mg32+@andrew.cmu.edu (Michael Ginsberg) (08/30/89)
Just a note: One reason why you'd want to turn off concurrent DMA (read: the hard drive is being read while the floppy writes) is that, while real IBM's and good clones have seven channels, all dependable, many clones have only one, or seven with only one really being proven in. I have seen many people make incrementals daily and full backups weekly only to find all the backups are sludge. One thing I would suggest to everyone is to make a full and an incremental backup, and then restore it (just like everyone should pull the plug on a UPS before they assume it'll handle glitches as you write to the hard drive... but that's another story). Another thing, many people find that the dma works at a slower speed... I had an AT clone that did perfect backups at 8 mhz, but at 12 it failed miserably (of course never telling you that it was dying). Many backup programs (such as FastBack Plus, but I don't really trust this program in general so...) will test your system for you, trying to make backups and restore at various DMA speeds, and it then recommends to you one of the speeds... But after all this rambling, MAKE SURE you check to see that you can really restore your backups before depending on them! --Michael Ginsberg Carnegie Mellon University, Pittsburgh, Pennsylvania, USA ARPA: mg32+@andrew.cmu.edu BITNET: mg32%andrew@cmccvb UUCP: ...!harvard!andrew.cmu.edu!mg32+
burkett@csd4.csd.uwm.edu (Edward W Burkett) (08/30/89)
In article <8Yym5V200WB58jE1Az@andrew.cmu.edu> mg32+@andrew.cmu.edu (Michael Ginsberg) writes: >Just a note: One reason why you'd want to turn off concurrent DMA (read: >the hard drive is being read while the floppy writes) is that, while real >IBM's and good clones have seven channels, all dependable, many clones have >only one, or seven with only one really being proven in. . . >Another thing, many people find that >the dma works at a slower speed... I had an AT clone that did perfect backups >at 8 mhz, but at 12 it failed miserably (of course never telling you that it >was dying). Many backup programs (such as FastBack Plus, but I don't really >trust this program in general so...) will test your system for you, trying to >make backups and restore at various DMA speeds, and it then recommends to you >one of the speeds... I have one of those clones (LE Model D) that have a bad DMA. I have been using FastBack Plus for sometime now and it seems to be a very reliable program. When I test the DMA, I find that I can only make backups at the slow speed :-(! I would like to replace the defective chip but I need to know more about its location. Is the DMA chip part of the mother board, floppy disk controller or harddisk controller. I assume it is on the mother board. Secondly, can it be easily purchased and replace by the owner or would it have to be put in by a repair technician? Any help would be greatly appreciated. ******************************************************************************** burkett@csd4.csd.uwm.edu (Edward W Burkett) University of Wisconsin -- Milwaukee Department of Biological Sciences ********************************************************************************
bb16@prism.gatech.EDU (BOSTATER,Scott) (08/30/89)
In article <4048@csd4.csd.uwm.edu>, burkett@csd4.csd.uwm.edu (Edward W Burkett) writes: > I have one of those clones (LE Model D) that have a bad DMA. [stuff deleted] > I would like to replace the defective chip but I need to know more about its > location. Is the DMA chip part of the mother board, floppy disk controller or > harddisk controller. I assume it is on the mother board. Secondly, can it be > easily purchased and replace by the owner or would it have to be put in by a > repair technician? > The DMA chip is on the motherboard. (At least it is on the original PC/XT/AT's and most clones). What you would want to replace is the Intel 8237 High Performance Programmable DMA Controller chip. Chances are that its solderd directly to the motherboard. Its a 40 pin chip so unless you have the proper hardware to desolder that large a chip, a PC technician might be a good investment. You could cut the leads off the chip and unsolder each pin individually, but that topic has already been discussed in comp.hardware.mutilation :) As far as the location of the chip, It should be the only 8237 on the board, happy hunting :) -- Scott Bostater GTRI/RAIL/RAD (Ga. Tech) "My soul finds rest in God alone; my salvation comes from Him" -Ps 62.1 uucp: ...!{allegra,amd,hplabs,ut-ngp}!gatech!prism!bb16 Internet: bb16@prism.gatech.edu
axaris@cs.buffalo.edu (Vassilios Axaris) (08/31/89)
Hello, My unsolved problem is that when FastBack Plus performs the DMA test, it fails completely on my 10/6 MHz AT compatible (by Mitac). When I say it fails, I mean not even the slow DMA speed test is passed. Any hints on what is going wrong? Vassilios.
mg32+@andrew.cmu.edu (Michael Ginsberg) (08/31/89)
That's an easy one... The DMA (which, incidentally stands for Direct Memory Access) chip is on the motherboard. If you have a 80286 or '386 system, you're in trouble. If you have a 8086 or 8088 (read: PC or XT clone) you can easily get a replacement fairly cheaply (I think at most 20 bucks, if not much cheaper). If you want to save money, at the expense of time, get a computer magazine with lots of electronics adds (such as Computer Shopper) and order a chip. If you're lazy and don't mind paying whatever they feel like charging you, the makers of Fastback plus (and if you read the manual you'd know this) offer a "DMA upgrade" deal where they'll sell you a replacement chip. NOTE: THIS chip is only good in 8086/8088 systems. If you don't have their phone number, and can't get it through information (if you own the software, it's in their docs, and the bit about the DMA chip, is, if I can recall, in the appendix of the docs), write back and I'll dig it up. Other than that, if you've got a few bucks (and the clone-makers haven't done something really mean like solder the poor chip onto the board (in which case you've got to desolder it) it's a five minute job. --Michael Ginsberg Carnegie Mellon University, Pittsburgh, Pennsylvania, USA ARPA: mg32+@andrew.cmu.edu BITNET: mg32%andrew@cmccvb UUCP: ...!harvard!andrew.cmu.edu!mg32+ Good luck!
karl@ddsw1.MCS.COM (Karl Denninger) (08/31/89)
>----- >Response 3 of 3 (5210) by axaris at cs.buffalo.edu on Thu 31 Aug 89 01:36 >[Vassilios Axaris] (7 lines) > >Hello, > >My unsolved problem is that when FastBack Plus performs the DMA test, it fails >completely on my 10/6 MHz AT compatible (by Mitac). When I say it fails, I >mean >not even the slow DMA speed test is passed. Any hints on what is going wrong? Your DMA controller is badly hosed. Yell at the place you got the system from and demand satisfaction. Our clones easily pass the test at high speed.... and backup flawlessly with FB+! We use the package in-house, and highly recommend it to customers. I have _never_ lost a file due to backup set corruption, although we have had several disks that have had bad spots -- which would have rendered the backup useless with other backup tools. -- 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"
andyross@ddsw1.MCS.COM (Andrew Rossmann) (09/01/89)
(About Fastback + failing on even slow DMA:) The computer I use at work (a CompuAdd 386/20) also fails immediately on the test. Nevertheless, I manually set the DMA to fast. I have backed up and restored this disk with no problems. Could it be the memory cache or the 1:1 interleave controller? Andrew Rossmann andyross@ddsw1.MCS.COM
burkett@csd4.csd.uwm.edu (Edward W Burkett) (09/01/89)
A little information about Clones, DMA and FastBack+.
I have followed up on a suggestion to contact Fifth Generation Systems, Inc.
about DMA chips that do not pass the *fast* FastBack+ test. After talking to
a Tech. at FGS, Inc. I learned that there are several clones (Leading Edge
is one) and (believe it or not) IBM's that do not have defective DMA chips but
rather the architecture of the system limits the access to the DMA chip by
a clock or timing problem. Well, those of you who understand explain it to
the rest of us.
As it turns out, if you have an 8086 or 8088 machine, FGS, Inc. supplies a
chip that will fix the timing problem. To install this chip you simply
pull the original CPU, insert this 40 pin chip and then piggyback the original
CPU (or as they suggested a V20) onto the new chip. The Tech. assured me
that this would fix the DMA problem for any 8086/8088 machine and does not
influence the performance of the machine in any other way. Where they get
you is that the chip cost $40.00 which includes shipping.
Does anybody know anything about this chip? The reason I ask is that FGS, Inc.
also suggested replacing the 8088 with a V20. They will sell it to you for
$30.00. Well, I can buy one for $11.00!!!! Hmmmm! Maybe the DMA fix chip
is also available at a lower cost from a different source.
An interesting point was brought out in my discussion with FGS, Inc.
Since my original posting, I have gotten MANY replies that the DMA problem was
actually a defective DMA chip and that it simply (or not so simply, if it is
soldered) had to be replaced. One knowledgable person said that it was
due to a slow machine. This was why I called FGS, Inc. The point is
--- investigate a problem like this thoroughly before you spend a LOT of money
on someones advice that may not fix the problem---.
For those of you who are interested, you do not have to own FastBack to buy this
DMA fix chip. They will sell it to anyone.
FGS, Inc. address is:
Fifth Generation Systems, Inc.
11200 Industriplex Boulevard
Baton Rouge, LA 70809
The chips' part number is 132000
Include a check for $40.00 and a return address.
If you would like to talk to them, their phone number is 504-291-7283.
P.S. (They also said.. DO NOT force a faster speed than FB+ says is ok because
you will LOSE data....maybe not the first time....but you will lose data)
HOPE THIS HAS BEEN HELPFUL!
********************************************************************************
Ed Burkett burkett@csd4.csd.uwm.edu (Edward W Burkett)
University of Wisconsin
Department of Biological Sciences
********************************************************************************