Subject: Re: Slow 486 seems really slow
To: Jonathan R. Hinds <jon@fork.yi.org>
From: Andrew Gillham <gillhaa@ghost.whirlpool.com>
List: port-i386
Date: 02/14/2000 18:34:00
Jonathan R. Hinds writes:
> 
> No, I do not have a memory filesystem mounted on /tmp. I will try this. Is
> this what everybody does on non-pci ide hardware to speed up compiling? Or
> is there still something horribly wrong with this machine?

How about installing 'bonnie' or 'iozone' from pkgsrc and running some
quick benchmarks?  Or, just use 'dd' for some quite simple benchmarks
as well.

On a P-133 with 48MB: 1.4Q, GENERIC (so no SOFTDEP)
wdc0 at isa0 port 0x1f0-0x1f7 irq 14
wd0 at wdc0 channel 0 drive 0: <QUANTUM FIREBALL1080A>
wd0: drive supports 8-sector pio transfers, lba addressing
wd0: 1039 MB, 2112 cyl, 16 head, 63 sec, 512 bytes/sect x 2128896 sectors
wd0: drive supports PIO mode 4, DMA mode 2

dd if=/dev/zero of=junk bs=1m count=20
20971520 bytes transferred in 8 secs (2621440 bytes/sec)
dd if=junk of=/dev/null bs=1m
20971520 bytes transferred in 9 secs (2330168 bytes/sec)

A quick run of bonnie (only 20MB):
              -------Sequential Output-------- ---Sequential Input-- --Random--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
           20  1931 98.4  2702 76.0  1053 33.0  1749 95.0  2423 80.3  74.5  5.7


On this system a compile of GENERIC (1.4Q) took:
3060.0u 433.9s 1:01:27.24 94.7% 0+0k 16951+29287io 1057pf+0w

So, while 12 hours sounds terrible, something like 2-3 hours wouldn't be
unreasonable I guess.

-Andrew
-- 
-----------------------------------------------------------------
Andrew Gillham                            | This space left blank
gillham@whirlpool.com                     | inadvertently.
I speak for myself, not for my employer.  | Contact the publisher.