[comp.unix.i386] Adaptec 1542 kernel fine-tuning

larry@nstar.uucp (Larry Snyder) (06/27/90)

billbr@xstor.UUCP (Bill Brothers) writes:

>Bus-on/off generally won't affect the floppy unless running it on
>the same DMA channel. Here at Storage Dimensions, we are regularly
>using 3 host-adapters (not pure AHA-1542... modified for us) at full
>tilt driving a full complement of disks. 

What on the 1542 are you changing for your drives?

-- 
      Larry Snyder, Northern Star Communications, Notre Dame, IN USA 
            uucp: iuvax!ndmath!nstar!larry  -or-  larry@nstar
 Public Access Unix Site (219) 289-3745 / lots of files & free PEP feeds!

randyt@asdnet.uucp (Randy Terbush) (06/28/90)

In <190@xstor.UUCP> billbr@xstor.UUCP (Bill Brothers) writes:

>larry@nstar.uucp (Larry Snyder) writes:

>>After playing with the various parameters in my kernel to tweak
>>the throughput - I have settled (for the time being) on the following:

>>#define AHA_BUSON       9 
>>#define AHA_BUSOFF      6

>Larry mentions his playing with the DMA rate. I would recommend running
>a DMA test with any configuration that is going to be used with real
>live data. Even one failure is not acceptable when it comes to my data.
>But what he says is true. If your machine can handle it, turning up
>the knobs dramatically improves performance.

I have been tuning my new 1542/ISC2.2 installation also.  Regarding
data transfer rates, I was told that UNIX is more tolerant of lost
data in copying due to its methods of file copying/checking, etc.
More so than other OS's such as DOS.

Is there any truth in this?

-- 
Randy Terbush, 711 Eastridge Place, Boise, ID  83712

UUCP: asdnet!randyt
INET: asdnet!randyt@sunstone.idbsu.edu