Programmer/director gamedev. 20+ years shipping games in multiple countries in different roles. Ex-Massive, Funcom, Ubisoft Montreal, Milestone Srl, Bohemia Interactive, now at CDPR. Lots of grey hairs from code, pipelines, bones and bugs.
I only post about gamedev, code, tech and an occasional reminder here and there that the planet is burning.
This is a quite good "reference sheet for optimization techniques", with benchmarks done on their usefulness:
https://hackaday.com/2024/07/13/c-design-patterns-for-low-latency-applications/
There is a link to the paper on the article and sample code on GitHub:
https://github.com/0burak/imperial_hft
It would be interesting to run this code (or equivalent) in dev kits and see if the results are similar.
All these large language models (LLM's) doing coding is great and all, but there is quite a leap from simple apps to large architectures with tons of micro-services and whatnot.
But WHAT IF instead of focusing on LLM's writing the code we don't get these LLM's to *give more contextual compiler errors* instead?
What if MS makes a "compiler assistant" in VS that parses not only the error but also the code around it to give contextual advice?
Could MS... finally make clippy useful?
