[comp.periphs] faster without cache than with cache?

cjp@beartrk.beartrack.com (CJ Pilzer) (07/14/90)

I have an Adaptec ACB-2322D ESDI disk controller with a Maxtor XT4380E drive.
When running the PC Mag benchmark 5.1 software, the results show that there
is an improvement in every disk performance test with the controller read
ahead cache enabled except the DOS DISK ACCESS test.  This test simply uses
DOS interrupts to create, read and write disk files of various record lengths.
With this test, the results are about 14 secs with the cache disabled and
47.18 secs with the onboard read ahead cache enabled.  This is slower than
the PC AT MFM drive which does 42.68secs.  Can any one explain this result
and what it means in the overall computer performance?

BTW the other tests in the test suite of disk performance are:

	 File create
	 sequential write
	 sequential read
	 random write
	 random read

of groups of records as follows:

	 520 records/512 bytes each
	 64  records/4096 bytes each
	 16  records/16384 bytes each
	 8   records/32768 bytes each

In all these tests the disk performance was better in each part with the
cache enabled.

This is a puzzlement to me.
-- cj