Intel Hyper Threading Performance With A Core i7 On Ubuntu 18.04 LTS

[phoronix]

Phoronix: Intel Hyper Threading Performance With A Core i7 On Ubuntu 18.04 LTS

Following the news yesterday of OpenBSD disabling Intel Hyper Threading by default within its OS over security concerns and plans to disable Simultaneous Multi Threading for other processors/architectures too, here are some fresh Intel HT benchmarks albeit on Ubuntu Linux. The OpenBSD developer involved characterized HT/SMT as "doesn't necessarily have a positive effect on performance; it highly depends on the workload. In all likelihood it will actually slow down most workloads if you have a CPU with more than two cores." So here are some benchmarks using a current-generation Intel Core i7 8700K six-core processor with Hyper Threading.

http://www.phoronix.com/vr.php?view=26469

[brent]

Nice comparison. And the results were as expected. My bullshit detector was instantly triggered when the OpenBSD developers proclaimed that SMT doesn't offer any performance advantage. If that is the case on OpenBSD, it only shows once more how badly OpenBSD performs with multiple cores.

It might be interesting to see how SMT speedup is on Ryzen, compared to Intel. Does such a comparison exist yet?

[speculatrix]

that fits in with my personal benchmarking of HT. In rare circumstances I got 40% improvement, mostly I got 30% ish.

on my VM host at home which has a Haswell Xeon quad core with HT, I am careful to allocate pairs of sibling cores to the same VM.

[starshipeleven]

It might be interesting to see how SMT speedup is on Ryzen, compared to Intel. Does such a comparison exist yet?

Does OpenBSD even boot on Ryzen?

[schmidtbag]

Considering HT added about a little less than 30% performance on average, it's nice to see the 8700K is a little less than 30% more expensive than the 8600K, making the price differences very proportionate. I'm curious if the Ryzen 1300X vs 1500X is the same way.

[Weasel]

HT can offer 100% performance theoretically (like a real core) if you are completely latency bottlenecked, because then most of the execution units will be idle and available for the virtual cores to use. You'd need an extreme "chain" of computations depending on previous one's result for that to happen though, since modern CPUs have very long pipelines.

[skunk]

would have been interesting to see some more server oriented benchmarks (apache, nginx, pgbench, phpbench, openssl, redis) since it's on servers where side channel attacks are more worrisome...

[paulpach]

I guess "doesn't necessarily have a positive effect on performance" is marketing spin on "We expect 30% performance degradation on common workloads".

If you look hard enough, you can find a workload here and there where it does not have an effect, so his statement is technically true.

[pinguinpc]

@Michael

Good tests but if you can make similar test with games will be good, also add ryzen with/without ht

[chuckula]

The OpenBSD guys being a-holes and spreading FUD for no reason? [Oh wait, there is a reason: Free publicity for an OS that's niche even inside of the BSD niche]
That's new!

It's also amusing how much they harp on Intel for being the sole source of all security problems when Hyperthreading is just their implementation of SMT that's used by... let's see here...

1. IBM Power architecture. But clearly IBM just put all that SMT in to be stupid because people spending a few million dollars on their servers don't actually evaluate SMT.
2. SPARC (but at least SPARC is basically dead now).
2. The only AMD product line that's worth mention in RyZen (funny how SMT is BAD for Intel but mysteriously good when AMD "invented" it 16 years later).
3. The newest version of MIPs
4. The only ARM server processor that's even being advertised as somewhat worthwhile for datacenters.

So, in conclusion, Intel is full of idiots because Hyperthreading but everybody else is a genius because SMT.