• Please review our updated Terms and Rules here

pentium overdrive replacing 486dx2 fails to complete boot

cmc

Experienced Member
Joined
Feb 23, 2014
Messages
107
Location
South Bend, IN
Hello Friendly VCFers

I have a 486 class system with the following configuration:

Intel 486dx2-66
DTK PKM-0038S motherboard
36mb ram
Cirrus GD5426 VLB video card
DTK PTI-250DW VLB IDE card
(and sound and network cards removed for the time being)

It is rock solid stable with the dx2.

For fun and historical interest I picked up a Pentium Overdrive 83mhz. It fit nicely in the board and I altered to jumper settings as specified on the motherboard to indicate P24T. I also changed the clock speed to 80mhz, which is the closest option available to 83mhz. See the settings here:
http://stason.org/TULARC/pc/motherboards/D/DTK-COMPUTER-INC-486-PKM-0038S-VER-6-02.html

The machine posts, checks memory, runs in BIOS, etc., just fine, but will not load an OS. It stops right after the System Configurations page (see attached picture). It just sits there forever.

I have checked and rechecked the jumpers. Perhaps it has something to do with the slightly-wrong clock frequency?

I thought to ask the forum before I do anything drastic.

IMG_20150718_231034252.jpg
 
When I installed an Overdrive some time ago I set it up for P24D; not P24T. I think P24T is different in some respect regarding the cache. The bios tests the cache at the point your system is hanging so this might be indicative of the problem.

Take this with a grain of salt though; I may not have got it right all those years ago.
 
I've played with the 83 version on and off and I never liked it, too many problems with the software. Try booting with a system floppy and see if can at least get to a command prompt A:>, and then we can go from there.
 
When I installed an Overdrive some time ago I set it up for P24D; not P24T. I think P24T is different in some respect regarding the cache. The bios tests the cache at the point your system is hanging so this might be indicative of the problem.

Take this with a grain of salt though; I may not have got it right all those years ago.

I did some quick research and it seems the only difference between the two is the P24T has write-through cache while the P24D has write-back cache and is slightly faster at the same clocks.

WB cache can cause problems because not all 486 systems supported it. I had a Evergreen Am5x86 upgrade chip in the late 90s where you could select the cache mode, and WB always caused weird issues like hangs or crashes on the machine I had it in.
 
You changed the clock speed to 80MHz? I think that is not correct. An 80MHz 486 would use a 40MHz bus clock doubled. The Pentium Overdrive 83 is supposed to use a 33MHz bus (at 2.5X). I suspect the reason your chip isn't working is because you are overclocking it. Don't put much faith in the speed that the BIOS reports.
 
Thanks for all this advice.

On the issue of P24T vs. P24D, I was under the impression that the 486dx2 was the P24D socket. I suppose it is possible that this board treats the overdrive as a drop-in replacement -- no jumper changes needed. That would also solve the bus clock issue. However, the clock speed would still be 66, and not 83. The board has no 83mhz option...

I will try this out and report!
 
You changed the clock speed to 80MHz? I think that is not correct. An 80MHz 486 would use a 40MHz bus clock doubled. The Pentium Overdrive 83 is supposed to use a 33MHz bus (at 2.5X). I suspect the reason your chip isn't working is because you are overclocking it. Don't put much faith in the speed that the BIOS reports.

That makes a lot of sense. But I wonder how to set 83mhz when the closest options at 66, 80, and 100. Perhaps the MB doesn't support the overdrive after all. I have heard there are boards that were made before Intel finalized the specs. The board is from Nov. 94 or so, so it is possible.
 
But I wonder how to set 83mhz when the closest options at 66, 80, and 100. Perhaps the MB doesn't support the overdrive after all.
That doesn't sound right. The normal bus speeds for a 486 were 33, 40, and 50. Perhaps your board has it labeled at 2x for a DX2? I'd try the "66" setting, and that might actually give you a 33mhz bus, which should be what the 2.5x Overdrive 83mhz expects.

Most Overdrives were intended as drop-in replacements for existing DX or DX2 486 CPUs. Not all boards could have their bus speeds or voltages adjusted, so it was critical that one got the right Overdrive to match the system.
 
The settings for the DX-33 or the DX/2-66 should be identical, the clock doubling is handled internally (not set by jumpers). However, in order to use some of the advanced features of the POD-83 it would be best to use a motherboard specifically designed for it.
 
Well, I have reset the jumpers to 486dx2 mode, and reinstalled the original 486. Everything worked fine. Then I pulled the 486 and dropped in the overdrive, and the machine now doesn't even post! So, I do have to set the jumpers to P24T after all it seems.

I am thinking that this is one of those boards that is not really compatible with the POD83, despite what is advertised. Or, it may be some obscure jumper settings that I will not know without a proper manual.
 
An update, the system will pass cache check by disabling the cache in BIOS, but will not read any disks... it will say disk boot failure even with a valid disk.

This thing is as temperamental as its reputation. It is also possible that the CPU is fried somehow.

I may just find myself a good ol' dx4 instead. :)
 
Last edited:
I have a P83OD as well, and in a late 486DX2-66 (Compaq Deskpro, PCI system), it works fine (did not have to change any settings, just drop it in). On an older board, which has the right socket, and according to the manual, has "P24T" support, it does not work.
I think the problem is that these motherboards were already on the market by the time the P83OD was released, and the boards were probably based on preliminary specs, and never actually tested with the real thing, so they don't work.
You see a similar thing with AMD Bulldozer CPUs. Quite a few boards originally marketed as "AMD FX-compatible" didn't work properly once the chip was finally released. They would need an BIOS update, but they were already EOL at that point.

I have to say, that *if* the P83OD works, it's a fantastic CPU, and quite close to a real Pentium 75 system in many cases. Completely puts those 486DX4/DX5 systems to shame. It can actually play Quake :)
 
That problem you describe is very common for the POD83. The problem is that POD83 uses "write back cache", and older motherboards are often not compatible with CPUs that have write back cache.

Most 486 motherboards support write back L2 cache, but only the later 486 boards support write back CPU cache. Maybe you can examine your chipset and tell us which one your motherboard uses. Also, check your manual to see if it supports any other CPUs that support write back cache. Perhaps there is just a error for the P24T jumper setting.
 
That problem you describe is very common for the POD83. The problem is that POD83 uses "write back cache", and older motherboards are often not compatible with CPUs that have write back cache.

Most 486 motherboards support write back L2 cache, but only the later 486 boards support write back CPU cache. Maybe you can examine your chipset and tell us which one your motherboard uses. Also, check your manual to see if it supports any other CPUs that support write back cache. Perhaps there is just a error for the P24T jumper setting.

Just had a look at the old motherboard/overdrive that I installed years ago (now sitting in an antistatic bag). Looking a the mobo manual it supports write back CPU cache. Seems like it was a quite late model 486 motherboard; I think it originally had a 133MHz processor in it.
 
I have to say, that *if* the P83OD works, it's a fantastic CPU, and quite close to a real Pentium 75 system in many cases. Completely puts those 486DX4/DX5 systems to shame. It can actually play Quake :)

The POD also is the only way a 486 machine can run Windows XP. Though the machine also needs to support 64MB of RAM minimum.
 
That problem you describe is very common for the POD83. The problem is that POD83 uses "write back cache", and older motherboards are often not compatible with CPUs that have write back cache.

Most 486 motherboards support write back L2 cache, but only the later 486 boards support write back CPU cache. Maybe you can examine your chipset and tell us which one your motherboard uses. Also, check your manual to see if it supports any other CPUs that support write back cache. Perhaps there is just a error for the P24T jumper setting.
Interesting... my motherboard does support write back cache. I don't have the manual, but in BIOS I can set L1 cache to write back or write through mode.

I was able to get it past the cache check by disabling the L1 cache entirely. However then it will not actually boot. The system does not detect the correct hard drive size (in the auto config in BIOS), and even with the correct settings put in manually, it says "disk not bootable" or something to that effect. The same message occurs for boot floppies.

The jumper settings for my IDE card are here:
http://stason.org/TULARC/pc/hard-di...E-AT-drives-PTI-250DW-UN-10.html#.VbOfHGCVtNl

I don't see anything obvious there... I assume that the jumpers should stay the same since the bus is still 33mhz.
 
CMC:

What I would do at this point is pull the POD and get my system back to square one. Once everything is copacetic, reinstall the POD and make yourself a check list, and keep track of the jumpers and BIOS setting. Eventually you will arrive at the right combinations at it will work. However, if it's anything like mine, I wouldn't expect jaw dropping performance. The reason that chip exists was just a stop-gap between the 486's and the Pentiums, which were very expensive at that point in time. Good luck!




 
CMC:
What I would do at this point is pull the POD and get my system back to square one. Once everything is copacetic, reinstall the POD and make yourself a check list, and keep track of the jumpers and BIOS setting. Eventually you will arrive at the right combinations at it will work. However, if it's anything like mine, I wouldn't expect jaw dropping performance. The reason that chip exists was just a stop-gap between the 486's and the Pentiums, which were very expensive at that point in time. Good luck!

Thanks, I'm following this advice. I have the system back as it was, with the dx2 66 in. I am now testing it to make sure it is still 100% stable as it was before.

I don't expect "true pentium" performance -- this is mainly to satisfy my 20 years of curiosity about the various 486 chips. My plan is to create a reference system around a dx2 66, and then modify different variables to see what /really/ mattered. I have different types of cache chips, disks, video controllers, video ram, etc. to try.

In addition to this POD chip, I have several things I'd always wondered about such as:
1) compare the AMD vs. Cyrix 5x86
2) compare a 75mhz 486 at 25mhz bus to a 66mhz at 33mhz bus
3) compare a real 80mhz, 40mhz bus 486 to a 66mhz overclocked to 80mhz (a friend once told me he did this and it worked the same, always wondered if it was true)
4) real world write back to write through cache performance difference
5) does 20ns to 15ns cache make any difference at all (some claimed it did, I doubt it)
MORE!

Anyway, now that prices are relatively low, I can finally know all these things. Could be a few years yet before I answer my similar questions about the i5 vs i7... :p
 
Yeah it's nice being able to do stuff now at a fraction of the cost needed back in the day. One reason I have a couple of 5x86 cpus and collected various 486 cps to play with. I've done the same with early pentium 1/ socket 7/super socket 7 class cpus as well. Guess I just have have more of an interested in these than any other '90s setups.

There's been a some extensive and comprehensive benchmark testing on both types over the years.
 
Back
Top