🚀 go-pugleaf

Thomas Leitner (tom@finwds01.tu-graz.ac.at) wrote:
: Brian Tao (taob@gate.sinica.edu.tw) wrote:
: : In article <4216sm$4ic@ia.mks.com>, Andy Toy <andy@mks.com> wrote:
: : >
: : >Here's some more bonnie data from a SPARC 20 model 502MP, 160MB RAM:
						               ^^^^^^
: : >Drives tested were Seagate Barracuda 2 & 4, Hawk 4, and Micropolis 1926.
: : >
: : >bonnie -s 100:
               ^^^^
: : >              -------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
: : >ST12550N  100  1664 92.8  4492 59.1  3593 64.6  1522 95.4 12110 94.6 821.8 106.5
               ^^^^
: : >ST15150WD 100  1709 94.4  5907 80.9  4406 81.7  1526 95.6 10777 89.4 1321.8 160.5
               ^^^^
: : >ST15230N  100  1676 93.3  3705 52.2  3166 57.4  1503 95.0 13140 97.2 775.0 96.1
               ^^^^
: : >M1926-15  100  1544 83.0  2414 29.2  1748 27.3  1389 86.5  9430 70.0 532.6 63.9
               ^^^^

: Can anyone provide these figures for a Conner CFP4207S which is supposed
: to be the most quiet from what I heard so far.

This is for the 4.3 GB Conner 7200 U/min. Machine is a SS10 running 
Solaris 2.4 11/94 with 4x90 Mhz Hypersparcs and 160 MB Memory. 
Note, the load average is relatively high, but the 4 GB Harddisk
hadn't traffic.

# nice -20 bonnie -s 30
10:40am  up 5 day(s), 23:40,  7 users,  load average: 4.26, 4.07, 3.95
              -------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
           30  1505 98.4  5060 90.1  3652 96.4  1595 99.4  9570 97.5 1374.6 179.4

# bonnie -s 100
 10:20am  up 5 day(s), 23:21,  7 users,  load average: 3.41, 3.73, 3.56
              -------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
          100  1527 89.4  4045 82.6  1272 37.8  1476 86.3  6606 81.0 249.1 39.0

# nice -20 bonnie -s 300
 10:35am  up 5 day(s), 23:35,  7 users,  load average: 3.79, 4.00, 3.89
              -------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
          300  1542 98.2  4387 93.7  1330 49.6  1505 97.7  6196 73.0 110.7 25.0

You'll see, that the values for 30 MB are extremely good, due to caching
(RAM). So, if you want to have reasonable results, you should run bonnie
on a file size, that is 2 or three times as large, as your amount of
system memory. Otherwise you'll get wrong results, results that doesn't
represent the real _Disk_Performance_. 

The values above mine seem to be very wrong. The system memory was 
160 MB, the file size of the bonnie tmp-file ONLY 100 MB.

I think the tester didn't know, what he wanted to test ...
Disk or Memory performance ;-))

So beware ....

	Andreas ///






--
andreas@wup.de        /\/\___ Wiechers & Partner Datentechnik GmbH
Andreas Klemm     ___/\/\/        - Support Unix -