Technical Question

RedFury · August 18th, 2002, 02:59 AM

Something struck me today, and I'm at a loss to figure out how to inform myself about it, so I turn to you guys.

What exactly is the difference between a ( for example ) 1600+ AMD and a 2100+, other than the obvious I mean.

What is it exactly that allows a faster chip to run faster than another chip in the same family ( XP for example )

Obviously we can overclock a processor, but why? And why does it heat up more than at a stock setting?

I'm a pretty logical thinking guy, but this escapes me. Perhaps it is beyond my scope and I should just go back to things I know, like making pancakes and fixing my cars.

OuTpaTienT · August 18th, 2002, 04:10 AM

The main difference is one can run faster than the other.

When cpu's are made they are then tested and given a speed rating based on how fast they test. And as the manufacturer continues to fabricate the same family of chips they get better at the process as they go along, thus they start producing more high quality chips which means faster chips.

Also, from the beginning of the fabrication process they might be getting mostly slower chips (1500, 1600, etc) but they also will get a number of really good faster chips. And what they do is stockpile the faster ones until they are ready to release that speed of cpu.

At least that's my understanding of it. I'm sure you can get much more technical about what goes on, but in a simplistic nutshell that's it.

They release the chips that the market is calling for. For example, many, if not most, of the old Slot-A Athlon cpus were actually much faster chips than they were sold as. This was because AMD got such a high quality chip from their process they actually didn't have enough slower chips to fill the market demand. So they took their chips and marked them down a few speed ratings and sold them. I bought a 500Mhz Athlon and upon opening up it's case I discovered it was actually a 600Mhz chip. Then not too long after that I got a 700Mhz Athlon that turned out to be a 900Mhz chip. Ahhhh, the good old days.

RedFury · August 18th, 2002, 11:41 AM

You're kidding, right?

Geez, I figured it would be more technical than that

Theophylact · August 18th, 2002, 04:42 PM

It's an old story.

The Intel 486 came in two versions: the 486, with math coprocessor, and the 486SX, "without" coprocessor. The chip itself was the same (why have two fabrication facilities?) but one version had an extra pin to enable (or disable, I can't remember which) the coprocessor.

Back in the days when IBM didn't sell you its machines, only lease them (that is, before the antitrust consent decree), they had a punch card reader that came in two versions: a less-expensive, slower one and a more-expensive, faster one. You could pay for an upgrade, though, and if you did, what happened was that an IBM systems engineer would come, open a side panel, and switch the drive belt from one set of pulleys to another.

It's the kind of thing that lends credence to urban legends like the 200 Mpg Carburetor.

John Prophet · August 18th, 2002, 05:30 PM

Is that what AMD is doing now with the XP and MP? I have seen the threads showing how to make an XP SMP compatible by connecting the certain bridges on the chip.

Theophylact · August 18th, 2002, 05:51 PM

Basically, yeah.

PartsMan · August 18th, 2002, 05:54 PM

Quote:

Originally posted by Theophylact
It's an old story.

The Intel 486 came in two versions: the 486, with math coprocessor, and the 486SX, "without" coprocessor. The chip itself was the same (why have two fabrication facilities?) but one version had an extra pin to enable (or disable, I can't remember which) the coprocessor.

REALLY? The story I heard was that the difference between the 486DX and the 486SX was that all 486s had math co-processors but the ones labeled and sold as SXs the math co-processors had failed in the QC process .... so instead of throwing them away they just renamed them sold them for a slightly reduced price and sold the ones that worked properly for more $ as DXs.

Epyon9283 · August 18th, 2002, 05:58 PM

Quote:

Originally posted by PartsMan

REALLY? The story I heard was that the difference between the 486DX and the 486SX was that all 486s had math co-processors but the ones labeled and sold as SXs the math co-processors had failed in the QC process .... so instead of throwing them away they just renamed them sold them for a slightly reduced price and sold the ones that worked properly for more $ as DXs.

Thats what I heard too.

Theophylact · August 18th, 2002, 06:07 PM

No, there was a different pinout, though there seems to be a dispute as to whether this actually constituted crippleware. See here and also here.

OuTpaTienT · August 19th, 2002, 06:06 AM

One of those links says "OverDrive - 486DX-2 with a slightly different pinout for use in 486SX systems."

I had a few Intel 486DX-2 OverDrive CPUs and they were completely interchangable with a true Intel 486DX-2 CPU. Not only could the hardware not see the difference between the OverDrive & non-OverDrive chip, but extensive benchmarking proved them to be identical in performance numbers.

However I never owned an "SX" machine so I couldn't do any testing there (that's IF there were any SX specific motherboards...I don't recall.)