Processor question

[Tommy]

I have two identical Pentium III SL52Rs. Both work, but I wonder if one could be having some issues since they were both used, I don't know their history. Here's what makes me wonder.

 

First I had the one in a Dell Optiplex GX110 (It originally came with a 733MHz processor, so this was an upgrade I bought), it worked but sometimes seemed just a bit sketchy, like it would stumble just a bit at times.

 

But here's what makes me wonder. I put this same one in my HP Kayak XM600 using a slotket adapter. This first one out of the Dell first made me clear the BIOS using the dip switch, that *could* be expected. However, the system fan seemed to run a bit funky. It would throttle up and down a lot for no apparent reason, and it wasn't normal for this machine. The fact I left the side off and it's still cool should allow plenty of cool air flow to go through the case. Windows also seemed like it would have to think a bit longer than usual in order to complete commands.

 

So, I took the other identical PIII and placed it in the slotket and put it into the Kayak and booted it up. It works perfectly. The system fan doesn't play around, it works normally, and Windows doesn't seem to act funky either. However, I've read where if a processor goes, it just goes and that's it. But is that really the truth, or could you have a damaged processor that still works but isn't reliable? I want to put it through a few tests now to push the processor to its limits to see what happens.

[Tripredacus]

However, I've read where if a processor goes, it just goes and that's it. But is that really the truth, or could you have a damaged processor that still works but isn't reliable?

In my experience, it is usually the CPU that is the most reliable part in a computer (except maybe the power cord). I've definately need CPUs operate in the way you describe, but then I've also seen where a CPU may act strange in one board and then fine in another board.

It was just recently a co-worker was working with a notebook that had problems working properly in WinPE. No part swaps fixed it except for changing the CPU. It was actually the first time I've encountered a scenario where the CPU was at fault for WinPE problems.... usually it is bad RAM.

[jaclaz]

Maybe the CPU has been contaminated since it has been in contact with a Dell

More seriously once upon a time (and I would not be so sure things have changed) processors were manufactured in the same manner and it was a post manufacturer factory test that made the selection (and the labeling) according to speed capabilities.

At least in those early days the manufacturing seemingly involved some kind of "magic" and each CPU came with slightly different capabilities.

I see as plausible that one of your two Pentiums if tested would result of a "lower grade" than the other, and by sheer coincidence/luck one motherboard can "deal" with it and another one cannot.

jaclaz

[TELVM]

.. I've read where if a processor goes, it just goes and that's it. But is that really the truth, or could you have a damaged processor that still works but isn't reliable? ...

 

Processors are very complex things, they aren't just black or white as in 'go' or 'no-go', there can be a lot of grey shades in between.

 

Moreover, reliability of high-mileage specimens like venerable PIIIs can be affected by 'diseases' like electromigration.

 

Might be that particular SL52R is just unreliable by now.

[Tommy]

Thanks for the tips, you guys. I've extensively tested out the other SL52R and it seems to be running more reliably. I played Majora's Mask on it and it was actually quite playable on it whereas it wasn't so much on the other SL52R, it just acted quirky. I myself would have to agree with TELVM as that's what I was thinking myself. LOL, jaclaz. Truthfully, I like the older Dell computers but because of the bad experience I had with a few of them, it might make me think a bit more before I'd purchase another one from them. And it's possible you could be right too. But something makes me wonder and it could be a simple matter of Googling it, but I noticed some Coppermines had full Intel printing on the chip whereas the ones I have have very minimal markings, almost like it was a different revision of the PIII. For example: http://www.pchub.com/uph/laptop/85-29928-7116/Intel-SL3XT-Pentium-III-600MHz-CPU-Old-Type-.htmlvs http://www.amazon.com/SL52R-Intel-Pentium-1-75V-Processor/dp/B001033DL4. Maybe voltage requirements falls into play? I don't really know.

[TELVM]

... I noticed some Coppermines had full Intel printing on the chip whereas the ones I have have very minimal markings, almost like it was a different revision of the PIII. For example: http://i.pchub.com/i/Intel-SL3XT-Pentium-III-600MHz-CPU-SL3XT-b-29928.jpg vs http://ecx.images-amazon.com/images/I/51FVX0-3oiL.jpg Maybe voltage requirements falls into play? I don't really know.

 

Don't worry that was just an aesthetic change, they dropped the Intel logo and switched to more discrete markings around 2001:

 

·····

 

No real significance, except it tells us if the CPU was manufactured prior to or from 2001 onwards.

Edited April 5, 2015 by TELVM

[Tommy]

I didn't think it really mattered much but since I didn't get into computers until the early Pentium 4 days, I never really got to learn much about Pentium IIIs too much. I like how the PIII is connected to a slotket. I was surprised how well the one I have works. I saw on ebay but a tad expensive but probably not so much for what it is, a tualatin adaptor. But I wonder if it would work on an existing slotket. I don't really see why not but then you never know with these sorts of things. For what I use my Kayak for, a 300MHz boost might be a bit useful.

[TELVM]

Some slotkets for Coppermine can run a Tualatin straightforward, provided some CPU pins are modified. My DIY pin-modified Tualatin 1400S on a MSI-6905 v2.0 slotket works like a charm. But modding the CPU is a hairy job.

 

There's a Korean guy who sells already modified Tualatins, ready for plug & play on said slotkets: http://www.ebay.com/itm/Tualatin-P-IIIs-1-4GHz-512K-include-Socket-Adapter-/281238323137

 

Not really worth the trouble nowadays, except for some crazy retrocomp enthusiasts.

[Tommy]

Some slotkets for Coppermine can run a Tualatin straightforward, provided some CPU pins are modified. My DIY pin-modified Tualatin 1400S on a MSI-6905 v2.0 slotket works like a charm. But modding the CPU is a hairy job.

 

There's a Korean guy who sells already modified Tualatins, ready for plug & play on said slotkets: http://www.ebay.com/itm/Tualatin-P-IIIs-1-4GHz-512K-include-Socket-Adapter-/281238323137

 

Not really worth the trouble nowadays, except for some crazy retrocomp enthusiasts.

Wow, thanks for the information! It's kinda worth it for me because I'm more of the classic junky of computers. I'd rather try forcing something to work on an older PC than something that's almost guaranteed to work on today's fast computers. I like the challenge!

[TELVM]

... I like the challenge!

 

That's the spirit!

 

If you decide to try the commando way, this is what I did to make a Tualatin work on a MSI MS-6905 v2.0 slotket on a 440BX board:

 

 

 

 

Pins AN3, AK4 and AJ3 need to be isolated from the slotket. Also the isolated pin AK4 needs to be bridged to live pin AN11.

 

More info and pics.

Edited April 7, 2015 by TELVM

[pointertovoid]

I confirm (my first profession a loooong time ago) that chips can and do have local defects, which are not observed immediately. One good test is "Intel Burn Test", but it needs WIndows running. I wish I knew a Cpu test booted from a diskette, just like Memtest86 does.

 

One one mobo, I had errors when the displayed image changed from clear to dark (supposedly the Agp that overloaded a tired power supply). On an other, mainly the antivirus gave false positives, the rest seemed to run fine.

Though, most chip failures are broad and observed early.

 

About the fan, could it be that pinout differences between both Cpu let read a wrong temperature that triggers a strong cooling?