Intel Pentium 66Mhz SX835??!!!

[nikko]

OK, first, I thought this is a remark, but after a series of cleaning (with alcohol, acetone and naphtha), polishing (using paper towel) and microscoping. It's indeed laser engraved. The thickness of this CPU is normal which means it's not sandpapered before the engraving.

The condition of this CPU is not good because I got it form a recycling company. Those CPUs were heading China for gold extraction.

If this is a fake (like a remarker's qualification sample or something), then I think I can just quit this business.

If this is a real one, then, the question is: how could this be possible??

(The last image was taken using a dissecting microscope.)

[JAC]

I can see 3 additional discrepancies.

The date codes differences are unusual.

The additional date code ( as I understand it) is significantly younger than the top one. Again, very unusual.

Also, on this I could be wrong, but when comparing it to my SX835 and SX836, I notice the font on the 5 on your cpu looks slightly different.

My only suggestion is to test it.

[Marcin]

my P100 gold cap SX963 have engraved prints by laser too ceramic looks a bit different but this can be picture fault.

[nikko]

Don't trsut the first picture; I shrinked it, so the fronts look funny. All fonts were engraved, no question about it. It is not thinner than other Pentiums. It is hard to sand the gold top ones, because the metal part would stand out (the edge adjacent to the ceremic part).

If this is a remarked one, then the remarker needs to find a non-engraved P60 (Socket 4)and then mark it as P66 so he can sell it at higher price. I don't think a non engraved goldtop P60 (Socket 4) exists.

And again, if this is remarked, it will changed my view of remarked CPU, because basically, all of the CPUs I won could be fake.

If this is a MQ80486 I won off eBay, at US$250 each, I would suspect it as remakred. But I got this one out off a recycling plant at a down-to-earth-cheap price. It basically is not aimming at collor's market. The only motivation to do such thing is to mark P60 to P66 so they can sell it at higher price during that time. However, laser engraving is not something you & I can afford at 1994. Even none of those later remarked one used laser to remark the CPUs (if those remarker can affrod a laser engraving machine, they probably would find a better business).

I will try to find a motherboard to test it, but what's the point?
If it appears as P60, then what's with the 80502-66?
If it appears as P66, then what's with the SX835????

Attached is the date code part. The finger print on the left is mine because my hand was wet.

BTW, what do I need to get this Pentium 60/66 running? I know I need a mother board, but what about RAM (PC100?) & Video card (VL-bus)? Do I need I/O card for mouse and keyboard? oh damn... I need a large P/S 2 keyboard.

[Marcin]

AT power supply unit, motherboard with Socket 4, AT keyboard with DIN connector, pair of same SIMM 72 pin RAM memory and ISA or PCI VGA card

[JAC]

You protest too much. I dont think it is a fake cpu, I think it is unusual.

Faking gold top pentiums is not impossible. It has been done before.
I have one.

Finding the cpu in a scrap yard now does not mean a cpu is not fake. The most recent location of a cpu does not increase or decrease it's legitimacy.


The purpose of testing it to get the CPUID, if possible. Advanced testing is also possible, heat/load tests to compare it to intel data, or in some cases op code testing.

[nikko]

No, the recent location doesn't prove its legitimacy, but it certainly is useful for determining remarker's intention (if this is a remarked one). If this is a remarked one, then certainly, it's a P60 and the intention was to make more money by marking P60 into P66. I don't really care if it's a remarked one because I didn't pay a future for it and I am not selling it anyway. As John mentioned to Paul (chipcollector) in the other thread, a misprinted CPU does not have increased value. CPUs are not like coins or stamps. A misengraved CPU is definitely unusual and fun to see, but I don't think it have much inceased market value (unless some collectors are into this, anyone?). I meant, other than us, the geeks (probably less than 100 in the world), who knows what SX835 is?

My father was asking me why am I phoning people and ask for an old Pentium system.
"It's waste of money", he said "paying for those junks."
I said "But I need to test one of my CPUs."
"Just a CPU?"
"Yeah.. it's kinda strange. it's a 66Mhz, but also got a SX835"
I guess the last line made my father speechless. Try this line to your friends and see. I bet the reaction will not be much different.

CPUID testing is certainly useful, but I doubt those other tests you mentioned is useful for distinguishing between P60/P66. The only difference between P60 and P66 I can think of is the core voltage. Even if it does, it will not solve the problem. The text on the CPU is already a contradiction. Even if we can determine the core, the question remains. Unless, we found the core turned out to be a hybrid, which is unlikely.

If CPUID turned out to be a P60, then the question of whether this is a remarked remains. If it's remarked, then whoever did it did a INCREDIBLE job. If it is not remarked, then that 80501-66 is definately deceiving.

If CPUID turned out to be a P66, then the SX835 now have a new definition.

About the discrepancies, I put my not-so-rare CPUs in the storage, so I will need to dig out the 80501-60 SX835 Gold Top w/ Processor logo to compare this weekend.

My personal guess is that: this is a P60 SX835 capable of running at 66Mhz and the text on top remains mystery (it should be marked as SX836 right?).

I am complaining mainly because it is HARD for me to find a Pentium Socket 4 system for me to test this one. I know some recyclers might have the motherboard and parts, but given my status in Taiwan, I am almost certain that I will get a "special", collector's rate (like 400% increase or something) when I inquire.

I am currently trying to locate a Socket 5 system, so I can concentrate on finding a Socket 4 motherboard only (since I will need RAM, Power, Video Card). Damn.... I will need a heatsink & fan too. Ahh.. almost forgot the floppy drives.

[debs3759]

CPUID will NEVER tell you EXACTLY what chip you have. The speed of a processor can not be determined from the CPUID (although I haven't checked what features are on more modern processors, to see when/if that changed...).

The only difference between two chips of the same stepping and different speeds is very often how fast they were able to operate without faults once the die had been produced and tested.

As such, I will always insist that CUPID can be used to obtain a model number, but other tests are needed to determine the speed at which a CPU _IS_RUNNING_, but no test will tell anyone what speed it was intended to be run at.

[FDIV]

Well everybody makes mistakes. I suppose that goes for the engravers at Intel as well. Does make you wonder whether they messed up the sx-835 or the 66 mhz.

[nikko]

Thanks for the suggestions.

Debs: So does that mean the CPUID can only tell me that my Pentium is a 80501? It cannot determine the S-Spec nor the designated speed?

If that is the case, then it would be pretty much useless for me to test the CPUs. I just found someone who has a Socket4 machine that I can test the CPU (not only this one but also the ESs I bought before).

Or anyone has any idea on how to verify this CPU? I was thinking maybe I can x-ray the engraved part, but I don't see how that can answer the questions.

[debs3759]

It can tell you that it is a 80501 and what stepping it is. Each stepping covers more than one speed, and sometimes more than one sSpec per speed as well. There are probably other tests that could be done for the sSpec but not for the rated speed. For the purpose of testing processors, I usually consider this enough, but if I come across more powerful tests for older processors I'll at least put it into the code I keep on my site.

Unfortunately, the only thing that determined whether a core went into the 60MHz part or the 66MHz was whether when testing the wafer, it was found to perform correctly at the higher speed.