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

The issue is there's plan b: spill your data to L1 cache.

Whatever plan you have needs to be far faster than L1 cache to even make sense. I can imagine some very heavy register-pressure situations where registers are much better than cache accesses.




Join us for AI Startup School this June 16-17 in San Francisco!

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

Search: