VOGONS


Reply 40 of 52, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t

I'm testing some different CPUs in this board and here are my findings so far:

AMD 5x86 133 seems to only run at 100Mhz and benchmarks pretty low with a Speedsys score of around 35 (though my BIOS settings are at defaults and probably not optimal)
AMD 486DX4-100 runs at 80Mhz and seems flaky (should be 33x3... no idea how that equals 80Mhz)
Cyrix 486DX4100GP runs properly at 100Mhz, Speedsys benchmark ~36
Pentium Overdrive PODP5V83 runs properly at 83Mhz (startup message incorrectly says it is 100Mhz), Speedsys benchmark ~54

I am new to tweaking 486 boards, so I will take some pictures of the BIOS screens and post them here later if anyone has any suggestions for BIOS settings.

Also, for some reason every time I start up the POST screen is showing "EISA CMOS Checksum Failure", but it proceeds to the memory test as if nothing is wrong. The system is keeping time, date and settings normally. I have not tried any EISA cards yet, so maybe that is why?

EDIT: Here is how the BIOS is set up currently. Should system bus speed be set to 7.159Mhz? That is the default.

The attachment 20240610_174107 (Custom).jpg is no longer available
The attachment 20240610_174038 (Custom).jpg is no longer available
The attachment 20240610_174028 (Custom).jpg is no longer available
The attachment 20240610_174007 (Custom).jpg is no longer available

Now for some blitting from the back buffer.

Reply 41 of 52, by Skorbin

User metadata
Rank Newbie
Rank
Newbie
Ozzuneoj wrote on 2024-06-10, 19:47:

...

Skorbin wrote on 2024-06-10, 09:04:

Anyhow, if there is interest, I can check my old notes I have floating somewhere on my NAS.
If I am not mistaken, the small program should be there as well.

If you have a chance to do this, I would definitely appreciate having any other info about the board that you are able to dig up.

So far I have not found the files.
But when I tested the board in 2007 I wrote some information on the german Hardwareluxx forum.
Therefore I know that I had the 5V version (probably ver. 1.5).
The program I used to make my RAM known was called "setxmsto.com". I will check some old backups on the weekend, whether I still have it floating around somewhere.

Reply 42 of 52, by NJRoadfan

User metadata
Rank Oldbie
Rank
Oldbie

I think "System Bus Speed" is the speed the EISA slots operate at. If you are running a 33Mhz CPU bus, CPUCLK/4 should be safe. EISA officially targeted 8.33Mhz.

Edit: Here is the official AMIBIOS docs for the SiS EISA chipset.

Reply 43 of 52, by Anonymous Coward

User metadata
Rank l33t++
Rank
l33t++

You need to run the EISA config program and feed it the correct .cfg file for that EISA checksum error to go away. Any time to add an EISA expansion card, or move it to a different slot then you need to do the same. You'll need a cfg and ovl file for each EISA expansion card.

Does your board have a jumper to set the CPU multiplier? Your board shouldn't have any software issues stopping the am5x86 from running 4x.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 44 of 52, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t
Anonymous Coward wrote on 2024-06-10, 22:52:

You need to run the EISA config program and feed it the correct .cfg file for that EISA checksum error to go away. Any time to add an EISA expansion card, or move it to a different slot then you need to do the same. You'll need a cfg and ovl file for each EISA expansion card.

Does your board have a jumper to set the CPU multiplier? Your board shouldn't have any software issues stopping the am5x86 from running 4x.

Good to know about the EISA configuration! It should be fine then.

Also, there are no multiplier jumpers. I thought with 486 systems the multiplier was internal to the processor? The board has settings for CPU Type, bus speed and voltage. I'll just copy and paste the stuff from the manual:

CPU Jumpers […]
Show full quote

CPU Jumpers

CPU Type Jumpers
The 486EI system board supports three types of processors:
Type A: 486SX
Type B: 486DX, 486DX2 , 486DX4
Type C: 487SX , P23T, OverDrive Processors


The jumper settings for each type of CPU are as follows:

CPU J1 (2-pin) J2 (3-pin) J3 (3-pin)
Type A OPEN OPEN 2-3
Type B SHORT 1-2 1-2
Type C SHORT 2-3 1-2

CPU Clock Frequency Jumpers

The 486EI system board provides following jumpers for
different external clock frequencies to CPU ranging from
20MHz up to 50MHz to match to the external clock
frequency required by the CPU installed.

CPU
JO1 JO2 JL2

SX-25
DX-25
DX/2-50
SHORT SHORT OPEN

SX-33
DX-33
DX/2-66
DX/4-100
OPEN SHORT OPEN

SX-40
DX-40
DX/2-80
SHORT OPEN SHORT

DX-50
OPEN OPEN SHORT

Supply Voltage to CPU Jumper

CPU Type JP12

All DX, SX CPUs (5.0V) 1-3, 2-4

For DX/4 CPU (3.3V) 3-5, 4-6

So, I guess it's possible that the AMD chips need to be set to a different "Type" than a normal DX, DX2 or DX4? I have only used the Type 2 setting so far, and it has worked on everything except the AMD DX4 and 5x86 chips.

EDIT: I just saw your post here which explains it well. I will tinker a bit to see if anything changes but I don't see any standard multiplier jumper on this board.

EDIT2: Okay. Not sure why the AM486 DX4 100 seemed flaky earlier, but it seems to work fine even if it the startup message (after post) says it is only 80Mhz. Speedsys and chkcpu both show it as a "standard" 486DX at 100Mhz. Speedsys benchmarks at 34.74. The Cyrix is definitely the quicker of the two. I can't seem to locate an Intel DX4 in my collection for some reason. I thought for sure I had one.

Now for some blitting from the back buffer.

Reply 45 of 52, by Anonymous Coward

User metadata
Rank l33t++
Rank
l33t++

The multiplier is indeed internal to the processor, and one of the pins on the CPU is dedicated to changing it. That pin is normally connector to a jumper on the motherboard. It would be very unusual if your board supported 3.3V CPUs but didn't have any provisions for changing the multiplier.
Not sure about AMD CPUs, but on Cyrix chips the multiplier could often times be changed either through software or a jumper.

I seem to recall that some of the earlier AMD CPUs had a slightly non-standard pinout. The vast majority of of 486 type CPUs out there use the standard i486DX pinout (or an extension of that).

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 46 of 52, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t
Anonymous Coward wrote on 2024-06-11, 03:24:

The multiplier is indeed internal to the processor, and one of the pins on the CPU is dedicated to changing it. That pin is normally connector to a jumper on the motherboard. It would be very unusual if your board supported 3.3V CPUs but didn't have any provisions for changing the multiplier.
Not sure about AMD CPUs, but on Cyrix chips the multiplier could often times be changed either through software or a jumper.

I seem to recall that some of the earlier AMD CPUs had a slightly non-standard pinout. The vast majority of of 486 type CPUs out there use the standard i486DX pinout (or an extension of that).

There are locations for jumpers which have a permanent jumper wire in place from the factory. I'm not sure if any of these would control the multiplier though since they are undocumented. I can't seem to find anything multiplier related in any of the older revisions, though without 3.3v support on them that isn't really surprising.

I guess it could be possible to check continuity between the multiplier select pin (whichever that is) in the socket and the various permanent jumpers on the board to see if they are directly connected, but I don't know if that would convince me enough to start modding this poor thing. Afterall, the 83Mhz POD is plenty fast enough for this system and works fine (though that tiny fan is the worst noise I've heard from a CPU cooler in my life).

Also, I found an Intel DX4-100 and it works fine. Speedsys scored ~38, so it is the fastest of the 3 DX4-100 chips I have.

I also have a mystery CPU that I just tested at 3.3v and it seems to be working despite being detected as an Intel DX2-66. The bottom markings don't match any Intel DX2 chips I have seen online though.

EDIT: Aha... got the heatsink off of the mystery CPU and it turns out it is one of these Intel 486DX2-66 "PQ" chips manufactured by IBM. That would explain why the few markings on the bottom look like an early gold-bottom Cyrix 486. No idea why the thing seemingly works fine at 3.3v. Is that normal for a DX2?

Now for some blitting from the back buffer.

Reply 47 of 52, by Anonymous Coward

User metadata
Rank l33t++
Rank
l33t++

Intel DX4 scoring higher is than Cyrix/AMD is normal. Intel has 16kb vs 8kb, and some other small enhancements.
It's possible that your board is permanently wired for 3X mode. That might explain why the am5x86 is working at 100MHz, and most DX/2s don't have the multiplier pin and wouldn't care.
Using the multimeter is probably the best way to determine if a CLKMUL jumper exists.

If your POD5V83 fan is making a horrible noise, it's likely the bearings need greasing or are worn out. I've never had the pleasure to work on one of those fans, but it might be possible there is a plastic plug on the front or rear of the fan cage that lets you add grease.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 48 of 52, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t
Anonymous Coward wrote on 2024-06-11, 04:09:
Intel DX4 scoring higher is than Cyrix/AMD is normal. Intel has 16kb vs 8kb, and some other small enhancements. It's possible th […]
Show full quote

Intel DX4 scoring higher is than Cyrix/AMD is normal. Intel has 16kb vs 8kb, and some other small enhancements.
It's possible that your board is permanently wired for 3X mode. That might explain why the am5x86 is working at 100MHz, and most DX/2s don't have the multiplier pin and wouldn't care.
Using the multimeter is probably the best way to determine if a CLKMUL jumper exists.

If your POD5V83 fan is making a horrible noise, it's likely the bearings need greasing or are worn out. I've never had the pleasure to work on one of those fans, but it might be possible there is a plastic plug on the front or rear of the fan cage that lets you add grease.

I will check on that multiplier pin some time, just out of curiosity.

As for the PODP5V83 being loud, I actually have two of them and they both sound about the same and look to be in excellent condition (no dust). I don't think they're worn out... it's just that the fan is so tiny that the noise it makes is really loud and high pitch. I imagine the sound was tolerable back in the mid 90s when this thing was buried in 70 pounds of non-ventilated steel, cards, drives and bundles of sound-absorbing ribbon cables while being masked by the whirring of a constant 4000RPM fan in a full size AT power supply and a few spinning disks... but sitting out on my workbench in a silent office it has pretty much the worst noise profile I've heard from a good quality (name brand) product with a cooling fan. Seriously, it's bad!

... that said, I'm very glad both of them work. I got them in a big lot several years ago and was stoked that I found such desirable (and expensive) CPUs at the time. They will really help to remove CPU bottlenecks when\if I ever get around to doing some video card testing. 😀

Now for some blitting from the back buffer.

Reply 49 of 52, by Anonymous Coward

User metadata
Rank l33t++
Rank
l33t++

I only have one POD5V83. It has a damaged heatsink and no fan. I've only been able to test it at 33MHz to see if it works. When you have your system tuned, I'd be interested to see some benchmark results with that chip at 83.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 50 of 52, by Skorbin

User metadata
Rank Newbie
Rank
Newbie
Skorbin wrote on 2024-06-10, 22:00:
So far I have not found the files. But when I tested the board in 2007 I wrote some information on the german Hardwareluxx forum […]
Show full quote
Ozzuneoj wrote on 2024-06-10, 19:47:

...

Skorbin wrote on 2024-06-10, 09:04:

Anyhow, if there is interest, I can check my old notes I have floating somewhere on my NAS.
If I am not mistaken, the small program should be there as well.

If you have a chance to do this, I would definitely appreciate having any other info about the board that you are able to dig up.

So far I have not found the files.
But when I tested the board in 2007 I wrote some information on the german Hardwareluxx forum.
Therefore I know that I had the 5V version (probably ver. 1.5).
The program I used to make my RAM known was called "setxmsto.com". I will check some old backups on the weekend, whether I still have it floating around somewhere.

Ok, citing myself 😀

The program "setxmsto.com" is included in the XMSDISK package from Frank Uberto.
I'll attach the zip file, as I actually consider the whole package as a very good addition to an early 386/486.

The attachment xmsdsk.zip is no longer available

Reply 51 of 52, by Anonymous Coward

User metadata
Rank l33t++
Rank
l33t++

About your AMD DX/4, are you sure it's running at 80MHz?
I have a feeling it's actually running at 66MHz, but being misreported as 80. I seem to remember that some of the older AMD DX4s had their multiplier pins in a different spot than usual.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 52 of 52, by maxtherabbit

User metadata
Rank l33t
Rank
l33t
Anonymous Coward wrote on 2024-06-10, 22:52:

You need to run the EISA config program and feed it the correct .cfg file for that EISA checksum error to go away. Any time to add an EISA expansion card, or move it to a different slot then you need to do the same. You'll need a cfg and ovl file for each EISA expansion card.

Does your board have a jumper to set the CPU multiplier? Your board shouldn't have any software issues stopping the am5x86 from running 4x.

This is all correct, but I'll add if you have no EISA cards it is totally unnecessary (unless you just want to get rid of the nag text)