Re: No More Windows/Xbox Franken-Handheld?

On Thu, 6/5/2025 10:39 AM, Joel wrote:

Lawrence D'Oliveiro <ldo@nz.invalid> wrote:

On Tue, 3 Jun 2025 23:12:17 +1000, Daniel70 wrote:
>

Some time ago, don't MS produce a mobile-phone that ran some sort of
Windows OS.
>
Are they still about??

>
Gone.
>
Microsoft has had a long history of failure with Windows on non-x86
processors. That’s why it’s amusing to see it try yet again to try to put
Windows on ARM machines.

 
 
No, it's not a matter of what CPUs are involved, it's a matter of what
is practical in use, Winblows is overweight on any hardware platform.
It sucks.  Linux is going to outperform it on any machine you can
name.
 

It is possible to find technical deficiencies in the OS, but
you guys aren't doing it with this "hand waving".

The single most successful person at this topic, is the
guy who optimizes Chrome builds at Google, for the Windows platform.
Since the Chrome developers push a button on their powerful desktops,
and do one full build after another, he has been assigned the job
of making this as efficient as possible. His job is to reduce the
runtime of a Chrome build on Windows. Other staff do similar things
for whatever other OSes they support.

He has observed things which are deficient. And he has reported them.
And, they've been bandaided, but the information he gave, did not
cause any soul searching at Microsoft. "Whoa, this is bad, we should
fix this" said no one.

But just making the "Winblows is overweight" comment, that's not
going to cut it. Most of the SVCHOST, aren't using any cycles
at all. Only a few regularly use cycles.

What you are observing, at least some of it, is scroll throttles.
By using scroll throttles, fast machines are slowed down, so
people owning N100 processors don't feel bad about themselves.
Scroll throttles are things like graphics animations.

If your compute task does not spray the screen with graphics,
then a scroll throttle should not be present when you're running
SuperPI XS 1.5 for 32M.

Summary: It *is* possible to score points, but you haven't impressed
         me yet with your "research". Try harder :-) I know where
         the weaknesses are. Come on, sink my battleship :-)

         I did a series of benches between Windows versions, because
         a few kooks were insisting "every new Windows version ran faster
         than the previous one". Which is nonsense. You'd be surprised
         which OS won, but the difference of a couple percentage points
         is not significant. The answer then, is changing the OS version,
         does not make  SuperPI XS 1.5 for 32M  run faster. Do the
         fluffy parts of the OS design cause indigestion ? Yes.

   Paul