silver@cup.portal.com (Jim B Howard) (06/06/89)
Is there any way to speed up this Starboard SCSI controller of mine? The best rates I can get from it are 53k per second according to diskperf and autotest. I called Microbotics and they told me, "sorry, its not much faster than your floppy drive." I said, "you didnt mention that when you sold it to me" Is the problem with the driver? or is it just the miserable hardware? Has anyone found a way to speed this thing up? My drive is no problem , its a 277N (40 ms). Definatly faster than 50k/sec.
mikes@lakesys.UUCP (Mike Shawaluk) (06/07/89)
In article <19142@cup.portal.com> silver@cup.portal.com (Jim B Howard) writes: >Is there any way to speed up this Starboard SCSI controller of mine? >The best rates I can get from it are 53k per second according to >diskperf and autotest. I called Microbotics and they told me, >"sorry, its not much faster than your floppy drive." You don't say which version of AmigaDOS you are using, and/or which file system (original or Fast), if you are using AmigaDOS 1.3. I am assuming that the Starboard controller supports FFS, but since I don't own one, I may be wrong... Anyways, if you're not using the FFS, then you oughta be! That is, if you can... I have the same drive as you do (ST277N), except with a Supra 4x4 interface, and have the entire thing set up as FFS partitions (unfortunately on a 1000, I can't autoboot. :-( -- - Mike Shawaluk (mikes@lakesys.lakesys.com OR ...!uunet!marque!lakesys!mikes) "Where were you on the night of August 12?"
riley@batcomputer.tn.cornell.edu (Daniel S. Riley) (06/08/89)
In article <19142@cup.portal.com> silver@cup.portal.com (Jim B Howard) writes: >Is there any way to speed up this Starboard SCSI controller of mine? >The best rates I can get from it are 53k per second according to >diskperf and autotest. I called Microbotics and they told me, >"sorry, its not much faster than your floppy drive." >My drive is no problem , its a 277N (40 ms). Definatly faster than 50k/sec. On a decent sized transfer, I get around 100K per second with the Starboard (Stardrive) SCSI controller and an ST225N (which is a *slow* drive). 50K/sec is well out of line. The first obvious question is, are you using FFS? If not, get it. Assuming you are, what interleave did you format to? Most of the Seagate drives seem to prefer an interleave of at least 2, and the slower ones (like mine) might be better off at 3. Note that changing the interleave value in the mountlist does nothing for a SCSI drive--you have to redo the low-level format completely. -Dan Riley (riley@tcgould.tn.cornell.edu, dsr@lns61.tn.cornell.edu) -Wilson Lab, Cornell U.
pmf@mimsy.UUCP (Paul M. Franceus) (06/08/89)
There is a way to speed up the StarBoard interface - it's simple. Just low- level format it with an interleave of 3(or 2 maybe) instead of the default of 0. A customer of mine gets 187k off his ST277N with this setup. Paul StarLight Technologies
berry@stsci.EDU (Jim Berry) (06/08/89)
From article <19142@cup.portal.com>, by silver@cup.portal.com (Jim B Howard): > Is there any way to speed up this Starboard SCSI controller of mine? > The best rates I can get from it are 53k per second according to > diskperf and autotest. I called Microbotics and they told me, What interleave value did you use when you did the low-level format? Try 3. I'm getting 150-200K/reads with the same setup that you have (no, it's not blinding, but the controller only cost $90...) Good luck. -- ------------------------------------------------------------------------------ Jim Berry | UUCP:{arizona,decvax,hao}!noao!stsci!berry Space Telescope Science Institute | ARPA: berry@stsci.edu Baltimore, Md. 21218 | SPAM: SCIVAX::BERRY, KEPLER::BERRY
silver@cup.portal.com (Jim B Howard) (06/10/89)
My drive *WAS* low level formatted with a value of 2, and I am using Amiga-Dos 1.3 and the FFS. I still only get about 65k per second. I have tried interleave of 3, and it doesnt change the speeds in any direction but down. I talked to Microbotics for quite some time last nite, and they couldnt help me. Someone PLEASE tell me they had this exact same problem and have figured out the solution.
rob.sudbury@canremote.uucp (ROB SUDBURY) (06/12/89)
Try throwing an Addbuffer in your startup-sequence for the volumes oin your hard drive. I have a Comspec A2000 controller running on my A500 with a Subsystem expansion chassis and I average 120K/sec reads, my system is down right now while I have a charcoaled mother board replaced, and I hope to add a Addbuffers to my startup (with 1.3 WHEE) to increase my speed further. Supposedly the Comspec controller I have is capable of about 270k/sec(I even heard a figure of 390k/sec). Comspec is in Toronto, ON. RS * QNet 1.03a2: CONNECTed II 'THE BBS' - Peterborough, Can. (705) 743-7296.
karl@ddsw1.MCS.COM (Karl Denninger) (06/16/89)
>Response 5 of 5 (206) by silver at cup.portal.com on Mon 12 Jun 89 05:40 >[Jim B Howard] >My drive *WAS* low level formatted with a value of 2, and I am >using Amiga-Dos 1.3 and the FFS. I still only get about >65k per second. I have tried interleave of 3, and it doesnt change >the speeds in any direction but down. That performance SUCKS. I regularly get 650K/second on IBM clones (ATs) using a track cache controller. What the hell is Commodore and these other manufacturer's doing wrong if they can't even beat an ST506 (slow as a turtle) interface?! And I thought that Amigas were faster than IBM's.... guess not when it comes to disk performance. I'm greatly disappointed. Can others send or post their drive performance figures for me? I'd like to see what ELSE is out there. -- Karl Denninger (karl@ddsw1.MCS.COM, <well-connected>!ddsw1!karl) Public Access Data Line: [+1 312 566-8911], Voice: [+1 312 566-8910] Macro Computer Solutions, Inc. "Quality Solutions at a Fair Price"
mp1u+@andrew.cmu.edu (Michael Portuesi) (06/21/89)
I still have another 500 or so messages ahead of me, but I figured I would answer this one before I forgot about it. silver@cup.portal.com (Jim B Howard) writes: > My drive *WAS* low level formatted with a value of 2, and I am > using Amiga-Dos 1.3 and the FFS. I still only get about > 65k per second. I have tried interleave of 3, and it doesnt change > the speeds in any direction but down. > I have a Starboard with a ST-157N. Running Diskperf on a nearly empty partition give me read values of about 190K/sec. Microbotics includes a command called FASTMODE (or some such name; I can't remember now and I'm typing this on a MacII instead of my Amiga) that pulls some shortcuts on the SCSI protocol and speeds up transfers by a factor of two or so. The only restriction is that you cannot be in FASTMODE when you low-level format the drive. You can also specify FASTMODE directly from the mountlist. You have to change the Flags= value to 12336 (which is from memory and could be wrong -- I'll check the next time I'm in front of my Amiga). --M -- Michael Portuesi * Carnegie Mellon University INTERNET: mp1u+@andrew.cmu.edu * BITNET: mp1u+@andrew UUCP: ...harvard!andrew.cmu.edu!mp1u+ MAIL: Carnegie Mellon University, P.O. Box 259, Pittsburgh, PA 15213
scotth@sgi.com (Scott Henry) (06/22/89)
I just want to add to the list of "successfull" Stardrive owners. Here are
my fastest and slowest (out of 5) diskperf timings for my setup:
File create/delete: create 9 files/sec, delete 20 files/sec
Directory scan: 68 entries/sec
Seek/read test: 57 seek/reads per second
r/w speed: buf 512 bytes, rd 52428 byte/sec, wr 22405 byte/sec
r/w speed: buf 4096 bytes, rd 90394 byte/sec, wr 67216 byte/sec
r/w speed: buf 8192 bytes, rd 113975 byte/sec, wr 74898 byte/sec
r/w speed: buf 32768 bytes, rd 131072 byte/sec, wr 87381 byte/sec
File create/delete: create 8 files/sec, delete 27 files/sec
Directory scan: 86 entries/sec
Seek/read test: 70 seek/reads per second
r/w speed: buf 512 bytes, rd 54613 byte/sec, wr 23616 byte/sec
r/w speed: buf 4096 bytes, rd 93622 byte/sec, wr 77101 byte/sec
r/w speed: buf 8192 bytes, rd 113975 byte/sec, wr 87381 byte/sec
r/w speed: buf 32768 bytes, rd 137970 byte/sec, wr 100824 byte/sec
I am running a very early (October '85) Amiga 1000 (AmigaDos 1.3 w/FFS)
with 2MB Starboard-II and Stardrive SCSI. I am using an Adaptec 4070 and a
30MB (RLL) Miniscribe 8436 (? I forget the exact model number), 65ms
average access time. Here is an extract of my mountlist for the tested
partition:
/* MountList for V1.3 */
DH1: Device = StarDrive.device
FileSystem = l:FastFileSystem
Unit = 0
Flags = 0x03030
Surfaces = 4
BlocksPerTrack = 26
Reserved = 2
Interleave = 0
LowCyl = 0 ; HighCyl = 74
Buffers = 13
BufMemType = 5
GlobVec = -1
Priority = 10
Mount = 1
DosType = 0x444F5301
StackSize = 4000
#
Just for comparison purposes, here are the diskperf results for my
workstation at work (an IRIS 4D70G with 380MB ESDI drives):
File create/delete: create 11 files/sec, delete 38 files/sec
Directory scan: 1563 entries/sec
Seek/read test: 622 seek/reads per second
r/w speed: buf 512 bytes, rd 426828 byte/sec, wr 496171 byte/sec
r/w speed: buf 4096 bytes, rd 486202 byte/sec, wr 587986 byte/sec
r/w speed: buf 8192 bytes, rd 475904 byte/sec, wr 543303 byte/sec
r/w speed: buf 32768 bytes, rd 509842 byte/sec, wr 569878 byte/sec
All of the diskperf results were taken with my normal working
configuration. On the Amiga, this is a very non-standard workbench setup:
it includes interlace/overscan, dmouse, dcron (for amiga/uucp), Wshell/
Arexx, ARP, other stuff that I don't remember right now. My one real
complaint with the Stardrive software is that it Forbid()'s too much when
doing disk I/O.
---
Scott Henry <scotth@corp.sgi.com>
** I don't care whether the company likes my opinions or not, they're mine! **
--
Scott Henry <scotth@sgi.com>
#include <std_disclaimer.h>silver@cup.portal.com (Jim B Howard) (06/23/89)
To the person who recommended using Fastmode, believe me, I have. Still no results better than 70k per second.
nix@tolsun.oulu.fi (Tero Manninen) (06/27/89)
Maybe this isn't much related to previous posts in this subject, but the posted MountList for a hardisk reminded me some problems with my own hackings.. I was looking for a good backup program for my 30M Supra HD and decided to use MRBackup 2.4 (didn't care about the slowness). Because it uses standard AmigaDOS disk format the usual 837k of OFS disks seemed so small, I made a MountList for FF1: (same as DF1: but uses FastFileSystem). Now everyting seemed to run well, but after backup I checked the contents of one backupped file (type opt h file.Z) and found that it contained only carbage :-( So, what could went wrong ? I suspect BufMemType, but what do all these numbers mean ? (Chip, Fast, All) Are there some other types ?? E.g. in the posting that I followuppedthere were a number 5 in this field :-? Thanks for any help. Tero Manninen BTW, I got the same garbage results with Matts Backup.
new@udel.EDU (Darren New) (06/28/89)
In article <672@tolsun.tut.fi> nix@tolsun.oulu.fi (Tero Manninen) writes: >Now everyting seemed to run well, but after backup I checked the contents >of one backupped file (type opt h file.Z) and found that it contained >only carbage :-( The file probably did not contain garbage. Since the file ended in .Z you must have backed up with the compression on. The compressed file will indeed look like garbage until you decompress it. Find a file without the .Z (I think MRBackup does not compress one-block files). Especially look for a .info, which is usually small. Either that or redo the backup with the compression turned off. -- Darren
nix@tolsun.oulu.fi (Tero Manninen) (06/29/89)
In article <18597@louie.udel.EDU> new@udel.EDU (Darren New) writes: ->In article <672@tolsun.tut.fi> nix@tolsun.oulu.fi (Tero Manninen) writes: ->>Now everyting seemed to run well, but after backup I checked the contents ->>of one backupped file (type opt h file.Z) and found that it contained ->>only carbage :-( -> ->The file probably did not contain garbage. Since the file ended in .Z ->you must have backed up with the compression on. The compressed file will ->indeed look like garbage until you decompress it. Find a file without the ->.Z (I think MRBackup does not compress one-block files). Especially look ->for a .info, which is usually small. Either that or redo the backup with ->the compression turned off. -- Darren Ah, but those compressed *.Z files contained big chunks of zeroes and pars of another backupped files. I found this out by looking one script file (my S:Startup-Sequence to be exact). I thought compress would shrink those big fields of same bytes rather than leave those alone. The thing that I would like to get is now the correct MountList for a FFS floppy. Tero Manninen
jesup@cbmvax.UUCP (Randell Jesup) (07/01/89)
In article <[206.6]comp.amiga;1@ddsw1.MCS.COM> karl@ddsw1.MCS.COM (Karl Denninger) writes: >>Response 5 of 5 (206) by silver at cup.portal.com on Mon 12 Jun 89 05:40 >>[Jim B Howard] >>My drive *WAS* low level formatted with a value of 2, and I am >>using Amiga-Dos 1.3 and the FFS. I still only get about >>65k per second. I have tried interleave of 3, and it doesnt change >>the speeds in any direction but down. > >That performance SUCKS. What do you expect using a absolute cheapest and simplest controller that exists? >I regularly get 650K/second on IBM clones (ATs) using a track cache >controller. What the hell is Commodore and these other manufacturer's doing >wrong if they can't even beat an ST506 (slow as a turtle) interface?! A real controller (A2091/A590, Hardframe, GVP, even A2090) can do at least 650K/sec READING FROM A FILE (not raw device speed). With a good SCSI disk, some of them can do >1Meg/sec from a file, >1.4Meg/sec raw. This is without expensive track caches on the controller. I suspect your 650K/sec is mostly for reading from the cache, not actually from the disk. If you simply read a 5Meg file sequentially, what speed do you get? >And I thought that Amigas were faster than IBM's.... guess not when it comes >to disk performance. They are, especially when you look at normal PC controllers (most can't do more than 200K/sec no matter what). -- Randell Jesup, Commodore Engineering {uunet|rutgers|allegra}!cbmvax!jesup