Going by year doesn't help when they have more than one microarchitecture in their product line at a time. If their naming conventions were a bit cleaner and they didn't rebadge older chips to prevent the appearance of having holes in their product lineup, then we could almost go by model numbers, but that's definitely not viable. As difficult as they may be to remember, the microarchitecture codenames are the most concise way to unambiguously specify the affected chips.
Absolutely. And have you run across the phrase "previously called xxxlake?" [1] was all I could lay my hands on quickly. It's confusing when for ex. you're trying to figure out Intel pmu msr addresses.
For the BSDs and Linux, microcode updates can be loaded by the OS. What about Windows? Are microcode updates made available via Windows, or only via BIOS updates if the system manufacturer decides to make updates?
If the latter, what do people do when their system manufacturer doesn't provide updates?
Phoronix certainly has strong technical background. However, I am not friend of their sometimes dramatic writing style. I call them the Linux yellow press :) Let's wait and see whether that SA comes and whether it brings something big.
A nice test would be to compare a "cpuid -1 -r" before and after applied microcode updates. To check if new bit(s) is set somewhere.
Worth might be to check if IA32_ARCH_CAPABILITIES contain new bits set as well on some platforms which would indicate they are not affected. Via the NO_xxx flag.
13 comments
[ 3.4 ms ] story [ 42.1 ms ] threadhttps://en.wikipedia.org/wiki/Coffee_Lake
Intel 2017+ CPUs (Coffee Lake+).
https://ark.intel.com/content/www/us/en/ark/products/codenam...
If the latter, what do people do when their system manufacturer doesn't provide updates?
And that results in news pieces like this.
Worth might be to check if IA32_ARCH_CAPABILITIES contain new bits set as well on some platforms which would indicate they are not affected. Via the NO_xxx flag.