I have gold top 90.

Cheers,

EvieSigma wrote on 2020-06-07, 23:30:

I wonder, did they make gold top Socket 7 Pentiums or just Socket 5? I'm pretty sure Pentium 75 and 100 existed in both Socket 5 and Socket 7.

"Socket" is a property of motherboard, not CPU. Socket 5 has 320-pins, Socket 7 has 321 pins.

Regular Pentiums have 296 pins (no matter if PGA, CPGA or FCPGA) so they all fit in either socket. And yes. 60-100 MHz do exist in both the gold top or ceramic variant.

Any Pentium CPU that has staggered PGA that doesn't support 2.5 and 3.0 multipliers was designed for Socket5. P75, P90 and P100 for certain. Not sure about P120 and P133. However, Socket7 is backward compatible with Socket5 chips, so there's no problem there. Socket7 CPUs (P150,166, 200 and MMX chips) on Socket5 motherboards is another story. The easiest way to do it is with one of those Evergreen Spectra adapters. But, if you use a Socket7 CPU without split rail voltage, you can probably modify your motherboard to accept them.

mpe wrote on 2020-06-08, 07:22:

EvieSigma wrote on 2020-06-07, 23:30:

I wonder, did they make gold top Socket 7 Pentiums or just Socket 5? I'm pretty sure Pentium 75 and 100 existed in both Socket 5 and Socket 7.

"Socket" is a property of motherboard, not CPU. Socket 5 has 320-pins, Socket 7 has 321 pins.

Regular Pentiums have 296 pins (no matter if PGA, CPGA or FCPGA) so they all fit in either socket. And yes. 60-100 MHz do exist in both the gold top or ceramic variant.

Thanks, I didn't know all Pentiums had the same pin count so I thought maybe Socket 7 chips had an extra pin.

Yes. It is tricky to consider Pentium CPUs as being of "Socket 5" or "Socket 7" type as socket standards are separate from CPU requirements.

When Intel say "the P120 has been designed for Socket 5 and P133 for Socket 7", they say so because they know that that a motherboard that adheres to Socket 7 specification must be able to supply enough power draw to support P133. Not because it requires any S7 features. Similarly, the P200 MMX requires a split-voltage with core one 2.8V, but not necessarily a Socket 7 (although that's usually the case as the split voltage delivery is part of the S7 spec and virtually no one was putting PGA-296s or Socket '5s any more when the MMX came to market).

I highlighted the differences here:

On this picture, we have Pentium 233 MMX which is PGA-296 CPU (just like all other Intel Pentiums), Pentium 166 Overdrive (PGA-320 orSocket 5) and AMD K6 (PGA-321 or Socket 7).

Technically only the K6 is a "Socket 7 CPU" that wouldn't fit in any other socket (PGA-296 or Socket 5). And even then the extra pin has no electrical function whatsoever.

Yep, All of my AMD K6 have the same pins as your middle and right pictures to ensure Socket 7. If you look close it looks like the POD 166 does have the "mid" pin opposite of the "tang" pin like the AMD (added: or there is a place it could be if not there, there is a gold spot)

i still cant find a reliable tutorial on how to detect fake cpus from that era with pictures and everything

for example on this topic it is mentioned that the fake cpu had a darker color ceramic while the original has like a light color ceramic

while other people say that the light colores ceramic is probably fake.

I would think it was only worth faking common CPUs when they were still new since the prices varied quite a bit between speeds. These days people probably only fake high value rarities.

Never heard of fake CPUs, especially not during that era. Can't have been that common, but keep in mind it would be really easy to buy a CPU now, and make it look like a fake, then try to sell it as a "rare fake," or use one cpu to fake another more expensive one in the present day. Just saying.

midicollector wrote on 2023-09-18, 00:32:

Never heard of fake CPUs, especially not during that era. Can't have been that common, but keep in mind it would be really easy to buy a CPU now, and make it look like a fake, then try to sell it as a "rare fake," or use one cpu to fake another more expensive one in the present day. Just saying.

It certainly would be rarer back then, given online commerce wasn't really a thing, and you'd be likely to go through a reputable dealer/PC store, that would get the parts through official channels.
If they sold fakes, it would have been a big deal, given you're protected by your local laws (unlike today if you're buying from countries with looser laws, that are hard to get enforced if you're not a citizen there).

That said, sellers on shady markets selling bootlegs was a thing, especially in poorer countries were hardware was too expensive for most to be able to afford, and fake items would be more likely to pop up there.
That's also were items like those "fake" Pentium MMXes that were mobile chips stuffed into an adapter for Socket 7 use would be more likely to be sold (though those could actually be very decent, given laptop chips were generally better binned chips than desktop chips, specifically because they needed to be energy efficient and cool running).

I think that faking CPUs was part of the reasons why Intel started to lock the multiplier back in the day ...

SETBLASTER wrote on 2023-09-17, 20:40:

i still cant find a reliable tutorial on how to detect fake cpus from that era with pictures and everything

for example on this topic it is mentioned that the fake cpu had a darker color ceramic while the original has like a light color ceramic

while other people say that the light colores ceramic is probably fake.

The easiest way is to first check the part number and see if it actually exists.
I have a cermic SL27X when it should be PPGA (plastic)... 🤣
And a SL23W marked 233MHz (when the real one is 200MHz).

Check out my thread where I talk a little more about them and their multipliers.

The ceramic fakes are easy to identify because the fake paint starts to chip off if you clean it with isopropyl alcohol (hard).

The typography/ink is slightly different, and if you have two chips (fake and original) with the same part number, you can tell there's some difference.