Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-26 Thread Robin Cutshaw
On Wed, Feb 21, 2001 at 10:44:42AM -0800, Peter Wemm wrote: There's a problem here. I tried to configure an SMP kernel but when it booted the fxp0 (Compaq dual eepro100 adapter) got timeout errors and wouldn't work. I went back and did the config/make on the GENERIC kernel and booted

Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-22 Thread Alfred Perlstein
* Robin Cutshaw [EMAIL PROTECTED] [010221 06:07] wrote: OK, I set softupdates on the disk/partition that the build source/target is on. It made no difference in timing. I then created a memory disk, set softupdates on it, and mounted it as /tmp. AMAZINGLY, the build went from 2:50 to

Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-22 Thread Peter Wemm
Robin Cutshaw wrote: On Mon, Feb 19, 2001 at 12:21:26PM -0800, Peter Wemm wrote: Any ideas as to why it would take almost three times longer to build on FreeBSD? This is probably a silly question, but you did recompile the kernel for SMP, right? Actually, I was using the

Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-22 Thread Julian Elischer
Robin Cutshaw wrote: On Mon, Feb 19, 2001 at 12:21:26PM -0800, Peter Wemm wrote: Any ideas as to why it would take almost three times longer to build on FreeBSD? This is probably a silly question, but you did recompile the kernel for SMP, right? Actually, I was using the

Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-21 Thread David Malone
On Wed, Feb 21, 2001 at 09:36:18AM -0800, Alfred Perlstein wrote: * Robin Cutshaw [EMAIL PROTECTED] [010221 06:07] wrote: OK, I set softupdates on the disk/partition that the build source/target is on. It made no difference in timing. I then created a memory disk, set softupdates on

Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-21 Thread Robin Cutshaw
On Mon, Feb 19, 2001 at 12:21:26PM -0800, Peter Wemm wrote: Any ideas as to why it would take almost three times longer to build on FreeBSD? This is probably a silly question, but you did recompile the kernel for SMP, right? Actually, I was using the stock GENERIC UP kernel. I

Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-21 Thread Robin Cutshaw
On Wed, Feb 21, 2001 at 09:02:07AM -0500, Robin Cutshaw wrote: Have you tuned the FreeBSD kernel? It still ships with a worst-case configuration so that it runs optimally on i386 cpus. :-( Copy GENERIC to something else and remove all but 'cpu i686', rebuild and install. Also, get rid

Build timings - FreeBSD 4.2 vs. Linux

2001-02-19 Thread Robin Cutshaw
We just got a couple of Compaq 8500 quad Xeon PIII 700 boxes as daily build servers for the XFree86 tree. I loaded SuSE Linux 7.0 on one box and FreeBSD 4.2-RELEASE on the other. I was surprised to see the large difference in build times. The Linux box compiled in 1 hour and 4 minutes. The

Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-19 Thread Dag-Erling Smorgrav
Robin Cutshaw [EMAIL PROTECTED] writes: Any ideas as to why it would take almost three times longer to build on FreeBSD? Yup: 4.x sucks at SMP. Try the comparison again with uniprocessor kernels - I expect you'll see a much smaller difference. DES -- Dag-Erling Smorgrav - [EMAIL PROTECTED]

Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-19 Thread Robert Watson
HAve you turned on soft updates on your object and target file systems? Synchronous file system events can have a large impact on complex compiles; using -pipe can mitigate the effect fairly significantly. If you want to compare Linux and FreeBSD with more similar file system semantics,

Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-19 Thread Peter Wemm
Robin Cutshaw wrote: We just got a couple of Compaq 8500 quad Xeon PIII 700 boxes as daily build servers for the XFree86 tree. I loaded SuSE Linux 7.0 on one box and FreeBSD 4.2-RELEASE on the other. I was surprised to see the large difference in build times. The Linux box compiled in 1

Re: Build timings - FreeBSD 4.2 vs. Linux

2001-02-19 Thread Jordan Hubbard
Yup: 4.x sucks at SMP. Try the comparison again with uniprocessor kernels - I expect you'll see a much smaller difference. I rather doubt that SMP has anything whatsoever to do with this. 4.x's SMP implementation may be far from optimal, but I've done a lot of my own uniprocessor vs 2 vs 4