RFC: Optimizing for 386

[Trond Eivind Glomsrød]

Joseph D. Wagner wrote:

>>If you think there will be a performance gain, then rebuild /one/
>>RPM with the optimizations you want and show how much faster it is.
>>    
>>
>
>I actually did that once.  I recompiled KDE from scratch.  36 hours.  No noticeable difference.  Then I recompiled the underlying Qt libraries from scratch.  5 hours.  A few seconds here, a few seconds there.  Then I recompiled XFree86 from scratch.  4 hours.  Now we're starting to talk increased responsiveness.
>  
>
What aboyt trying just XFree (x.org), then?

>The problem is EVERYTHING from the ground up has to be optimized to feel the cumulative effect.  There is no way you are going to convince me that being optimized for a 386 is OK when I felt the effectiveness of optimizing the graphics programs.
>  
>
Optimization isn't being done for i386 now... it's being done for P4.  
Yes, you can do that and still have the binaries run on i386... the only 
issue then is missing instructions, most of which aren't automatically 
used anyway (MMX, SSE). CMOV could be useful, but I'm not sure of 
more... SSE/MMX are used already in X.

>However, because I didn't do my tests in a "scientific" manor with precise measurements, the developers won't even consider my results.
>  
>
"Feeling" faster just doesn't work, especially when one has just done 
something one believes should have done a lot of difference. That said, 
there are graphic benchmarks available.

> Who? Who is still using a 386? A 486? Do anyone on this list have 
> anything less than a 586? Who are these victims of my push for better 
> optimizations?

I doubt anyone has less than a 586 (and some 686 chips are still 
considered 586 in some ways, as they lack the useful, but optional to 
the 686 instruction set, CMOV), but tuning for pentium is, AFAIR, 
harmful to code running on i686, compared to just optimizing for i386.