Ubuntu 20.04 vs. Windows 10 WSL/WSL2 Performance In 170+ Benchmarks

[phoronix]

Phoronix: Ubuntu 20.04 vs. Windows 10 WSL/WSL2 Performance In 170+ Benchmarks

Earlier this month was a look at the Windows 10 May 2020 Update performance for WSL/WSL2 with many benchmarks and testing on an Intel Core i9 10900K. Here is a follow-up round of testing this time with HEDT performance in the form of running an AMD Ryzen Threadripper 3970X and running even more benchmarks up to 172 in total for this comparison of Ubuntu 20.04 LTS against WSL and WSL2 performance on this newest Microsoft Windows 10 update.

http://www.phoronix.com/vr.php?view=29317

[lucrus]

Boring. No, not the article. Ubuntu is boring. And running Ubuntu inside WIndows is boring²...

[andyprough]

People who want performance would not use Ubuntu or WSL, if they get any productivity at all they will be happy. It would be interesting to throw a performance-tuned Arch or Slackware setup into the mix.

[KaoDome]

HyperV was breaking virtualbox and VmWare workstation.
[...]
So if you want to use a third party and WSL2, your choices are now gone. All third party solution had to modify their software to use Windows WHP APIs.

Here's the thing, Hyper-V acts as a type 1 hypervisor (like Xen or ESXi), not type 2. When it's enabled the whole OS is running under it so virtualization extensions (VT-x, AMD-V) aren't accessible, that's the reason 3rd party hypervisors weren't compatible with it, nested virtualization has been traditionally tricky.

However since VMware Workstation 15.5.5 it's possible to use VMware's virtualization when Hyper-V is enabled. Maybe they make use of WHP or they're virtualizing without hardware primitives (although if that's the case I'm sure there'd be a hefty performance penalty), but it's not necessary to boot with Hyper-V disabled to get VMware's solution working.

I think VirtualBox is also able to work when Hyper-V is enabled (there are references to it in their change log), but I don't use it so take it with a grain of salt.

[Tuxee]

People who want performance would not use Ubuntu or WSL, if they get any productivity at all they will be happy. It would be interesting to throw a performance-tuned Arch or Slackware setup into the mix.

Actually, I work with my Linux desktops and do not run benchmarks all day. That said the performance differences between the various distributions are completely irrelevant for productive work. Anyway, I assume that wasn't your point - that was Ubuntu trolling.

[andyprough]

Actually, I work with my Linux desktops and do not run benchmarks all day. That said the performance differences between the various distributions are completely irrelevant for productive work. Anyway, I assume that wasn't your point - that was Ubuntu trolling.

You made my point. You are able to get work done and you are happy, even though you are not running a high performance system. If you wanted a high performance system you would almost certainly have a different setup. I'm in the same boat - I use MX Linux, a distro that I'm happy with, but not one that wins speed tests.

But if I am going to performance test them, it would make more sense for me to performance test actual competitive setups. For example, Phoronix frequently benchmarks Clear against default setups of Ubuntu and Fedora, which doesn't make sense, because Ubuntu and Fedora have zero chance of winning. Test Clear against a highly performance tuned Arch setup and see if Clear is actually giving a real performance benefit, or if the benefit that had been captured was simply Clear beating slower distros.

[Tuxee]

You made my point. You are able to get work done and you are happy, even though you are not running a high performance system. If you wanted a high performance system you would almost certainly have a different setup. I'm in the same boat - I use MX Linux, a distro that I'm happy with, but not one that wins speed tests.

But if I am going to performance test them, it would make more sense for me to performance test actual competitive setups. For example, Phoronix frequently benchmarks Clear against default setups of Ubuntu and Fedora, which doesn't make sense, because Ubuntu and Fedora have zero chance of winning. Test Clear against a highly performance tuned Arch setup and see if Clear is actually giving a real performance benefit, or if the benefit that had been captured was simply Clear beating slower distros.

In this very case it makes sense from the benchmarkers perspective since it shows Ubuntu both with and without an additional abstraction layer. And if WSL2 is 15% slower than bare metal its 25% slower than Clear.

[sheldonl]

Actually, I work with my Linux desktops and do not run benchmarks all day. That said the performance differences between the various distributions are completely irrelevant for productive work. Anyway, I assume that wasn't your point - that was Ubuntu trolling.

I 100% agree. Any time I see those ubuntu comments it's trolling. I ran Linux as my primary desktop since 1998, debian and ubuntu (and now KDE Neon) have been performant enough for me to do everything I needed to do w/o issue - and convenient enough so I can do my work without having to mess around with a badly behaving OS and have things just work. That, is a major productivity gain compared to how things used to be in the early days.

[Jigglywiggly]

The other problem with wsl2 using hyper v is that even though the newest version of VMware works with it, the performance is terrible from what I have read. So I will stick with wsl1.

[vegabook]

Boring. No, not the article. Ubuntu is boring. And running Ubuntu inside WIndows is boring²...

Oh because, Linux itself is not boring?

LInux is basically corporate now. Own it. Stop pretending that whatever marginal non Ubuntu fake rebel distro is gonna somehow take you back 20 years to the good ol' days. That's over.

If you want seriously not boring, visit Hackaday and do program some assembler. Learn Forth or something. Linux no longer marks you out as some kind of savant. And Ubuntu knows that. It's not such a stretch to say that Linux IS ubuntu. They've all copied it.