Even with Win10 I can run my stresstest successfully, simply playback (but not performing actual work or changes in the DAW): https://www.tonstudio-forum.de/blog/ind … cks-de-en/

Every system and workload is different. To finally judge, which is better for your type of applications, a side by side
comparison on the same system using the same type of diskdrive/SSD for Win7 like for Win10 is the best thing to do.

Synthetik benchmarks sometimes support us to measure performance, but it can also fail.
I remember reading about systems with SSDs in RAID0 (Stripe). According to synthetik benchmarks
the performance is much higher with 2 SSDs in a RAID0. But when you work with real applications
it doesnt give you this big boost anymore, i.e. Application does not load that much faster.

In Win10 its the same NTFS version 3.1 like in Win XP in regards to the filesystem itself.
What can be different is, how the different subsystem in the kernel interact with each other.
That can create differencies.

On my Win7 system with a Win10 parallel installation, if I remember back right, the difference of
a SSD benchmark (AS SSD) was not that much.

Then you should fine tune your system, that the CPU cores do not become blocked by i.e.
- too many ISR and DSPs (good driver, enable only required HW, ....)
- not required service and processes
- avoid CPU clock changes, Energy saving, CPU core parking ....
to get a CPU which can react agile to a workload and has enough time to process important audio related prosesses.

For Win7 there was a good tuning guide out but the author was disappointed that nobody honored his brick license.
For Win10 I do not know any good tuning guide. Which is an additional reason why I prefer to stay on 7.
Its for me easier, because there I simply know what to do.
On Win10 I would have to cross check every step, whether its still feasible, if things appear to be the same.
That simply much work that I personally would like to avoid if possible.
I avoided this work as much as possible for the parallel installation itself, by using the Windows 10 upgrade process.
But the fine tuning still needs to be done, as not all settings have been taken over by the upgrade process.
But it saved me the time having to install all applications, VSTs, VSTis .. that was already a big win.
But then Win10 brings additional things like the tiny applications running in the Start menue ...
This is work on top to be done for Win10, to look what additionally required for Win10.

This is the task of the process scheduler.
There are tools on the market which allow to tweak this if required like Bitsums "process lasso".
IMHO Only recommended for specialists ... not the full blood musician who is usually only a "casual admin".
But the other tool parkcontrol is great to disable at least CPU core parking ...

Other aspects in regards to Spectre
The Microcode Upgrade against Spectre version 2 had on my system with the same CPU less impact under Win10
because Windows 10 renders fonts not in kernel mode anymore but instead of this in user mode. This creates less context switches for the CPU which are "more expensive" for the CPU since the microcode upgrade for spectre.

If all has been fixed in Win10, then it might be more performant, especially when you need to apply the patches agains spectre. Whether you like Win10 or not for other things (EULA, privacy, GUI, Windows as a service, ...) this is a different topic.

As you can see the question which is better, has many aspects.

But I think most important is, that you need to perform your own test on your own HW, as systems,
configurations, combination of HW and workloads simply differ...