no 95W TDP AM2+
no 3 Core AM2+
no 6MB version (4x512KB+1x4MB)

to bad

Phenoms have often been over nine thousand in the past, but this development is truly ludicrous now.

Apparently, the table here is incorrect, hence the confusion. The one shown in the link at the beginning is right, though.

this is almost as bad as:

AMD ATI Radeon HD 4870 X2...

I quite like how the model number doesn't seem to distinguish the socket type, nor does there seem to be any reason why some of them end in 50.

I'm not exactly a fan of Apple, but this sort of stuff is exactly why they pretty much eclipse any other computer/electronics company in marketability.

Nobody remembers this kind of crap or knows what it means when they look at it, and almost no one is going to bother looking into it to find out. Just playing off the fact that people interested in computer parts are possibly "tech savvy" and will have no problem with it is assuming entirely too much.

I doubt very much that the 2.6 GHz Propus has 3 MiB of total cache. Propus has 0 MiB L3 and 0.5 MiB L2 per core, which means that like the 2.8 GHz Propus, it should have 2 MiB of total cache. I'm guessing ExpReview typoed it.

Edit: I note some people are having trouble with the numbering scheme:

Phenom X4 xxx50: Socket AM2+ (DDR2-only)
Phenom X4 xxx00: Socket AM3 (DDR3 in AM3 boards; DDR2 in AM2+ boards)
Phenom X4 20xx0: Deneb 4-core with L3 cache
Phenom X4 16x00: Propus 4-core without L3 cache
Phenom X3 14x00: Heka 3-core with L3 cache
Phenom X3 12x00: Rana 3-core without L3 cache.

Cyril, your chart has a mistake. The 20350 is AM2+, and the row below it, 20x00 at 2.6 GHz, is AM3.

Dear god, why?

I thought the X3 was only released to dispose of early Phenoms with the TLB cache problem. The bug was negated by disabling one of the cores.

I assume the 45nm model do not have the TLB problem so maybe this is a way to get rid of CPUs that have 1 faulty core (increases the yield).

OMG THESE ARE TEH FASTEST NOW!!!1!

"50" at the end means it uses the AM2+ socket and DDR2 memory as according to expreview. The chart here is not the same.

I wondered right after the first K10 releases why AMD started with 9x00 model numbers. It was easily predictable that they would run out of numbers quite fast.

Marketing people don't seem to think in longer term concepts - well, probably keeps their jobs safe... new numbering scheme every 2 years... what an awful time it must have been (for them) when CPUs were named from 8086 to 80486.

Silly AMD, tricks are for kids!

Besides, don't they know that when you run out of numbers, you start adding letters, not more numbers?

Ah why can't they call the crippled cores (no L3 cache) something other than Phenom, to reduce the confusion. They could call it the "Awesom" or something.

And 5 digit codes? Ugh. They're already way way too confusing.

As an aside: I see articles and news items on other techy news sites that just list code numbers for (Intel especially) CPUs now. Please don't do it here, they're meaningless to me, I can't mentally map from code number to "MHz, Cache Size, Number of Cores, etc", and I'd feel sorry for the person that could.

So there is a trade, right? Put EXTREME, in the name, or add more numbers.

I was hoping to hear a little more about Dual Core Phenoms. I don't think that dual core is quite dead and it's probably a good place for AMD to be competing.

Furthermore, even if AMD does hit these clock speeds, they're not going to catch up with Intel's best Penryn, which wouldn't be so bad if Nehalem wasn't coming around the corner.

If 45nm X4s out perform Core2 quads then I might end up with an AMD chip after all. That possibility surprises me.

Have to wait and see about power draw and overclocking.