some interpretation is in order here...

for all of the tests except NetBSD-1.5Y, both runs are reading
the file from disk.  (NetBSD-1.5X is doing something funky,
let's just ignore that since it no longer matters.)
thus, both runs take the same amount of time.

for NetBSD-1.5Y, the second run has the whole file still in memory,
so copying that out to the disk is very fast.  keeping the whole file
in memory is great for this benchmark, but there are other situations
where it's not so great, so this will probably change.

the more important question I have is:  why does it take 
NetBSD almost 4 times longer to read the file from disk?
(ie. each of the first runs.)  were the source and destination
filesystems on different physical disks or the same one?
did you wipe the disk(s) and start from scratch for each OS,
or do something else like installing all three on the same disk?
did you lay out the partitions the same way for each OS?

-Chuck


On Sat, Sep 22, 2001 at 07:15:54PM +0200, mauzi@expertlan.hu wrote:
> some new test results that may be interesting:
> 
> disk-to-disk copy, ``time cp bigfile bigfile2'' where
> ``bigfile'' is as large as 128MB.
> 
> test machine: pIII-800 / 256MB / IBM DTLA ATA harddisk
> 
> TEST 1: NetBSD-1.5.1 (options BUFPAGES=16384)
> 
> first run:	64.4s real
> second run:	63.8s real
> 
> TEST 2: NetBSD-1.5X (UBC, prior to Chuck's UVM updates)
> 
> first run:	64.6s real
> second run:	36.3s real
> 
> TEST 3: NetBSD-1.5Y (UBC, with Chuck's UVM updates, dirpref)
> 
> first run:	63.3s real
> second run:	 8.5s real
> 
> TEST 4: FreeBSD 4.3-RC5
> 
> first run:	18.3s real
> second run:	18.9s real
> 
> TEST 5: Linux 2.2.19
> 
> first run:	17.2s real
> second run:	17.0s real
> 
> the bufcache performance with 1.5Y is very impressive.
> 
> -- mauzi
> 
> Gergely EGERVARY
> System Administrator
> Business Polytechnic, HUNGARY