rwm@atronx.UUCP (Russell McOrmond) (02/23/90)
Hi,
Seems a message to CBMVAX bounced (This is the first time), but I assume
it would go if sent a second time - I think this is best posted public anyways.
I have a friend that has a DC WREN II which is a 155M 16.5ms access drive. This
drive was hooked up to a 2090 for a few years, a 2090a for a couple of months,
and is now working on a 2091 controller.
The upgrade to the 2091 controller was made in order to get a speed increase
with access to the Drive. This did not seem to happen with this unit, and
now I question anyone out there (Expecially Dave H) as to any possibilities
why, and any possible fixes.
While it is no definitive test, we used DiskPerf3 on this drive, on an
empty partition, with all 3 controllers. I will show the figures given for
three different tests.
2090A - 140MPartition, FFS
Create : 15 Delete 45
Directory Scans : 92 entries/sec
Seek/Read test : 97 seek/reads per second
R/W speed: Buf 1K Rd 57508 wr 53137
Buf 8K Rd 308404 wr 181833
Buf 32K Rd 455902 wr 251658
2090A, 10M partition, OFS
Create : 10 Delete 27
Directory Scans : 40 entries/sec
Seek/Read test : 68 seek/reads per second
R/W speed: Buf 1K Rd 40124 wr 17672
Buf 8K Rd 47662 wr 21283
Buf 32K Rd 48695 wr 21952
2091, 100M partition, FFS
Create : 12 Delete 29
Directory Scans : 104 entries/sec
Seek/Read test : 53 seek/reads per second
R/W speed: Buf 1K Rd 49617 wr 46125
Buf 8K Rd 218453 wr 174762
Buf 32K Rd 499321 wr 296766
As these 'indicate' (Yes, I realize that DiskPerf values are kinda like
'MIPS' values - Doesn't show much to anyone) the 2091 did not help with actual
seeking (Much lower value than 2090a, and only showed an increase in transer
rate when working with much larger buffers (P.S. the 'Buffers' for the partition
itself were all equal between these tests).
Now, I have spoken to other people with the A2091, and they have mentioned
RADICALLY higher ratings with the same drive moved from the 2090a - is this
something that could be specific to the CDC drive? Any thoughts would be
appreciated.
:Later
--
Opinions expressed in this message are my Own. My Employer does not even
know what these networks ARE.
Russell McOrmond rwm@atronx fts1!atronx!rwm alzabo!atronx!rwm
FidoNet 1:163/109 [(613) 231-7144 ] BBS: (613) 230-2282
Amiga-Fidonet Support 1:1/109doug@xdos.UUCP (Doug Merritt) (02/24/90)
In article <2032.AA2032@atronx> rwm@atronx.UUCP (Russell McOrmond) writes:
:
: I have a friend that has a DC WREN II which is a 155M 16.5ms access drive. This
:drive was hooked up to a 2090 for a few years, a 2090a for a couple of months,
:and is now working on a 2091 controller.
:
: [ drive is very slow w/ 1K blocks, very fast w/ huge blocks ]
:
: Now, I have spoken to other people with the A2091, and they have mentioned
:RADICALLY higher ratings with the same drive moved from the 2090a - is this
:something that could be specific to the CDC drive? Any thoughts would be
:appreciated.
The Wrens are pretty fast, well made drives; I doubt very much that a 40K
transfer rate has anything to do with the design of the Wren.
But it could easily be that he's using an inappropriate interleave factor.
E.g. a 4:1 interleave factor might show exactly this kind of performance
difference. Reformat with 1:1 and retest. And also with 2:1, to compare.
Doug
--
Doug Merritt {pyramid,apple}!xdos!doug
Member, Crusaders for a Better Tomorrow Professional Wildeyed Visionarylphillips@lpami.wimsey.bc.ca (Larry Phillips) (02/25/90)
In <05034.AA05034@alison>, mjl@alison.UUCP (Martin J. Laubach) writes: > But I have a different question, concerning the A590: has anybody >succeeded in using the A590 with some other SOTS peripheral? We tried >it here with different brands of memory expansion boards -- but the A590 >always locked up the system. > And: please don't tell me that "Commodore never supported more than one >expansion device on the A500" -- I'm using it on an A1000... You have answered your own question, at least in terms of support. CBM supported one PIC (Plug In Card) on the expansion bus. That you were able to run more than that is a fact, but does not change the fact that not only was it unsupported, but that not all combinations of peripherals would run on all A1000s. I have known people who never were successful in getting more than one hng on there, and others who have hung 3 or 4 out there with no problems. You can try multiple peripherals, but when you enter this realm, you are completely on your own. -larry -- Gallium Arsenide is the technology of the future; always has been, always will be. +-----------------------------------------------------------------------+ | // Larry Phillips | | \X/ lphillips@lpami.wimsey.bc.ca -or- uunet!van-bc!lpami!lphillips | | COMPUSERVE: 76703,4322 -or- 76703.4322@compuserve.com | +-----------------------------------------------------------------------+
mjl@alison.UUCP (Martin J. Laubach) (02/25/90)
> The upgrade to the 2091 controller was made in order to get a speed > increase with access to the Drive. This did not seem to happen with this > unit, and now I question anyone out there as to any possibilities why, > and any possible fixes. I don't think the 2091 (or 590) is faster than the 2090 -- my experiences have been that the 2090 outperforms a 2091, especially on small data transfers (the 2090 has an additional Z-80 doing all the DMA and SCSI setups, while on the 2091 all that stuff is directly handled by the main processor). On large transfers (>64K), the 2090 still is a bit faster than the 2091 (something like 650 vs. 500 K/s). But I have a different question, concerning the A590: has anybody succeeded in using the A590 with some other SOTS peripheral? We tried it here with different brands of memory expansion boards -- but the A590 always locked up the system. And: please don't tell me that "Commodore never supported more than one expansion device on the A500" -- I'm using it on an A1000... mjl // Usenet: mjl@alison.at or ..uunet!mcsun!tuvie!alison!mjl \X/ Fido: 2:310/3.14
stan@teroach.UUCP (Stan Fisher) (02/27/90)
In article <05034.AA05034@alison> mjl@alison.UUCP (Martin J. Laubach) writes: > > I don't think the 2091 (or 590) is faster than the 2090 -- my >experiences have been that the 2090 outperforms a 2091, especially on >small data transfers (the 2090 has an additional Z-80 doing all the DMA >and SCSI setups, while on the 2091 all that stuff is directly handled by >the main processor). On large transfers (>64K), the 2090 still is a bit >faster than the 2091 (something like 650 vs. 500 K/s). > > mjl // Usenet: mjl@alison.at or ..uunet!mcsun!tuvie!alison!mjl > \X/ Fido: 2:310/3.14 I beg to differ. It's pretty hard to make a blanket statement about which controller is faster _period_. Example: You've seen situations that make you believe the 2090 is faster than the 2091. I dropped a 2091 into a 2500/20 in place of a 2090A and compared diskperfs with virgin fresh formats on a Micropolis 158 meg SCSI drive and saw speed improvments of around 20% overall. This was the exact same hardware except for the 2091/2090A swap, both to the same drive with a fresh format. I really can't say which controller handles a well used/fragmented partition with lot of data on it. So to me.. The 2091 is decently faster. Around 950K (max) diskperfs on my setup. Stan Fisher - stan@teroach.phx.mcd.mot.com - asuvax!mcdphx!teroach!stan Motorola Microcomputer Division, Tempe, Arizona - Voice (602) 438-3228 Call our User Group BBS "M.E.C.C.A." running Atredes 1.1 @ (602) 893-0804