Source: Hacker News
Article note: This was one of the things I was most interested in as soon as Intel started talking about microcode patches for localized over voltage issues.
Test protocol isn't completely conclusive because it's a single device data set and there's potentially some context sensitivity, but looks like a good first pass. Possibly related to handling of branch-intensive code?
Basically: slight regressions, near-negligible on most jobs. Most noticible on cryptographically intense loads at around a 10% hit.
Comments 