only occasionally. The system has a lot to do and we're pushing what it can do for itself.

I like these problems and am always trying to find new ways to solve them. That's really it, I'm wondering if it's best to focus on learning to work the hardware as is, or learning to twist existing libraries into new problems.

For example if you look at this: https://weightythoughts.com/p/cuda-is-still-a-giant-moat-for...

It actually denigrates anyone who would consider starting at low level CUDA kernels. I liked your message though, you had a great point: what do I imagine all those CUDA programmers do if not write CUDA?




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: