PDA

View Full Version : Pentium Divider Flaw



wct097
February 28th, 2011, 11:00 AM
I'm sure everyone in this forum remembers the FDIV flaw. I grabbed this machine out of the back of a pickup just before it got tossed in the dump about 10+ years ago. Neat piece of nostalgia.

http://i298.photobucket.com/albums/mm266/wct097/DSCN9258.jpg

http://i298.photobucket.com/albums/mm266/wct097/DSCN9261.jpg

Relatively certain this is legit. The guy who discovered it worked (works?) for a local college. Wonder if it has any value. I've never plugged it in, nor do I know if the processor was replaced as part of the recall.

Chuck(G)
February 28th, 2011, 11:18 AM
I'm curious if you've taken a look at the CPU to see if it's one with the bug or if it's been replaced.

wct097
February 28th, 2011, 11:25 AM
Have not checked the CPU. Not really even sure how to do so. My assumption is that it was replaced.

linuxlove
February 28th, 2011, 11:32 AM
So what brand is this system?

wct097
February 28th, 2011, 11:42 AM
http://i298.photobucket.com/albums/mm266/wct097/DSCN9260.jpg

Micron

http://i298.photobucket.com/albums/mm266/wct097/DSCN9255.jpg
http://i298.photobucket.com/albums/mm266/wct097/DSCN9254.jpg

I'm guessing it was a higher end machine in it's day. Appears to have a SCSI hard drive.

http://i298.photobucket.com/albums/mm266/wct097/DSCN9259.jpg

Chuck(G)
February 28th, 2011, 11:54 AM
Have not checked the CPU. Not really even sure how to do so. My assumption is that it was replaced.

Usually, the heatsink is held on with a spring clip. If you don't see any, try unscrewing the fan and see if that gives any clues.

If this is a historic machine, I'd assume that the original chip would be left in.

If it boots Windows, here's an easy test (http://www.rtgsoftware.com/news/0195.htm).

wct097
February 28th, 2011, 11:59 AM
Usually, the heatsink is held on with a spring clip. If you don't see any, try unscrewing the fan and see if that gives any clues.

If this is a historic machine, I'd assume that the original chip would be left in.

If it boots Windows, here's an easy test (http://www.rtgsoftware.com/news/0195.htm).

Well, that depends on the good Doctor's thought process at the time. Did he immediately swap in the new chips Intel sent him? He probably used the computer for research (how he found the problem), and would need it functioning properly. Then again, he did sign it several years after the fact.

Will try plugging it in when I get home. I don't have an AT keyboard. Will poke around the office and see if I can find a PS/2-AT adapter somewhere.

Unknown_K
February 28th, 2011, 12:01 PM
Are the buggy chips worth much? I have a few early P60/P66 chips that might have the bug.

Neon_WA
February 28th, 2011, 05:30 PM
Are the buggy chips worth much? I have a few early P60/P66 chips that might have the bug.

Some are worth $50+ due to their sSpec rarity (not due to FDIV bug) the more common FDIV sSpecs go for $15

I would be interested in what sSpecs you have :-)
I have 16 of the 27 production sSpecs (socket 4, 5 & TCP) containing the FDIV bug

Neon_WA
February 28th, 2011, 05:37 PM
I'm sure everyone in this forum remembers the FDIV flaw. I grabbed this machine out of the back of a pickup just before it got tossed in the dump about 10-15 years ago. Neat piece of nostalgia.


Wonder if it has any value.

I would think so... it would to me.. as FDIV items is one of my aims in my collecting
The system itself isnt worth much.. but its place in history is its value

There are a few FDIV collectors.. but they keep to just the chips

I live in Australia.. but I have a friend in Oregon that ships items to me that I buy off US ebay
Will cost me a mint to ship (any idea of weight?)

... but if you do consider selling.. let me know

Unknown_K
February 28th, 2011, 07:20 PM
Where/how do I look up the sSpec?

Caluser2000
February 28th, 2011, 07:45 PM
Interesting http://www.trnicely.net/pentbug/pentbug.html It appears the chip was replaced on 4 Nov 1994 and tested fault free. Though it may be possible the original chip was refitted for historic reasons. There's a calculation on the page to work out whether or not the chip is a faulty one or not.

Neon_WA
February 28th, 2011, 08:53 PM
Where/how do I look up the sSpec?


type the sSpec into identify section on this page
http://www.cpu-world.com/search.html

then click on sSpec for more information particular to that sSpec

but here is a list of all production chips with FDIV


Socket 4
SX753 A80501-60
SX835 A80501-60
SX842 A80501-60
SX926 A80501-60
SZ949 PCPU5V60
SX754 A80501-66
SX828 A80501-66
SX836 A80501-66
SX837 A80501-66
SZ950 PCPU5V66

Socket 5
SX878 A80502-75
SX874 A80502-90
SX879 A80502-90
SX885 A80502-90
SX909 A80502-90
SX921 A80502-90
SX922 A80502-90
SX923 A80502-90
SX942 A80502-90
SX943 A80502-90
SX944 A80502-90
SZ951 PCPU3V90
SX886 A80502-100
SX910 A80502-100
SX956 A80502-100
SX960 A80502-100

TCP
SX951 TT8050275

Chuck(G)
February 28th, 2011, 09:08 PM
Without the failing chip, it's just another Micron P1 system, it seems to me.

wct097
March 1st, 2011, 06:46 AM
I would think so... it would to me.. as FDIV items is one of my aims in my collecting
The system itself isnt worth much.. but its place in history is its value

What would be the most appropriate way to determine it's value?



I live in Australia.. but I have a friend in Oregon that ships items to me that I buy off US ebay
Will cost me a mint to ship (any idea of weight?)

No clue, but if it came down to it, I could obviously price it.


... but if you do consider selling.. let me know

Will do. My biggest problem there is assessing a value. I think my first step will be to see if I can find an AT keyboard or adapter and see if I can fire it up.


Without the failing chip, it's just another Micron P1 system, it seems to me.

That could very well be the case. Again, I have no basis for estimating what it's value (if any) is. I suppose I could email the professor for details and to verify it's authenticity.

Raven
March 1st, 2011, 06:48 AM
I recommend powering up the PSU disconnected from the system before powering it up. This way if anything is wrong with the PSU you'll know before you subject the rest of the system to it.

kishy
March 1st, 2011, 07:23 AM
Didn't I read somewhere on this site that powering up a PSU with no load can damage it?

Chuck(G)
March 1st, 2011, 08:57 AM
No, but most PSUs won't function without a load. Hook an old hard drive or four to one when powering up. Most PSUs will simply die if they're bad. The chances of one actually ruining anything are very small.

hargle
March 1st, 2011, 09:53 AM
Well, that depends on the good Doctor's thought process at the time. Did he immediately swap in the new chips Intel sent him? He probably used the computer for research (how he found the problem), and would need it functioning properly. Then again, he did sign it several years after the fact.

Wait, are you saying you've got the machine that the bug was originally discovered on? I apparently work with a guy who worked with the guy who discovered how the problem was introduced in a logic lookup table of some sort. Apparently my co-worker's previous co-worker was the one who tracked it down enough for intel to actually verify that there was a problem. It all sounds very urban legend to me, but I suppose if you're in the tech industry the odds of running into someone like this are pretty good...

Chuck(G)
March 1st, 2011, 10:14 AM
You know, instead of speculating about this, why not drop the good Doctor an email? (http://www.trnicely.net/). Who knows? He may be interested in getting this system back...

wct097
March 1st, 2011, 12:12 PM
Wait, are you saying you've got the machine that the bug was originally discovered on? I apparently work with a guy who worked with the guy who discovered how the problem was introduced in a logic lookup table of some sort. Apparently my co-worker's previous co-worker was the one who tracked it down enough for intel to actually verify that there was a problem. It all sounds very urban legend to me, but I suppose if you're in the tech industry the odds of running into someone like this are pretty good...

Exactly. That's what it written & signed on the inside of the case at least. Dr. Nicely was a professor at Lynchburg College. I live in Lynchburg, have never met the man, but came across this when my brother was helping his former boss clean out his basement and take a bunch of things to the dump. The former boss, I'm told, was friends with Dr. Nicely.


You know, instead of speculating about this, why not drop the good Doctor an email? (http://www.trnicely.net/). Who knows? He may be interested in getting this system back...

Scored an AT-PS2 adapter today, so I'm going to try to power it up and see what the story is. If it doesn't boot or I can't figure out if the processor was replaced I may drop him an email.

Unknown_K
March 1st, 2011, 03:41 PM
Checked the loose P60 I have and it doesn't have the FPU bug, the other chips are in systems (One packard Bell and one PS/2 Model 90) so I will check those later.

wct097
March 1st, 2011, 03:57 PM
It booted up fine. Running Windows 3.1. Don't have a serial mouse, so I could only fiddle with as much as I could get to using the keyboard. Would seem that it no longer has the original flawed chip in it, but that's not surprising. Didn't really expect it to.

http://i298.photobucket.com/albums/mm266/wct097/IMAG0142.jpg

k2x4b524[
March 1st, 2011, 06:54 PM
no, but if this is the maching that DID discover the bug, it may have some sentimental value to him. I know if it were me that were in his shoes and i was reading this thread, i may want it back, besides, that hard drive may have some of his original work on it, shoot him a line, nothing bad can come of it short of he's not interested in it anymore, or even worse, he is deceased or something.

wct097
March 2nd, 2011, 06:31 AM
no, but if this is the maching that DID discover the bug, it may have some sentimental value to him. I know if it were me that were in his shoes and i was reading this thread, i may want it back, besides, that hard drive may have some of his original work on it, shoot him a line, nothing bad can come of it short of he's not interested in it anymore, or even worse, he is deceased or something.

I'll drop him an email. I doubt it has a sentimental value to him as I believe it was given/sold to someone else before heading to the dump after a basement cleanout. I don't think it was unintentionally lost or stolen.

wct097
March 3rd, 2011, 09:33 AM
Dr Nicely replied to me and verified the authenticity of the machine. An excerpt from his response:


The flawed processor was swapped out when
Intel sent me a corrected replacement in
early November of 1994. I then donated the
flawed processor (along with a few other
new and corrected ones that Intel sent) to
my employer, Lynchburg College, which they
used to construct new systems at a discount.
A flawed processor from a machine owned by
one of my colleagues in the Mathematics
Department was also donated, after we
swapped one of the new processors into
his system. The two systems constructed
by the College with the flawed processors
remained in service (in the student labs)
for several years. I had a standing offer
during that time of (considerable) extra
credit for any of my students that could
demonstrate an error caused by the flaw
(other than a variation of the ones already
known). Nobody ever collected. I assume
the College eventually junked those two
machines (I retired in July, 2000).

Anyone know what the going price for a flawed Pentium is? Anyone have a non-rare version they'd be willing to part with?

Neon_WA
March 3rd, 2011, 02:26 PM
Anyone know what the going price for a flawed Pentium is? Anyone have a non-rare version they'd be willing to part with?

A ceramic P60 about $20 - 30 (sSpec SX753)
A Gold top P60 about $15 without the word "Processor".. with about $25 (sSpec SX835)

big issue at the moment.. is gold scrappers have pushed prices up

wct097
March 4th, 2011, 02:18 AM
A ceramic P60 about $20 - 30 (sSpec SX753)
A Gold top P60 about $15 without the word "Processor".. with about $25 (sSpec SX835)

big issue at the moment.. is gold scrappers have pushed prices up

Would you be willing to sell me one of the cheaper 60mhz versions, so that I can install it in this PC to give back to Dr Nicely?

Neon_WA
March 4th, 2011, 03:29 PM
shipping from Oz isnt cheap.. but I will see what I can pick up on US ebay. I have some CPU lots on the way to my friend in Oregon that may have one in it.
But if not... one will come up soon

wct097
March 17th, 2011, 06:08 AM
Nobody has a flawed Pentium chip they're willing to part with?

luckybob
March 17th, 2011, 11:03 AM
check my "do you have ocd" thread. I just picked up a huge pile of processors, one my be the s-spec you are looking for.

per
March 17th, 2011, 12:22 PM
check my "do you have ocd" thread. I just picked up a huge pile of processors, one my be the s-spec you are looking for.

According to Wikipedia, a total of 39 different s-specs are affected by the bug. See http://en.wikipedia.org/wiki/Pentium_FDIV_bug#Affected_models for a complete list.

Neon_WA
March 17th, 2011, 04:43 PM
the last 2 I got in CPU lots were both uncommon sSpecs.. one worth $50+ the other $100+ and I dont think you want to part with that much

but still keeping my eyes open for a common one

Neon_WA
March 17th, 2011, 04:53 PM
check my "do you have ocd" thread. I just picked up a huge pile of processors, one my be the s-spec you are looking for.

the machine originally had a P60

and you have 2 SX835s that would do the job. The one with missing print probably his cheapest choice


According to Wikipedia, a total of 39 different s-specs are affected by the bug. See http://en.wikipedia.org/wiki/Pentium_FDIV_bug#Affected_models for a complete list.

dont believe everything you read on Wiki... there are many missing off that list

per
March 18th, 2011, 12:18 AM
dont believe everything you read on Wiki... there are many missing off that list

Of course, but these 39 are at least known to have the bug. Intel never provided a full list of affected s-specs, so the only way to actually make a complete list is to test every single s-spec CPU of the early Pentium series. This is quite a lot of CPUs, and it would be very difficult for somebody to get every single one.

Neon_WA
March 18th, 2011, 12:33 AM
Of course, but these 39 are at least known to have the bug. Intel never provided a full list of affected s-specs, so the only way to actually make a complete list is to test every single s-spec CPU of the early Pentium series. This is quite a lot of CPUs, and it would be very difficult for somebody to get every single one.

I am aware of 57 (30 qSpec & 27 sSpec)
so far I have only managed to get a total of 16 for my collection as the rarer ones now go for $400+

per
March 18th, 2011, 05:53 AM
I am aware of 57 (30 qSpec & 27 sSpec)
so far I have only managed to get a total of 16 for my collection as the rarer ones now go for $400+

Then I would suggest that the tables on Wikipedia should be updated.

By the way, how do you know if one is rare or not? I guess the one I have (SX835 if I remember correctly) is a common one as it already is in the table.

wct097
March 23rd, 2011, 03:27 AM
Turns out that my coworker has a display case with a variety of old processors in it. He thinks two of them are the flawed versions. We're going to check today to see if any of them work and are indeed the flawed variety.

The question is; does an original Pentium need a heatsink & fan? The processor that is installed has one, but it appears to be glued/attached directly to the processor and not readily removable. The chips we have available don't have heatsinks attached.

wct097
March 23rd, 2011, 04:49 AM
Found an SX835 chip that's been in my coworker's office for years in a display case with a variety of other processor chips. Installed in the machine and tested positive as flawed. Going to have to chase down a proper heatsink & fan now. The one in the machine now isn't properly secured.

Neon_WA
March 23rd, 2011, 08:19 PM
Found an SX835 chip that's been in my coworker's office for years in a display case with a variety of other processor chips. Installed in the machine and tested positive as flawed. Going to have to chase down a proper heatsink & fan now. The one in the machine now isn't properly secured.

Good one :-) if produced before week 19 1994 it should have the "Processor" marking
most socket 4s only had heat-sink, but one with fan would be better as they run so hot.. but not as hot as the P66

Neon_WA
March 23rd, 2011, 08:29 PM
Then I would suggest that the tables on Wikipedia should be updated.

By the way, how do you know if one is rare or not? I guess the one I have (SX835 if I remember correctly) is a common one as it already is in the table.

I have signed up to Wiki and updating the tables is on my list of things to do.

As for rarity.. Some non FDIV socket 4s are rarer, while some FDIV socket 5s are even rarer still. But basically any socket 4 or 5 qSpec is rare.. FDIV or not.

It is a bit hard to pin down rarity.. like SX836 FDIV P66 a year ago.. there was only one known to exist
I have 4 and there is about 8 in other people's collections

Neon_WA
March 23rd, 2011, 09:44 PM
here is a basic list of all FDIV Pentiums


Q0280*(Pentium* A INTEL-P5)*
Q0283*(Pentium* A INTEL-P5)*
Q0305*(Pentium* PENTIUM(TM))*
Q0315*(Pentium* PENTIUM(TM))*
Q0335*(Pentium* PENTIUM(TM))*
Q0336*(Pentium* PENTIUM(TM))*
Q0352*(Pentium 60MHz*A80501-60)*
Q0353*(Pentium 66MHz*A80501-66)*
Q0394*(Pentium 60MHz*A80501-60)*
Q0395*(Pentium 66MHz*A80501-66)*
Q0399*(Pentium* PENTIUM(TM))*
Q0400*(Pentium 60MHz*A80501-60)*
Q0412*(Pentium 60MHz*A80501-60)*
Q0413*(Pentium 66MHz*A80501-66)*
Q0466*(Pentium 60MHz*A80501-60)*
Q0467*(Pentium 66MHz*A80501-66)*
Q0540*(Pentium 75MHz*A80502-75)*
Q0541*(Pentium 75MHz*A80502-75)*
Q0542*(Pentium 90MHz*A80502-90)*
Q0543*(Pentium 90MHz*A80502-90)*
Q0563*(Pentium 100MHz*A80502-100)*
Q0587*(Pentium 100MHz*A80502-100)*
Q0601*(Mobile Pentium 75MHz*TT8050275)*
Q0606*(Mobile Pentium 75MHz*TT8050275)*
Q0611*(Pentium 90MHz*A80502-90)*
Q0612*(Pentium 90MHz*A80502-90)*
Q0613*(Pentium 90MHz*A80502-90)*
Q0614*(Pentium 100MHz*A80502-100)*
Q0628*(Pentium 90MHz*A80502-90)*
Q0677*(Pentium 100MHz*A80502-100)*
SX753*(Pentium 60MHz*A80501-60)*
SX754*(Pentium 66MHz*A80501-66)*
SX828*(Pentium 66MHz*A80501-66)*
SX835*(Pentium 60MHz*A80501-60)*
SX836*(Pentium 66MHz*A80501-66)*
SX837*(Pentium 66MHz*A80501-66)*
SX842*(Pentium 60MHz*A80501-60)*
SX874*(Pentium 90MHz*A80502-90)*
SX878*(Pentium 75MHz*A80502-75)*
SX879*(Pentium 90MHz*A80502-90)*
SX885*(Pentium 90MHz*A80502-90)*
SX886*(Pentium 100MHz*A80502-100)*
SX909*(Pentium 90MHz*A80502-90)*
SX910*(Pentium 100MHz*A80502-100)*
SX921*(Pentium 90MHz*A80502-90)*
SX922*(Pentium 90MHz*A80502-90)*
SX923*(Pentium 90MHz*A80502-90)*
SX926*(Pentium 60MHz*A80501-60)*
SX942*(Pentium 90MHz*A80502-90)*
SX943*(Pentium 90MHz*A80502-90)*
SX944*(Pentium 90MHz*A80502-90)*
SX951*(Mobile Pentium 75MHz*TT8050275)*
SX956*(Pentium 100MHz*A80502-100)*
SX960*(Pentium 100MHz*A80502-100)*
SZ949*(Pentium 60MHz*A80501-60)*
SZ950*(Pentium 66MHz*PCPU5V66)*
SZ951*(Pentium 90MHz*PCPU3V90)*