Re: No More Windows/Xbox Franken-Handheld?

On Fri, 6/6/2025 1:39 AM, Joel wrote:

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

On Thu, 5 Jun 2025 20:08:47 -0400, Paul wrote:
>

But just making the "Winblows is overweight" comment, that's not going
to cut it.

>
It’s absolutely true. Just try running Windows 11 versus a typical Linux
distro on the same hardware, running the same cross-platform applications.
Or even not cross-platform: consider how the Steam Deck is running so many
Windows-specific games better under Linux+Wine than Windows-based
competitors can manage running them natively.

 
 
I had a PC ideal for Windows 11, until it wasn't anymore, which is why
I'm firmly in Linux on a Win11-capable PC.  :Paul is arguing from the
wrong perspective.  It's fair to say I could boot Win11, it's not fair
to say it'd be as sleek and responsive as Linux, otherwise what
would've motivated me to change so much?
 

I have some strict definitions of what performance is.

For example, if an ecosystem makes me add retpoline to the
application, that slows down the application a bit. I could then
make the claim that "in the year 2005 it ran in 10 seconds,
in the year 2025 it takes 12 seconds on the same hardware". And
that's a tax forced by spectre and meltdown.

In gaming, it has always been the case that a game developer
can do things, to disadvantage one platform or ecosystem,
with respect to another. For example, Google makes an
OpenGL to DirectX3D layer for the cake. If I'm a lazy dev,
I "write once" and only do an OpenGL version. Maybe I
include the Google module as a "tax" I make the Windows user
pay. If instead, the game had been built with a DirectX3D
stack on the Windows side, no Google tax in sight, now
we can compare the speed of each ecosystem and make comments
about which ecosystem is actually faster (because of the choices
that ecosystem made for its graphics environment). We need people
to make versions that "run best" on the respective OS, not
make one version that "sorta runs on the OS I don't happen to like".

If I do a shit job of design, and throw in the odd pork
pie to dieadvantage a platform, we can't go around making
statements without making note of what the stack looks like
that was used.

I'm sure you've noticed, that a certain web browser, on Windows
the graphics are 5x slower than they are on Linux. If you look
at the graphics stack on Windows with that web browser, it's had
a pork pie inserted. I could take my own pack of Windows Ninjas,
pull that pork pie out of there, and make a more competitive
test case.

Just working at the "this is so bloated" level, does nothing
for me. I know some of these things could be fixed by the
application developers, if they were incented to do it.

I was reading something earlier today, about a certain topic,
and the claim was made that "some companies write custom code
to fix this". I don't know if the claim is true. Some of these
topics are very hard to research, to see if there is actually
a solution that works better. If everyone reaches for the
"convenience library with the tax included", of course it's
going to suck.

If you want to take shots at the OS that count, you should
be able to identify which specific parts of the OS are bad.
And that information is out there. But saying "it's bloated",
that's not helping anyone.

   Paul