Anonymous Coward
Veteran Member
Maybe it's just my imagination, but is the cache memory really 35ns? That seems awfully slow for a 486. I think the slowest I've ever seen is 25ns.
- VCF South West - June 14 - 16, Davidson-Gundy Alumni Center at University of Texas at Dallas
- VCF West - Aug 2 - 3, Computer History Museum, Mountain View, CA
- VCF Midwest - Sept 7 - 8 2024, Schaumburg, IL
- VCF SoCal - Mid February 2025, Location TBD, Southern CA
- VCF East - April 2025, Infoage Museum, Wall NJ
-
Please review our updated Terms and Rules here
Getting back into older PC's with my first 486 Build in almost 30 years
- Thread starter VERAULT
- Start date
VERAULT
Veteran Member
From the first photo in this thread of the mainboard? Well thats how this board came I honestly couldnt tell you otherwise.
maxtherabbit
Veteran Member
could you run CACHECHK and let us know what the access times are like? I'd be curious to know what timings that memory was setup for
VERAULT
Veteran Member
This? : https://www.philscomputerlab.com/dos-benchmark-pack.html
When I get a few parts I ordered in and have time to get back to the machine.. Sure I can do that.
Anonymous Coward
Veteran Member
I checked the part numbers for those cache chips. They really are 35ns.
For a 33MHz 486, that's pretty rough. Normally at 33MHz, nothing slower than 20ns is used.
It would be unfortunate to alter the original condition of this board, but I would think you could get a pretty significant speed boost by swapping in some 15ns parts and adjusting BIOS timings accordingly.
MS62256A - Mosel Vitelic Corporation Datasheet PDF VIEW DOWNLOAD - DATASHEETBANK.COM
MS62256A Datasheet : 32K x 8 High Speed CMOS Static RAM, MS62256A PDF VIEW Download Mosel Vitelic Corporation , MS62256A 1 page Datasheet PDF, Pinouts, Data Sheet, Equivalent, Schematic, Cross reference, Obsolete, Circuits
www.datasheetbank.com
For a 33MHz 486, that's pretty rough. Normally at 33MHz, nothing slower than 20ns is used.
It would be unfortunate to alter the original condition of this board, but I would think you could get a pretty significant speed boost by swapping in some 15ns parts and adjusting BIOS timings accordingly.
VERAULT
Veteran Member
Im willng to try, Gotta a source for those 15ns cache chips?
maxtherabbit
Veteran Member
Don't get ahead of yourself, get some numbers first. Cache timings aren't always adjustable, and if they're not you would be spending money for nothing. Plus sometimes the chips you have will run faster than their rated speed without issue.
Anonymous Coward
Veteran Member
It's easy enough to know if the cache timings can be adjusted or not...just check in the CMOS setup. It's not like we're dealing with exotic stuff here. AMI Color BIOS and OPTi 491 are pretty well understood.
I used to buy cache memory on eBay, but this was many years ago. People I talk to tell me that these days eBay is full of Chinese fakes. It might be easier/cheaper to salvage the chips from another board.
Maybe those 35ns chips are very good quality and work on really tight cache timings. But if they only work at their rated specs, I would think it would be not much better than running cacheless. Even my 20MHz 386sx with cache came with 25ns chips.
I'd be really interested to see some benchmarks on your system.
I used to buy cache memory on eBay, but this was many years ago. People I talk to tell me that these days eBay is full of Chinese fakes. It might be easier/cheaper to salvage the chips from another board.
Maybe those 35ns chips are very good quality and work on really tight cache timings. But if they only work at their rated specs, I would think it would be not much better than running cacheless. Even my 20MHz 386sx with cache came with 25ns chips.
I'd be really interested to see some benchmarks on your system.
VERAULT
Veteran Member
Ill have an update on this within the next couple hours. I just need to finish soldering up some cards and getting them in the mail.
VERAULT
Veteran Member
..... Ok lets get to the updates... Im not happy.So the RAM came in:
No matter what combination I try it wont work. I learned that this machine wont boot with any less than four of the ram sockets populated with 1MB simms which need to be installed with the first simm closest to the CACHE chips.
So if those wont work what do I do? I cant return them. Thats money lost I guess. Any ideas?
On to cachechk. For some reason if I boot into DOS and run it it freezes. The program with start. In the bottom left it will have XMS present then the next line down VCPI present.. then it locks. So I need to bypass my startup files/use a boot disk then run Cachechk -v memory managers causing an issue? I have no idea.
HEre is the output:
maxtherabbit
Veteran Member
Wait, you only bought 2 SIMMs? The 486 having a 32bit data bus requires they be installed in matching sets of 4.
So the RAM came in: View attachment 1273412
No matter what combination I try it wont work. I learned that this machine wont boot with any less than four of the ram sockets populated with 1MB simms which need to be installed with the first simm closest to the CACHE chips.
So if those wont work what do I do? I cant return them. Thats money lost I guess. Any ideas?
On to cachechk. For some reason if I boot into DOS and run it it freezes. The program with start. In the bottom left it will have XMS present then the next line down VCPI present.. then it locks. So I need to bypass my startup files/use a boot disk then run Cachechk -v memory managers causing an issue? I have no idea.
HEre is the output: View attachment 1273419
Just buy 2 more
VERAULT
Veteran Member
HAha! is that it? Its strange how I have so many memories of building and using these but damn getting old sucks.
Ill buy two more right now.
DONE....
Any thoughts on the cache?
Ill buy two more right now.
DONE....
Any thoughts on the cache?
VERAULT
Veteran Member
Just found this thread and seems my nic isnt so cut and dry. https://forum.vcfed.org/index.php?posts/1366657/
VERAULT
Veteran Member
I have had a nightmare of a time setting this machine up. Conflict after conflict. I will make an update but first I have something that might be a serious problem. Why is it when I directory a floppy disk. Remove it, and insert a new disk and do DIR, it returns the same results as the first disk even though its not? I have to goto the B: for a second, then back to the A: and then it will read the new disk. What is going on here? This seems very ungood.
syzygy
Experienced Member
Agent Orange
Veteran Member
Were both floppy's formatted on the same drive? If not, that could be an issue.
VERAULT
Veteran Member
No I dont have that option. Ill look into it.
some were some were not but they all behave like this.
modem7
10k Member
There is an assumption that 1.2M drives are configured to generate a 'disk changed' signal (on pin 34 of the cable) when the floppy is changed. If that signal is not being generated, DOS (and it would be versions 3.0 and later) assume that the disk has not been changed, and will use a cached copy of the directory.
See if your 1.2M drive is configured to generate a 'disk changed' signal.
A workaround is described at [here].
BTW. In at least early versions of DOS, a CTRL-C key combination would force a fresh read of the floppy (like what we had to do in the CP/M OS).
maxtherabbit
Veteran Member
Yes as other's have said this is definitely a problem with the disk changed line. You can add a DRIVPARM directive to force DOS to ignore the signal as suggested, but that is treating the symptom not the cause.
The cause could be anything from a port conflict on port 3F7 to a bad floppy drive logic board to a misconfigured jumper on the floppy drive. Most likely the latter