Wednesday, November 14, 2018

New comment by dragontamer in "Memory-Level Parallelism: Intel Skylake versus Apple A12/A12X"

AMD's optimization guide for Zen (https://developer.amd.com/wordpress/media/2013/12/55723_SOG_...) suggests:

> The LS unit can track up to 22 outstanding in-flight cache misses.

So not as much as Apple, but more than Intel in this regards. I probably could test this, but I'm tired today, lol.

----------

10 outstanding loads per core seems like a small number, especially compared to contemporary processors (Apple or AMD's). Maybe Intel knows something we don't.



from Hacker News - New Comments: "WordPress" https://ift.tt/2QCCp7H
via IFTTT

No comments:

Post a Comment

MCP Discovery API – Let AI agents find the right tools automatically

Article URL: https://mcp-discovery-production.up.railway.app Comments URL: https://news.ycombinator.com/item?id=46661173 Points: 1 # Comm...