On Wednesday 06 August 2003 02:23 pm, Andreas Raab wrote:
> Yes, that I'd be curious about as well. I was seriously wondering
> how removing CSE would affect other parts of the system. The VM
> isn't just interpret() and while speeding it up typically helps
> it's important to balance the use of compiler flags with the
> overall performance.
>

You've got a point there.

Just re-compiling gnu-interp.c with -O2 -fno-gcse results in faster 
tinyBenchmarks and slower macroBenchmarks:

-O2

0 tinyBenchmarks '112676056 bytecodes/sec; 4026448 sends/sec'
MacroBenchmarks run #(9362 110430 21756 10669 0 7882 7882)

-O2 -fno-gcse

0 tinyBenchmarks '113777777 bytecodes/sec; 4083052 sends/sec'
MacroBenchmarks run #(9536 114383 21892 10933 0 8200 8200)

However, I wonder whether we've really got the best optimization 
flags. I'm using gcc 3.3 now, and it looked like -O3 resulted in 
worse tinyBenchmarks performance. I should try it with the 
macroBenchmarks.

>
> > -----Original Message-----
> > From: squeak-dev-bounces at lists.squeakfoundation.org
> > [mailto:squeak-dev-bounces at lists.squeakfoundation.org] On
> > Behalf Of John M McIntosh
> > Sent: Wednesday, August 06, 2003 11:10 PM
> > To: The general-purpose Squeak developers list
> > Subject: Re: [VM] Should probably add -fno-gcse to the GCC flags
> >
> >
> > Ok, I get 2% better send/sec, but do you know what it does to the
> > macro-benchmarks?
> >
> > On Wednesday, August 6, 2003, at 12:52  PM, Ned Konz wrote:
> > > The GCC manual recommends trying the -fno-gcse (omit global
> > > common subexpression extraction) flag when you're using the
> >
> > computed gotos. --
> > ==============================================================
> > ==========
> > ===
> > John M. McIntosh <johnmci at smalltalkconsulting.com> 1-800-477-2659
> > Corporate Smalltalk Consulting Ltd.
> > http://www.smalltalkconsulting.com
> >
> > ==============================================================
> > ==========
> > ===

-- 
Ned Konz
http://bike-nomad.com
GPG key ID: BEEA7EFE