• Please review our updated Terms and Rules here

Pentium Overdrive 83MHz Overclocking

D

DamienC

Experienced Member
Joined
Aug 2, 2010
Messages
72
Location
Southern NJ USA
I finally got my 83MHz Pentium Overdrive system working properly. It runs nice and stable at the stock 83MHz speed.

I'd like to squeeze a little more performance out of this thing through overclocking though. Normally, the PODP runs with a 2.5x multiplier and 33MHz bus, giving the 83MHz clock speed. From what I've gathered online, the multiplier can't be changed and I'm stuck with changing the bus speed to overclock.

My goal speed is 100MHz, which I can achieve by setting the bus to 40MHz. My system booted this way, but it was very unstable: Quake bombed immediately at runtime with a page fault, and Doom would run but there would be random wall textures everywhere then it would crap out with texture-related errors.

I'm pretty sure it's because the video card I'm using is PCI (Diamond Stealth 3D 2000) and the PCI bus is I think unstable at anything above 33MHz.

I'd still like the 100MHz clock speed. Is there a way to get this system running stable at this speed? Here's what I think are my only options:

  • Somehow figure out how to change the multiplier to 2x with a 50MHz bus, or 3x with a 33MHz bus. I doubt changing the multiplier is possible, but it would probably run the most stable with 33MHz x 3.
  • Swap out the PCI video card for an ISA or VLB video card and run the system with a 40MHz bus. I'm reluctant to do this simply because I don't feel like hunting down another card.

Any ideas?
 
DamienC said:
I finally got my 83MHz Pentium Overdrive system working properly. It runs nice and stable at the stock 83MHz speed.

I'd like to squeeze a little more performance out of this thing through overclocking though. Normally, the PODP runs with a 2.5x multiplier and 33MHz bus, giving the 83MHz clock speed. From what I've gathered online, the multiplier can't be changed and I'm stuck with changing the bus speed to overclock.

My goal speed is 100MHz, which I can achieve by setting the bus to 40MHz. My system booted this way, but it was very unstable: Quake bombed immediately at runtime with a page fault, and Doom would run but there would be random wall textures everywhere then it would crap out with texture-related errors.

I'm pretty sure it's because the video card I'm using is PCI (Diamond Stealth 3D 2000) and the PCI bus is I think unstable at anything above 33MHz.

I'd still like the 100MHz clock speed. Is there a way to get this system running stable at this speed? Here's what I think are my only options:

  • Somehow figure out how to change the multiplier to 2x with a 50MHz bus, or 3x with a 33MHz bus. I doubt changing the multiplier is possible, but it would probably run the most stable with 33MHz x 3.
  • Swap out the PCI video card for an ISA or VLB video card and run the system with a 40MHz bus. I'm reluctant to do this simply because I don't feel like hunting down another card.

Any ideas?
Click to expand...

So the surface-mount components on top of the ceramic package don't have any input for setting the multiplier? I'll have to pull out my own and take a look too. A little more to supplement your correct information (i.e. 2.5 multiplier) are the CPUID and stepping values I have found:

CPUID: S-spec: MHz: Stepping: (C) Markings:
1531h SZ953 63 B1 '89 '94
1531h SZ990 63 B2 '89 '94
1532h SU013 63 C0 '92 '94
1532h SU014 83 C0 '92 '94

Notes:

63MHz internal clocking is a 25MHz FSB x 2.5
83MHz internal clocking is a 33MHz FSB x 2.5

Despite the 'B' stepping versions, the Socket 3 POD63 didn't have the Pentium FDIV bug like 'B' and prior steppings of Pentium Socket 4 and Socket 5 CPUs...
 
I have a POD 83 that will do 100 stable (40x2.5). I think the motherboard may be a factor because I don't remember being able to achieve this speed with the CPU on a board I had way back years ago. But I have a MSI-4144 PCI 486 mobo that it seems happy in.

The only issue I have is that the mobo's L2 cache doesn't seem to operate when the POD is installed. It's enabled and the BIOS says it's there, but it doesn't show up in cachemem. The result is that while the 100 MHz POD is the clear winner for 3D games due to its FPU, for apps that don't leverage the FPU (like Windows) a Am5x86 160 is faster. PODs are strange CPUs and they often have compatibility compromises/issues even with mobos that supposedly completely support them.

Regarding 40 MHz bus issues, there are a few things you could try. I'd start with trying another CPU and see if that works. If so, you know it's the POD that's the problem. There are lots of other things to try too. Different RAM, set the BIOS as conservatively as possible, and of course try one card at a time and in different slots.
 
Last edited:
You will have to play around with the bus speed to overclock the unit, as the the multiplier is hard wired.
But in saying that, the chip fakers did successfully hack this on socket 5 overdrives. So theoretically this could be done to a POD83 :)

The Bus/Core ratio is set by two pins & on the POD63 & POD83 they are both flagged as 0 giving a Bus/Core ratio of 2/5 (33bus/83MHz)
by making Pin X flagged as 1 you get a Bus/Core ratio of 1/3 (33/100)
by making Pin Y flagged as 1 you get a Bus/Core ratio of 1/2 (33/66)
by flagging both pins as 1 you get a Bus/Core ratio of 2/3 (33/50)

the fakers carried these hacks under the die cap (mainly so they werent so obvious) but the hack could be carried out at the pins.
I have pinouts for socket 4, 5 & 7 Pentiums & Pentium Overdrives, but will need to track down the pinout diagram for the POD63/83 to help any further :)
 
swaaye said:
...The only issue I have is that the mobo's L2 cache doesn't seem to operate when the POD is installed. It's enabled and the BIOS says it's there, but it doesn't show up in cachemem. The result is that while the 100 MHz POD is the clear winner for 3D games due to its FPU, for apps that don't leverage the FPU (like Windows) a Am5x86 160 is faster. PODs are strange CPUs and they often have compatibility compromises/issues even with mobos that supposedly completely support them...
Click to expand...

It's a known issue that can be fixed by a few different ways: http://ibmmuseum.com/ohlandl/76-77/7677_Lacuna.html#Interposer...
 
Neon_WA said:
...I have pinouts for socket 4, 5 & 7 Pentiums & Pentium Overdrives, but will need to track down the pinout diagram for the POD63/83 to help any further...
Click to expand...

There is a data sheet out there, I have it saved away somewhere...
 
Tetrium said:
Heres some info:
http://mysite.verizon.net/pchardwarelinks/486pin.htm
Includes a link to an intel download about the POD83
Click to expand...

thx for the link mate :)
I will have to do a bit of research as some of the pin naming appears different :dontgeti:
but as the pins only need to be flagged as 0, they maybe one of the NC (not connected) pins which won't make it easy.

I intend to buy some ODs intended for gold recovery and open them up and learn how to carry out the hacks that the fakers developed.
 
Neon_WA said:
thx for the link mate :)
I will have to do a bit of research as some of the pin naming appears different :dontgeti:
but as the pins only need to be flagged as 0, they maybe one of the NC (not connected) pins which won't make it easy.

I intend to buy some ODs intended for gold recovery and open them up and learn how to carry out the hacks that the fakers developed.
Click to expand...

Think it would be possible to change the default multi? I'd love to have one I could clock at 33x3 instead of 33x2.5 :)
 
I have really, really tried everything with these chips. If you remove the fan, the chip will work at 1x multiplier. It was really cool to use it at 66mhz bus on one of my via motherboards. I have a total of 3 chips + 1 on the way. 2 of my chips will not do 100mhz, no way no how. I manually built a vrm for them and tried 3.3, 3.6, 4.0, and 4.25 volts. 1 of my chips seems to do 100mhz at stock voltage and fan, but I have not stability tested it yet.
 
If anybody finds docs or has an idea on how to get one to work on a Socket 1 or 0 board let me know. I planned to make an adapter, then ended up not doing it..
 
joekster said:
I have really, really tried everything with these chips. If you remove the fan, the chip will work at 1x multiplier. It was really cool to use it at 66mhz bus on one of my via motherboards. I have a total of 3 chips + 1 on the way. 2 of my chips will not do 100mhz, no way no how. I manually built a vrm for them and tried 3.3, 3.6, 4.0, and 4.25 volts. 1 of my chips seems to do 100mhz at stock voltage and fan, but I have not stability tested it yet.
Click to expand...
Which POD are you referring to? A 1x multi on the POD83 would result in a 33Mhz speed. 66Mhz 1x multi sounds more like socket 5/7
 
The pentium overdrive for 486 systems does have 2 multipliers. the standard one is 2.5x. If the fan is removed (or doesnt work anymore), the chip will default to 1x - multiplier I assume as a safeguard.

I got one more overdrive chip (now a total of 4). below are my overclocking results:

chip A: not stable at 100mhz. tried 4v vrm as well.
chip B: not stable at 100mhz. tried 4v vrm as well.
Chip C: seems stable at 100mhz with built-in 3.3/3.6v vrm. I did have a problem installing windows98, but that could be unrelated??? Am able to install and run windowsxp for hours straight.
Chip D: stable for 5-10min at 100mhz with the built-in 3.3/3.6v vrm. I suspect that with better cooling and/or a 4v vrm it would be completely stable at 100mhz.
 
Joekster could you do some benchmarks using Landmark 6 on Chip C and/or D for me? I'd like to compare that with my various chip measurements.
 
Has any progress been made in overriding the locked multiplier on this chip? I was thinking even if you can change it to 1.5 then you can put it on a board with 66mhz bus support and make it a 66/100.
 
Raven said:
Joekster could you do some benchmarks using Landmark 6 on Chip C and/or D for me? I'd like to compare that with my various chip measurements.
Click to expand...

If additional S-specs are found it would be helpful too. I know of only one for the 83MHz POD, SU014, a 'C0' core (there are three I know of for the 63MHz: SZ953, a 'B1' core, SZ990, a 'B2' core, and SU013, a 'C0' core). The earlier steppings not including an 83MHz version, and paired 63/83 'C0' cores implies that improvements and testing were needed to come up with the faster speed.
 
Neon_WA said:
You will have to play around with the bus speed to overclock the unit, as the the multiplier is hard wired.
But in saying that, the chip fakers did successfully hack this on socket 5 overdrives. So theoretically this could be done to a POD83 :)

The Bus/Core ratio is set by two pins & on the POD63 & POD83 they are both flagged as 0 giving a Bus/Core ratio of 2/5 (33bus/83MHz)
by making Pin X flagged as 1 you get a Bus/Core ratio of 1/3 (33/100)
by making Pin Y flagged as 1 you get a Bus/Core ratio of 1/2 (33/66)
by flagging both pins as 1 you get a Bus/Core ratio of 2/3 (33/50)

the fakers carried these hacks under the die cap (mainly so they werent so obvious) but the hack could be carried out at the pins.
I have pinouts for socket 4, 5 & 7 Pentiums & Pentium Overdrives, but will need to track down the pinout diagram for the POD63/83 to help any further :)
Click to expand...

Which are the pins that control the multiplier and how do you change the flags? Would you insulate the pin that you want to change from 0 to 1 so that it isn't seen?
 
joekster said:
The pentium overdrive for 486 systems does have 2 multipliers. the standard one is 2.5x. If the fan is removed (or doesnt work anymore), the chip will default to 1x - multiplier I assume as a safeguard.

I got one more overdrive chip (now a total of 4). below are my overclocking results:

chip A: not stable at 100mhz. tried 4v vrm as well.
chip B: not stable at 100mhz. tried 4v vrm as well.
Chip C: seems stable at 100mhz with built-in 3.3/3.6v vrm. I did have a problem installing windows98, but that could be unrelated??? Am able to install and run windowsxp for hours straight.
Chip D: stable for 5-10min at 100mhz with the built-in 3.3/3.6v vrm. I suspect that with better cooling and/or a 4v vrm it would be completely stable at 100mhz.
Click to expand...

Could you PLEASE tell us which VIA Socket 3 board you used to overclock the FSB to 66Mhz? Afaik only UMC chipsets support a PCI divider and overclocking the PCI bus to 66Mhz is darn right outrageous.
 
You must log in or register to reply here.
Back
Top