| Title: | AMIGA NOTES |
| Notice: | Join us in the *NEW* conference - HYDRA::AMIGA_V2 |
| Moderator: | HYDRA::MOORE |
| Created: | Sat Apr 26 1986 |
| Last Modified: | Wed Feb 05 1992 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 5378 |
| Total number of notes: | 38326 |
I'm looking for either a fix, or company to share my misery.
I have a Supra/500, Adaptec 4000, and ST225 (RD31). This weekend I
"upgraded" the Supra software from Series I to Series II version 1.09
- the latest.
Using Diskpera, the measured performance on reads has dropped from
the previous 130Kbytes/sec to a meager 55Kbytes/sec. This performance
degradation was simulataneous with reformatting of the drive.
Trying to go back by formatting with the old version Supra software
brings the guru. Whatever the new supraformat program did, it causes
the old version of supraformat to wolf its cookies!
I called Supra, 503-967-9081. Not impressed. Summary was:
-that I should set interleave to 2 or 3, the default is 1.
I replied that I have tried all interleaves (0,1,2,&3) with no real
differences in result. IN FACT, I seriously doubt that playing
with this setting has any effect!
-that it "beats him" whats wrong, I should be seeing about
175Kbytes/sec on reads.
-that they cannot simulate the problem as the can't duplicate
my configuration
-that nobody else has complained (I doubt this one, as I personally
know at least one other individual with the same results on a 30Meg
drive)
-one thing they helped with, although their methods are less
than optimum. They told me if I want to get the old format program
to work, I should first run the new format program and then power
of the system while its running! This, it seems, is in order to
trash the drive and then the old version of format will work again!
Things someone might do to help me:
-tell me if you have the same problem, if enough people do then
perhaps Supra will take us seriously
-tell me if you have no problem, perhaps we can figure out what
is different in configurations or procedures between us
-if anyone knows how I can decode the track zero information
that format is putting on the drive regarding partion info, etc,
I would be very interested in comparing the results of the two versions
of the format program.
Can anyone help?
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 2929.1 | Buffer size | MQOFS::LECOMPTE | Hubert Lecompte | Mon Sep 18 1989 19:38 | 21 |
Hi Dave,
I had a similar problem when I switched from series
I to series II.
My configuration is different though, I have a A1000 with a real
SCSI drive (MiniScribe 8051S).
The only way I was able to fix it was to "play" with the value
of buffers and interleave factor on the drive.
Doing this I currently have about 200K on a read and 65K on a write
using Diskperfa with 64 for buffer.
It look to me as if the new driver (or combination of driver, drive
setup and transfer rates) saturates at about 54K of read transfer.
Using SupraEdit you should be able to change the buffer value and
obtain more speed.
Good luck.
Hubert
| |||||
| 2929.2 | maybe this will save someone a week of free time! | CGOFS::WADLEIGH | Dave in Calgary, Alberta | Wed Sep 20 1989 20:57 | 38 |
re -1 - I tried the buffer size, thanks, but no performance difference.
I finally figured it out, by attacking the SupraFormat executable
with a debug utility, Adaptec manual in hand!
Here's what happens. If you turn off the MAP option, because you
know you have a defect free platter and want to save a couple hours
time.....then the SupraFormat program does not send the INTERLEAVE
settings you have specified on the screen. (SCSI Format command
coming from host adaptor to controller should be followed by bad
block info bytes and those followed by the interleave bytes)
The Adaptec card responds to the missing info, by trying to read
the current format from the drive and then reformatting to the same
tune. (If it cannot read it it will format as a 5MB default drive.)
So, because I once formatted with map ON and interleave of 1, the
drive was interleaved at 1 yielding only 55KBytes/sec reads. My
subsequent attempts to change interleave to 2 or 3 had no effect,
as the map option was turned off. This was true of both the old
and new revision format programs. (Even when I finally got the
old version format to stop GURUing and run, it did not return my
drive to its former performance level!! That's when I got REALLY
bent!!)
Having discovered this, I formatted with map ON and interleave
of 3 last night. Now I have 150KBytes/sec reads. There may be
room to further optimize through buffers etc, I will have to play
with it.
With map ON, the SupraFormat program carefully builds
the bad block info bytes and interleave bytes and sends them out
after the format commmand, as it should.
Bottom line is a 10% improvement with series II software so far,
in spite of the content free documentation that accompanied it.
I'm trying to decide if I should pay for another phone call to Supra
and let them know how their format program "really" works.
| |||||