VOGONS


Test:486 undocumented jumper settings

Topic actions

Reply 20 of 36, by retro games 100

User metadata
Rank l33t
Rank
l33t
5u3 wrote:
Tetrium wrote:

So anyone know of a small little program that I can boot from that can do a quick cpu-speed test from a batched bootable DOS disk?

CHKCPU <-- This one is really nice!

Excellent utility! 😀 It's interesting that the copyright datestamp is 1997 to 2010. A long and active development phase. 😎

Reply 21 of 36, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++
Tetrium wrote:

[So anyone know of a small little program that I can boot from that can do a quick cpu-speed test from a batched bootable DOS disk?

Not a speed test, but on the ULTIMATE BOOT CD (which I highly recommend to anyone) are the Intel CPU ID tools. They work really well and tell if your CPU has been overclocked or underclocked and by how much...

Reply 22 of 36, by retro games 100

User metadata
Rank l33t
Rank
l33t

I bought a couple of AMD-X5-133ADZ CPUs from ebay. Unfortunately, one appears to be dead. The other one seems to work. I tried it at 33 FSB, and then at 40 FSB. No problems. I then tried setting the FSB to the next highest setting on the Asus 486 board, which is the undocumented 48 FSB setting. Understandably, I see no BIOS POST on the screen, and so I tried:

* BIOS options set to "fail safe".
* Mobo's wait state jumpers set to 1 Wait and > 33 MHz. BTW, I'm using a PCI video card.
* Mobo's CPU jumper set to write thru cache (not WB).

One thing I could try next is to set the mobo's CPU voltage jumper to 3.6V, which is the Cyrix CPU option. Will this kill the 3.45V AMD CPU, or give it the necessary extra juice in order to run it at this crazy 48 FSB setting? Damn nuisance about the other CPU. It must be dead, because both CPUs are identical, and one works while the other doesn't. Both CPUs had badly bent pins on arrival. 😒

Edit: For the 48 MHz FSB test, I tried removing the PCI video card, and replacing it with a VLB card. No POST. The VLB card works OK, because if I drop the mobo's FSB back down to 40 MHz, the BIOS POST reports that the 133 MHz chip is running at 160 MHz (4x multi * 40 = 160).

Reply 23 of 36, by BastlerMike

User metadata
Rank Member
Rank
Member

Are you trying to run the thing at 4x 48 MHz ?
First you should make sure, that the board runs with this 48 MHz / 3x multiplier at 144 MHz without problems.
Use an ISA graphics card, they are less sensitive to clocking experiments.

Reply 24 of 36, by retro games 100

User metadata
Rank l33t
Rank
l33t
BastlerMike wrote:

Are you trying to run the thing at 4x 48 MHz ?
First you should make sure, that the board runs with this 48 MHz / 3x multiplier at 144 MHz without problems.
Use an ISA graphics card, they are less sensitive to clocking experiments.

Hehe yes, I am attempting to run this thing at 4x 48 MHz. I took your advice, and tested the board at 3x 48 MHz = 144 MHz. I set the mobo's bus speed jumpers to 48 MHz. BIOS POST OK! I then ran chkcpu (downloaded from 5u3's ckcpu120 attachment above), and it reported:

Internal CPU cpeed : 148.0 MHz
Clock multiplier : 3.0
Bus clock speed : 49.3 MHz

It's possible that a multi of 4x and a bus speed of 48-50 MHz is far too much for the mobo+CPU to cope with. Unless I alter the mobo's CPU voltage jumper, to 3.6V? Another idea is to try a UMC chipset based mobo, and set its bus speed jumper to 60 MHz, and set the mobo's CPU multiplier to 3x. That would yield a speed of 3x * 60 FSB = 180 MHz. BTW, unfortunately I do not own a single ISA video card.

Reply 25 of 36, by unmei220

User metadata
Rank Member
Rank
Member

3.6V is within specs, so don't worry about frying the processor. I remember that there was even people running it at 5V in those times. Don't know how much time they survived at that voltage though...

The Am5x86 needs a supply voltage of 3.45V +/- 0.15V

Reply 26 of 36, by retro games 100

User metadata
Rank l33t
Rank
l33t

OK, I changed the mobo's CPU voltage jumper from 3.45V to 3.6V, and tested the board+CPU on the 3x multi / 48-50 FSB jumper setting. The BIOS POST is displayed, and no burning smell! 😉 I then altered the mobo's CPU multiplier from 3x to 4x, but again I do not see a BIOS POST on the screen. Perhaps what I am attempting to do on this particular board may not be possible?

Reply 27 of 36, by unmei220

User metadata
Rank Member
Rank
Member

In MY opinion, I think you're asking too much of the poor CPU / chipset. Don't know which one of the two is the impeding factor here, unless both are. I hear the max these CPUs could achieve was 160Mhz (40Mhz X 4) ?. So I think trying to do near 200Mhz is beyond their reach. Are you using a ADW or ADZ 5x86 ? I think the ADZ model overclocks better because it can stand higher temperatures (85º vs 55º of the ADW model).

EDIT: Sorry. Just forgot that you already said they're ADZ ones. Just ignore my last remark.

Last edited by unmei220 on 2010-11-30, 14:51. Edited 1 time in total.

Reply 28 of 36, by retro games 100

User metadata
Rank l33t
Rank
l33t

It's definitely the ADZ model. The full code written on the CPU is AMD-X5-133ADZ.

Reply 29 of 36, by BastlerMike

User metadata
Rank Member
Rank
Member

3.6V is insufficient. From my experiences I can say that some Am5x86 need at least 4V to run at 180 MHz. Many of them won't do this speed even if voltage is increased to 5V !

Reply 30 of 36, by retro games 100

User metadata
Rank l33t
Rank
l33t

Ah, I see. Do you know how I can tell my Asus 486 mobo to send more volts (eg 4V or 5V) to this AMD 3.45V P75 chip? Do I "trick it" in to thinking that I have installed a lower speed Intel SX or DX chip - the ones that require higher voltage, eg 5V? These mad experiments help me to understand more about how old PC hardware works. 😀

Reply 31 of 36, by BastlerMike

User metadata
Rank Member
Rank
Member

The board autimatically detects if a 3,3V or 5V CPU is installed.
I'm not very familiar with such things, but to give 5V to a 3.3V CPU requires shorting some pins at the voltage regulator. I think JP4 and JP5 were intended for manual voltage setting. 'Unfortunately they were replaced with a switching transistor.

Reply 32 of 36, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie

The PVI-486SP3 has automatic voltage detection, so it would require a modification on either the socket or the CPU.

Further reading: Upgrading and repairing PCs, 5x86

160 MHz is the limit for my 5x86 as well.

Reply 33 of 36, by retro games 100

User metadata
Rank l33t
Rank
l33t

Thanks very much for everyone's interesting comments. At this point, I'm going to have a dig about in my box of 486 mobos, and see if I can spot one that had a voltage regulator adapter attached to it. Will such an adapter allow me to set the CPU's voltage, from typical values such as ~3.5V up to ~5V? If so, I could set it to ~4V or ~5V, and use it on the Asus mobo...

Also, check out the photo in this ebay auction! Underneath this P75 chip, there are 3 dip switches. I wonder if these dip switches control the CPU's voltage?

Reply 34 of 36, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
unmei220 wrote:

In MY opinion, I think you're asking too much of the poor CPU / chipset. Don't know which one of the two is the impeding factor here, unless both are. I hear the max these CPUs could achieve was 160Mhz (40Mhz X 4) ?. So I think trying to do near 200Mhz is beyond their reach. Are you using a ADW or ADZ 5x86 ? I think the ADZ model overclocks better because it can stand higher temperatures (85º vs 55º of the ADW model).

EDIT: Sorry. Just forgot that you already said they're ADZ ones. Just ignore my last remark.

Might be worth a shot with a very late 5x86 model of, say, 1997?
Btw...€42 for a 5x86? MUCH too expensive for my taste!

Reply 35 of 36, by BastlerMike

User metadata
Rank Member
Rank
Member

Those jumpers are not only for voltage setup. Most of these adapters use one for L1 WB/WT setting, one for multi selection (3x or 2x/4x), and one for switching between 3,3 V and 3,45 V

Reply 36 of 36, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Using an oscilliscope is probably the most reliable means to determine what FSB frequency is in use. You want to probe the motherboard just before the CLK signal hits the PGA socket. In many cases, there is a top-to-bottom layer change hole for the traces where I jam in a single stranded wire and connect that to the scope. Alternately, you can probe the PLL chip itself. If the PLL clock generator chip is in a DIP package, I sometimes pull the PLL chip out and wrap a little wire around the CLK_OUT pin and connect that to the scope. Then plug the PLL back in with the wire connected to the CLK_OUT pin.

The FSB of 48 MHz seems a bit awkward. I'd be curious what the scope shows for this setting.

Plan your life wisely, you'll be dead before you know it.