Fake Cyrix FasMath?

[feipoa]

There are two Cyrix FasMath CX-83D87-40-GP FPUs shown in the accompanying image. The FPU on the left I bought many years ago from within N. America. The FPU on the right I bought recently from eBay, with source location as China. They are listed as new, old stock here, http://www.ebay.ca/itm/162201026922

Does the font on the NOS FPU look odd? Did Cyrix change their font, or is this a counterfeit FPU?

I have benchmarked the new FPU and the FPU-based results are consistent with that of other 40 MHz FasMath FPUs (pictured on the left). IIT, ULSI, and Intel FPU's for the 386 platform have distinct benchmark results, so I know this is not a counterfeit CPU from one of those brands.

I bought this FPU, and paid a bit over market value in my opinion, because it was NOS. I was hoping to overclock it up to 50 MHz. To my delight, it held up well at 50 MHz; it was able to complete the Quake benchmark and several others.

[CPUShack]

the -40 is definitely odd, added afterwards, its possible Cyrix themselves remarked it, its a very late FPU, 9352, probably why it overclocked so well

[feipoa]

What do you think about the Cyrix USA? Doesn't it seem like there is too much spacing between the letters?

[CPUShack]

No, thats consistent with others I have seen, different assembly plant to different font not unusual there

[feipoa]

Attached is a photo of one with an even later datecode. The font looks like the later one I posted with "fake?", except the 40 looks correct.

[CPUShack]

yah looks good

[feipoa]

There is one other element of the 3 chips that I noticed which may or may not be of significance. If you look at the big Cyrix white paint on "real?", you will see there is this a lack of uniformity in the white, almost like a grain. Chip with "fake?" does not have this. Both of these chips were scanned on the same scanner right next to each other.

Now if you look at the 3rd FasMath, there is also this lack of white uniformity in the big Cyrix. Is this of any significance? All my other Cyrix FasMath chips, including a QFP piece, has this same lack of white uniformity in the big Cyrix.

[feipoa]

I looked at the "40" through 25x magnification and noticed that the "40" is a slightly different colour compared to the rest of the text. The rest of the text is a slight yellowish tint compared to the "40". I'm not sure what can be derived from this except that the "40" was deposited at a different time compared to the other text.