Fake socket 754 CPUs (AEP5AP) or not?

[KK_Fin]

I don't want to hijack Dome's thread about ADA3700AEP5AP so I made my own thread.

I was looking visual things in AEP5AP's (from my own collection and also from websites) and there were some 3200's and 3400's which looked like "normal" Athlons but especially 3000's and 3700' had something strange in markings.

I collected some of my findings in one picture, #1, #2, #7 and #8 have same matrix code and also font changes in serial number. #3 (AJ) have an extra space in serial number. Matrix code is missing in #4 and #5. #6 looks good, but serial number looks weird.

#5, #6, #7 and #8 are from cpu-world.com (so there is better pictures available), rest are from other places. #7 and #8 are kept fakes so they are mostly for comparison.

Are all these fakes or not? What do you think?

[socket]

1 - fake
2 - fake
3 - pre-production C0 sample. Same space between codes you can see on ADAAA2800ACN5
4 - eng sample
5 - first "production" version, no idea why it's missing matrix, earlier CPUs have it, but I would say it depends on factory where it was packaged.
6 - production version, it looks weird because there were little discrepancies between factories. Notice it also has black placer for labeling of mobile versions - but I'm not 100% sure it's not fake
7,8 - fakes.
btw notice difference between 1 and 2, despite being manufactured (according to fake labellers) 30 weeks apart, on different color of substrate, serial numbers are aaaaalmost the same

[xsecret]

1 - fake
2 - fake
3 - Genuine ES
4 - Genuine QS
5 - Early production
6 - Highly suspicious
7 - fake
8 - fake

So almost agree with socket. For 6-, without a dump, I'll say 80% chance fake, 20% chance odd production part.