[Search for users]
[Overall Top Noters]
[List of all Conferences]
[Download this site]
Title: | IBM PCs, clones, DOS, etc. |
Notice: | Intro in 1-11, Windows stuff in NOTED::MSWINDOWS please |
Moderator: | TARKIN::LIN ND |
|
Created: | Mon Jan 02 1995 |
Last Modified: | Fri Jun 06 1997 |
Last Successful Update: | Fri Jun 06 1997 |
Number of topics: | 3023 |
Total number of notes: | 28404 |
2938.0. "Upgrade Processor/Caching Question" by SMURF::TOMG () Mon Mar 03 1997 09:43
Over the weekend, I bought one of those "processor upgrade" kits
for 486 systems. The kit included an AMD 586 133Mhz processor and
voltage converter. The user simply replaces their existing CPU
with the whole package (CPU and voltage converter).
Anyway, installation was a breeze. Just take out the old
processor and replace it with the new one. This part was fine.
However, after the system memory size check, the system would
just hang. I traced this down to something to do with the
external cache. That is, disabling the external cache allowed the
system to boot just fine. Replacing the old CPU and enabling
cache also seems to work. It didn't matter whether the CPU
internal cache was enabled or not, enabling external cache with
the 586 present failed.
On the underside of the new CPU package, there is a switch pack
that allows setting of various clock multiplier and cache
strategies. For example, I could have any of the following
caching strategies at either 4x multiplier or 3x multiplier:
- pass through (use caching strategy of motherboard)
- write through
- write back
By default the configuration is 4x, pass through.
I tried all combinations without success. Using, I think, write
back allowed the system to add least start booting windows95, but
the system would hang at the windows startup splash screen.
If I set the multiplier to 4x, the chip was identified as
something like 80486DX2 100mhz, with 100mhz being the highest
speed the motherboard could report. At 3x multiplier, the system
reported the chip as DX4-S 100mhz, which looks ok, since at this
multiplier the speed should be ~100mhz.
Unfortunately, I don't have the motherboard documentation, so I
don't know if there is something that I could change on the
motherboard to make this work.
I am probably going to return the upgrade package, but I'm
interested in maybe finding out why this wouldn't work.
The system hardware configuration is:
486 motherboard (30 pin memory, 256k cache, vlb/isa)
Intel 486 66mhz processor
32mb memory
The motherboard is an older one where the BIOS doesn't support
LBA. Offhand, I would say it's probably 1994 or so.
It has the usual complement of hardware cards, including multi
function I/O, sound card, modem, etc.
Tom
---
Dictated using DragonDictate.
T.R | Title | User | Personal Name | Date | Lines |
---|
2938.1 | Chance occurance | JULIET::HARRIS_MA | Networks Sales Exec | Mon Mar 03 1997 12:53 | 13 |
| I don't think you are going to find an answer to WHY it didn't work.
I have heard about a dozen people tyry this avenue. Half failed, half
succeeded. There are also 5 or 6 different 486/586-133 upgrade kits on
the market (each a little different).
You *may* find that a different upgrade kit works. You may not. It
appears to be luck of the draw/chance whether you can get it working.
A common thread which seems to be noticed across the users who have
tried:
The newer the motherboard, the more luck you'll have.
Mark
|
2938.2 | wait state | SSDEVO::ASTOR | Subsystems Engineering Support | Mon Mar 03 1997 12:59 | 6 |
| Hi,
Is there a way to give the cache or memory a wait state or two?
Just a thought.
Kurt
|
2938.3 | | SMURF::TOMG | | Mon Mar 03 1997 14:58 | 13 |
|
re: .-1
I thought about that. The system is currently running with 1ws (I
think) for cache. Changing this to a higher number didn't help.
Good suggestion, though. :-)
Tom
---
Dictated using DragonDictate.
|
2938.4 | How Can One Tell? | EVMS::KRSNA::DKOSKO | Oh Lord, won't you buy me... | Mon Mar 10 1997 11:51 | 8 |
| This is most interesting since I'm currently running a similar system to the one
Tom has and have been considering the same upgrade strategy. The question is,
"how does one go about determining if the motherboard/bios combo in the existing
system will support the upgrade?"
Or is it simply trial and error?
dave
|