News for hackers. Stuff that matters.
Re:SIMD < SIMT < SMT: Parallelism in Nvidia GPUs (2011) by majke on Wednesday June 12, 2024 @04:20AM (#40655793)
vpgatherdd - I think that for newer CPUs it is faster than many loads + inserts, but if you are going to fault a lot, then it becomes slow.

> The VGATHER instructions are implemented as micro-coded flow. Latency is ~50 cycles.

https://www.intel.com/content/www/us/en/content-details/8141...

There are no comments on this comment yet.