Folding Forum

Windows always ran better in the beginning, until all the bug patches and updates installed.
Then it just slows down over time.
Expect the same with Windows 11.

The operating system has very little influence on hardcore number-crunching tasks such as FAH. Don't expect more than perhaps a few percent from a better SMP scheduler, but probably not really. Scheduling is a computer science problem that has effectively been solved to death and back, and everything else about FAH is highly optimized code that is parallelised and runs directly on the hardware (CPU/GPU), and the OS has almost zero influence.

The performance of native operating system tasks, such as filesystem and user-interface operations may still vary between systems, depending on the task, but not as much as it used to.

So don't expect an noticeable increase in FAH performance.

Cheers,
HG.

Hopfgeist wrote:The operating system has very little influence on hardcore number-crunching tasks such as FAH. Don't expect more than perhaps a few percent from a better SMP scheduler, but probably not really. Scheduling is a computer science problem that has effectively been solved to death and back, and everything else about FAH is highly optimized code that is parallelised and runs directly on the hardware (CPU/GPU), and the OS has almost zero influence.

The performance of native operating system tasks, such as filesystem and user-interface operations may still vary between systems, depending on the task, but not as much as it used to.

So don't expect an noticeable increase in FAH performance.

Cheers,
HG.

There is a significant difference in ppd between Windows and Linux, so the OS must make a difference. No way of knowing whether W11 is better than W10 until someone tests it and lets the community know. I plan to do that in the near future (I have modified the W11 ISO file in order to run it on hardware that's a little too old to work officially, so my results may not apply to officially supported hardware)

debs3759 wrote:
There is a significant difference in ppd between Windows and Linux, so the OS must make a difference.

And which one is faster?

I find it rather unlikely that there be a significant difference, unless either there is a substantial code difference between Linux and Windows clients (unlikely for the highly optimised assembly code), or the purportedly "slower" operating system does something immensely stupid, which is also unlikely. So unless I see a large number of unequivocal benchmark comparisons on the exact same hardware (dual boot), over several dozens of work units, I reserve my doubts.

Cheers,
HG.

Linux gives higher ppd. I can't comment on how or why (I assume they use the same code) but guess it's because Windows has a lot of processes working in the background.

debs3759 wrote:Linux gives higher ppd. I can't comment on how or why (I assume they use the same code) but guess it's because Windows has a lot of processes working in the background.

So just hearsay.

I'm not a fan of Windows, but this isn't convincing. I don't care about reasons in this case, but I do care about evidence. And I haven't seen any.

HG.

Hopfgeist wrote:

debs3759 wrote:Linux gives higher ppd. I can't comment on how or why (I assume they use the same code) but guess it's because Windows has a lot of processes working in the background.

So just hearsay.

I'm not a fan of Windows, but this isn't convincing. I don't care about reasons in this case, but I do care about evidence. And I haven't seen any.

HG.

It's more than hearsay when lots of Linux users say it. No good asking Windows users to say why Linux users say it is better.

debs3759 wrote: It's more than hearsay when lots of Linux users say it. No good asking Windows users to say why Linux users say it is better.

Of course Linux users would say that.

Unless you come up with hard numbers, it remains hearsay.

If lots of Windows users also say Linux is faster, it gains some credibility, but not much. Do they?

Evidence or STFU,
HG.

Hopfgeist wrote:

debs3759 wrote: It's more than hearsay when lots of Linux users say it. No good asking Windows users to say why Linux users say it is better.

Of course Linux users would say that.

Unless you come up with hard numbers, it remains hearsay.

If lots of Windows users also say Linux is faster, it gains some credibility, but not much. Do they?

Evidence or STFU,
HG.

I have locked this topic as it has gone of into yet another OS argument.

This comes up at least once or twice a year, there are even old posts where people have tested the effect of using different OS's. Generally Linux and OS X process a bit faster than Windows on the same hardware. The amount varies by different releases and patch levels, as I recall the reported numbers were in the 5-10% range. Lots of theorizing on the reasons, but little definite profiling. Persons who disabled some of the optional Windows services reported less of a difference.

There is also a known difference in how Windows manages PCIe bus access. This mainly affects users of Windows with multiple, fast GPUs.