the hardware:
ppro 200, natoma/triton II
128M RAM
buslogic bt958
scsi drive: seagate ST32550N 2G (narrow baracuda)
eide drive: quantum fireball 3.8G
i did two sorts of tests, the first tests were single bonnies with 300M
files onto both the scsi (gumby_s) and eide (gumby_h) drives. the second
tests were with 5 simultaneous 50M bonnies (gumby_[s|h]1-5).
results are below:
-------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
gumby_s 300 3293 42.8 4478 16.2 1862 11.4 3964 35.4 3685 6.4 62.4 3.4
gumby_h 300 3074 58.7 4405 35.9 1792 56.2 2738 82.1 3702 81.4 44.1 6.4
gumby_s1 50 586 7.6 827 3.0 287 1.7 431 4.0 528 0.9 18.6 0.5
gumby_s2 50 594 7.7 837 2.9 287 1.9 424 4.0 525 1.1 18.4 0.5
gumby_s3 50 588 7.7 892 3.0 288 1.8 420 4.0 524 0.9 18.9 0.5
gumby_s4 50 598 7.8 763 2.8 293 1.9 419 3.9 550 1.0 17.9 0.5
gumby_s5 50 589 7.7 765 2.7 287 1.9 436 4.2 534 0.8 17.9 0.5
gumby_h1 50 644 13.9 1391 8.1 279 7.7 426 12.8 532 10.0 13.3 1.3
gumby_h2 50 664 13.4 2044 9.5 282 7.1 449 12.0 630 12.0 10.0 1.3
gumby_h3 50 657 13.4 1356 8.2 279 7.8 433 12.9 503 9.7 13.2 1.3
gumby_h4 50 686 13.2 2040 9.0 295 7.5 456 13.3 627 11.5 10.0 1.4
gumby_h5 50 646 13.8 1333 8.1 279 7.8 398 11.8 499 9.7 14.7 1.2
is this really what i should expect? looks like the K/sec are about
equivalent, but with the SCSI using less CPU. this really surprises me -
especially with the 5-at-a-time runs; i thought SCSI was supposed to handle
that better than EIDE.
any thoughts?
-----------------------------------------------------------------------------
Matthew Nelson
Dynamics Technology, Inc.
21311 Hawthorne Blvd., Suite 300, Torrance, CA 90503-5610
Voice: (310) 543-5433 FAX: (310) 543-2117 Email: mnelson@dynatec.com