Is It Worthwhile Running Intel Alder Lake With mitigations=off?

[phoronix]

Phoronix: Is It Worthwhile Running Intel Alder Lake With mitigations=off?

Over the past month of trying out Intel Alder Lake processors on Linux, one of the questions that has come up a few times but not readily disclosed is whether it's still worthwhile on this latest-generation process to boot with "mitigations=off" to disable CPU security mitigations to help squeeze out some otherwise lost performance. Here are some benchmarks to answer that questions.

https://www.phoronix.com/vr.php?view=30722

[avem]

Looks like Intel has done a spectacular job of addressing performance regressions related to the use of mitigations. Kudos!

Hopefully AMD with follow suit with Zen 4/5. Zen 3 sees quite a significant performance boost from disabling mitigations.

[uid313]

Looks like Intel has done a spectacular job of addressing performance regressions related to the use of mitigations. Kudos!

Yeah, but I wonder if within a year there aren't a dozen new vulnerabilities exposed with need for additional mitigations.

[perpetually high]

Zen 3 sees quite a significant performance boost from disabling mitigations.

Source? From my limited pts/osbench benchmarks, practically zero effect disabling mitigations on Zen 3 (5800X). Might have been true before, but no longer true with latest microcodes/patches.

Phoronix post

Openbenchmarking link

[bug77]

Yeah, but I wonder if within a year there aren't a dozen new vulnerabilities exposed with need for additional mitigations.

There always are, no need to wonder.
But I expect the worst is behind us. There are a few years already since this new attack vector has been discovered, plenty of time for protections to be incorporated in the design phase and later. There will be further scenarios that will be exploitable, but they will be fringe cases compared to the initial batches.

[bobbie424242]

I am hallucinating or Phoronix has been changing a default ? Imagine that, CHANGING A DEFAULT ! Who ever do that in the real world ? /s

[jntesteves]

Source? From my limited pts/osbench benchmarks, practically zero effect disabling mitigations on Zen 3 (5800X). Might have been true before, but no longer true with latest microcodes/patches.

Phoronix post

Openbenchmarking link

I agree with you that "quite significant" is hyperbole on avem's part. But still, 2% could be considered significant, depending on each one's standards. I just did the quick math and Michael's benchmarks of the Core i9 12900K only exhibit 0.4% average performance improvement by going "mitigations=off", while your linked test of the Ryzen 5800X shows 2%.

[Slartifartblast]

Is it worth running the iGPU with DDR4 or DDR5 memory, please tell us Mike.

[perpetually high]

I agree with you that "quite significant" is hyperbole on avem's part. But still, 2% could be considered significant, depending on each one's standards. I just did the quick math and Michael's benchmarks of the Core i9 12900K only exhibit 0.4% average performance improvement by going "mitigations=off", while your linked test of the Ryzen 5800X shows 2%.

Thanks for working out that math on that, I was curious myself lol. And yeah I didn't mean to say "zero effect" on Zen 3 since I did see *a* difference but not as much as Alder Lake, which is really impressive by Intel.

[F.Ultra]

I am hallucinating or Phoronix has been changing a default ? Imagine that, CHANGING A DEFAULT ! Who ever do that in the real world ? /s

I know that you are joking, but this is far from the first time Michael did benches with mitigations=off