Looking At The Linux Performance Two Years After Spectre / Meltdown Mitigations

[phoronix]

Phoronix: Looking At The Linux Performance Two Years After Spectre / Meltdown Mitigations

Last week marked the two year anniversary since the formal public disclosure of the Spectre and Meltdown disclosures. To commemorate that anniversary, I was running some fresh benchmarks of various Intel desktop and server processors with the in-development Ubuntu 20.04 LTS to look at the performance impact today with the default CPU vulnerability mitigations and then again with the mitigations disabled at run-time.

http://www.phoronix.com/vr.php?view=28777

[johannesburgel]

Honestly, Michael: This kind of content should only be available to paying subscribers for the first week after publication.

[slacka]

NoScript + "mitigations=off" for the Win. Don't run untrusted software? No need to lose all your performance.

[MadCatX]

Am I reading this wrong or are some of the hardware mitigations on the 10xxx series just as inefficient as the software ones except that you cannot switch them off? Nicely done.

[F.Ultra]

Am I reading this wrong or are some of the hardware mitigations on the 10xxx series just as inefficient as the software ones except that you cannot switch them off? Nicely done.

Why do you think that the hardware mitigations would not have the same performance penalty? The fix is the same regardless of if it's done in hw or sw.

[nomadewolf]

Why do you think that the hardware mitigations would not have the same performance penalty? The fix is the same regardless of if it's done in hw or sw.

Are you sure that HW mitigations have the exact same impact as SW mitigations?
And, why?

[duby229]

Are you sure that HW mitigations have the exact same impact as SW mitigations?
And, why?

Well if you look at the bar graphs some of those results show the newer chip with hardware mitigations performing much worse than the chips without hardware mitigations. Even though the software mitigations don't much affect the result, the result is that some of those results for chips with hardware mitigations much worse than the chips without hardware mitigations even when using the software mitigations.

Just look at some of the GEGL, GIMP, and OSbench results.... The software mitigations don't much affect that one chip, but that one chip is performing much worse than the comparable older chip. I'd even go so far as to say the hardware mitigations on the newer chip are impacting performance much worse than the software mitigations are on the comparable older chip.

[rastersoft]

If I'm using a desktop computer just for desktop stuff, no server, and only with software from debian repositories, is it safe to disable all these mitigations? Or maybe only some? (and which ones?)

Thanks.

[MadCatX]

Well if you look at the bar graphs some of those results show the newer chip with hardware mitigations performing much worse than the chips without hardware mitigations. Even though the software mitigations don't much affect the result, the result is that some of those results for chips with hardware mitigations much worse than the chips without hardware mitigations even when using the software mitigations.

Just look at some of the GEGL, GIMP, and OSbench results.... The software mitigations don't much affect that one chip, but that one chip is performing much worse than the comparable older chip. I'd even go so far as to say the hardware mitigations on the newer chip are impacting performance much worse than the software mitigations are on the comparable older chip.

I wouldn't be so fast to jump to that conclusion. For that we'd have to benchmark comparable chips such as i9-9980XE against i9-10980XE. Some of the HW fixes seem to help without hurting the performance, see the ctx_clock test for instance. Robbing the user of the opportunity to get some performance back by switching the mitigations off seems rather idiotic.

@Michael: Would you consider to benchmark the effect of HW mitigations on the Cascade Lake chips, assuming you have the hardware to benchmark it on?

[Michael]

I wouldn't be so fast to jump to that conclusion. For that we'd have to benchmark comparable chips such as i9-9980XE against i9-10980XE. Some of the HW fixes seem to help without hurting the performance, see the ctx_clock test for instance. Robbing the user of the opportunity to get some performance back by switching the mitigations off seems rather idiotic.

@Michael: Would you consider to benchmark the effect of HW mitigations on the Cascade Lake chips, assuming you have the hardware to benchmark it on?

Unfortunately I don't have a 9980XE thus why that testing hasn't happened. Nor any extra funds just to buy an idetical chip in the 9000 series if it has hardware mitigations on certain steppings.