Windows 11 vs. Ubuntu Linux Performance Is Very Close On The AMD Ryzen 9 7950X

[phuclv]

Rule 0 of Linux benchmarking:
Use the performance governor, unless you want to measure power related stuff...

if that's the case then you'll also need to disable CPU scaling on Windows

[phuclv]

Win 11 got lucky, if michael had tested large file copy ...

do you know what mean/average are about? Winning in a test in 100+ doesn't change the final result significantly in any way. And copying large files isn't a typical use case anyway. Does everyone copy big files everyday? Even copying small files may not be common among typical users that only do web browsing at home

[MadCatX]

if that's the case then you'll also need to disable CPU scaling on Windows

Now that would actually be a pretty interesting test. Even more so with power consumption data.

[mdedetrich]

do you know what mean/average are about? Winning in a test in 100+ doesn't change the final result significantly in any way. And copying large files isn't a typical use case anyway. Does everyone copy big files everyday? Even copying small files may not be common among typical users that only do web browsing at home

There are also applications like robocopy if you need really fast file transfer. It's not something fundamental to Windows.

[Anux]

do you know what mean/average are about? Winning in a test in 100+ doesn't change the final result significantly in any way.

You should read my reply to stormcrow.
And we already established that it's a bug that might get fixed in the future. Although I'm not to confident looking back at the print server debacle or how Outlook has still the same errors it had 10 years ago and only got more with every version. :[]

And copying large files isn't a typical use case anyway. Does everyone copy big files everyday?

You really want to make this argument? It's at least much more common than Dacapo, Chess, LZ4, Y-Cruncher, Prime sieve ...

The good thing about such a wide test array is, that you can pick out the benches that are most important to your use case and make decisions based on that. If one OS wins because it's massively faster in Chess, Blender, Prime calculation, etc. but you never use those than the mean/average means exactly nothing for you.

Also there are applications even if they are 2x faster in a benchmark you wouldn't notice because it's already fast enough or your limited by another factor.

So if we are talking about speed impact on what the average user perceives, file copy is certainly much higher in the list then most of these benchmarks. If I had to wait 2x longer an a file transfer that would already take 2 min, it would bother me much more than a web page taking 30 ms longer to render.

There are also applications like robocopy if you need really fast file transfer. It's not something fundamental to Windows.

Sure I'm already seeing all those Win users firing up their cmd and fiddling with robocopy.

[ll1025]

I have never understood why these tests configure Windows with HVCI / VBS / MBEC when this is not the default configuration, and when Ubuntu has no equivalent security measure.

These mitigations are known to significantly affect performance, and it's not exactly an apples-to-apples comparison.

[ll1025]

Windows will never reach Linux performance. It would have to be rewritten completely. There's something strange going on in above tests. I bet the final Linux 6.1 would wipe the floor with Windows as usual.

The "strange" thing is Ubuntu finally enabled the retbleed protections that Windows had for a while. Windows is getting handicapped in these tests because it's running VBS / HEVC / MBEC-- the entire OS is virtualized for security reasons-- and I'm pretty sure it defaults to filesystem encryption. The fact that Ubuntu is neck and neck with it is a pretty bad indictment of Ubuntu.

There are lots of areas where Linux performance is great but there are also a lot of areas where Microsoft has done a ton of work to make it competitive.

[marios]

if that's the case then you'll also need to disable CPU scaling on Windows

I completely agree. I have no idea if Windows use scaling by default though...

[qarium]

Rule 0 of Linux benchmarking:
Use the performance governor, unless you want to measure power related stuff...

i really don't get it why is this performance governor not the default on all systems ?

is it sapotage to make sure microsoft wins benchmarks ?

[jacob]

Windows will never reach Linux performance. It would have to be rewritten completely. There's something strange going on in above tests. I bet the final Linux 6.1 would wipe the floor with Windows as usual.

Windows is not inherently slow, NTFS is. Linux will blow it out of the water in any benchmark that uses IO or filesystem operations, where NTFS would hold it back against Ext4, XFS or Btrfs. Which is ironic because for a while, NT (the predecessor of modern Windows) had better IO performance than then-Linux.