Subject: Re: Compiler timings on varous MVII NetBSDs etc.
To: Pierre-Michel Ricordel <Pierre-Michel.Ricordel@imag.fr>
From: David Brownlee <firstname.lastname@example.org>
Date: 01/23/2001 23:58:22
On Tue, 23 Jan 2001, Pierre-Michel Ricordel wrote:
> About that, but on other platforms, I am always amazed by the fluidity
> of SunOS 4.1 on a 4MB, diskless Sun 3/60, served by another old SunOS box.
> On the other side, I could not be able to compile the NetBSD 1.5 kernel
> in a finite time with a 8MB diskless Sparc.
> I think the 1.5 bloat will begin to be more and more visible to other ports
> (maybe the m68k-based ones ?)
Once again, compare like with like - try using the same vintage
gcc on both platforms, with the same source, and running a kernel
with the same support.
Better still, run the same application on both and see.
I agree there are issues (which is why tech-perform has been set
up), and we should also be providing 'lean' kernels with minimal
device support, but there are much better metrics than compiler
David/absolute -- www.netbsd.org: No hype required --