First signs of AMD 65nm

http://www.xtremesystems.org/forums/showthread.php?t=108884
this is a discussion and SOME info, lot of B.S. too due to not-so-informed people posting

http://www1.oc.com.tw/forums/msgexc...msgid=0&notop=0&allfrom=214071&page=1&itype=1
this is the original link, if someone can find a better translation than what's being posted at xs, please post it

I got a bit better translation. Still rough, but better.



Although said AMD already and bought ATi
But, I but actually anticipated the future will be able to have any brand-new CPU to be able to appear
Should better be also does the cartography core inside CPU
Such indicator does not need to spend too many money......
First has a look this lead ------
65 奈 rice new edition AM2 CPU

From seemingly looks like is all entirely alike with the present AM2 edition
The base needle foot also is

But this is the 65nm system regulation edition
The inside core design should have the trifle to be different
If will be the AMD future new secret weapon
Makes one anticipate but actually very much
The specification part is 2GHz as well as 512K L2
Strange? I also think am, therefore looks from the outer covering serial number


The above OPN code is dissimilar with the present edition
This kind of edition should all be the early project version
Under first has a look the voltage
Only has 1.25V
Must lower some compared to the present edition (present looking like is the 1.4v appearance)
ES did not need to explain
Is the project version
Forgot (Kazak as for the middle serial number on everybody...)
Most above AMD big Logo does not use the explanation
Key lies in the first row of code
1DO2000IAA556
This also saw with the present FX62 is dissimilar (FX62 to look like is ADAFX62IAA6CS)
Because this edition at present the main engine board does not recognize.... Oh
The on hand material are not many
Therefore everybody first dry addiction
The known part is as follows,

1DO "2,000" IAA556 middle 2,000 refers to 2000+
According to understood
This 65nm edition actual frequency is 2GHz
In other words
2,000 figures are the 2GHz meanings
But this naming principle is different with AMD the consistent PR
Has AMD had to get rid of PR the naming method? ? Really is the question....

Most front 1DO is indefinite
Possibly must wait for the accurate material only then to be clear
Perhaps has other greatly to be allowed the question and answer also help to tell everybody
At present little brother Zha Te material as follows:
Usual everybody saw is ADA to (FX62 is ADA, but also some many Athlon64 edition also is)
But ADO is represents 65W the low power loss, ADA may see all at least is 89W perhaps FX 125W
But here is 1DO, the little brother guessed should represents 65nm next generation edition brand
Therefore only then can appear the first character is 1
Because if is Sempron, the first character is S, the Athlon64 series is A

Back IAA also has above FX62 sees
First I refers to the package seal, with FX62 is same
In addition A refers to the vcore voltage, but here but actually indicates 1.25V, compared with FX62 has lowered some
Final A is refers to T-case the temperature upper limit, probably (this I looked up material discovery below 70, picture is 5000+ also is IAA, but its T-Case is 55-70, but FX62 is 55-63)

Most back 556 at present knew this 65nm the edition is 512K L2 Cache x2
In first 5 is refers constructs 512K L2 x2, if is 6 that is 1M x2 (FX62 is 6, therefore is 1M x2)
But back 56 refers to Part Denfinition in the official explanation, at present did not know what 56 do represent thing
According to new AMD Roadmap demonstration
Should be the next generation edition
Moreover has the possibility also to have 2M L3 Cache
While convenient provides several Roadmap the picture to give everybody the reference


Here has writes

In 2006 will make the regulation to change course 65nm from 90nm
This is the 65nm edition


512K L2
Also many 2M L3 Cache

New core overhead construction design



Finally, the little brother types very laboriously
Helps me to call well
Has the big trouble which may supplement many to help
Thanks everybody

I've translated this a little further but I may have missed some points so refer back to the origional translation is something seems whacked! I did use info from other sources to provide some fill ins and IMO where I interjected my opinions.

Although I've already said, AMD bought ATI, I actually anticipate AMD will be able to release a brand new CPU. The *cartography on the CPUs die should be better and should not cost a lot. First looks at the new 65nm edtion AM2 CPU.

From appearance it looks exactly like the current AM2 CPU and pin layout is the same.

The upcomming 65nm standard edition will have a slightly differnent internal core design which makes one think it will be AMDs secret weapon. The engineering sample is 2GHZ and 512K L2 Cache. Strange looking serial number.

The above OPN code is not like the current editions. This version will be the early production runs and run at 1.25V compared to the 1.4V current versions.
(Said something about the middle of the serial number being differnt lost him on the keys and code but it's not like the FX62).
The mobo does not recognize this version being the first dry run of chips on hand.
Therefore the part is known as "1DO "2,000" IAA556 middle 2,000 refers to 2000+" It is our understanding that the 65nm version is actually 2GHz.
The naming principle is different, Did AMD have to get rid of naming PR? Is really the question.

As for what the first letters 1D0 is not a definit but will have to wait for actual chips to be sure. We see ADA on the FX62 and ADO on the Athlon 64 version. ADO represents 65W and low power loss. ADA is as much as 89W and the FX is 125W. His little brother thinks the 1D0 represents the next generation in the brand and therefore can only have a 1 in the first character. If it is a Sempron then the first character is S and Athlon64 series is an A.
(Ooo possible name change, IMO! )
The last letters, IAA in the FX62 as seen above (on the pictures). The I refers to the Package same as the FX62, First A is the VCore which is 1.25 lower than the FX62 and the last A refers to the T-Case temperature maximum limit. I Looked this up and found it to be below 70 as the Picture shows a 5000+ is also a IAA and it's T-Case is 55-70 and the FX62 is 55-63.

The 556 is known as the 65nm version and 512K L2 Cache X2. The first 5 refers to the 512K L2 X2. If the first is a 6 then it is 1M X2 ( The FX62 has a 6 and therefore 1M X2). The ending 56 refers to the Part Definition andthe official explanaition did not say what it represented.

According to the new AMD Roadmap, next generation could possibly have 2M of L3 cache and he provided some pics of these roadmaps for our convience to use as a reference.

2006 is set to change from 90nm to 65nm and this is a 65nm edition. We can expect 512K L2 but some versions will/may have 2M L3 and a new core design.

IMO - longer pipes, faster micro instructions, L3 cache for multiple core effiency we've seen in other articles.

New core overhead construction design

My stomach is tingling!!

dan

According to the AMD roadmaps, the change to 65nm is occurring late this year. Maybe a holiday release? Either way, I think all of this is excellent news.

After looking at the links, I can see where AMD may have made a last minute change in the die. The author mentioned "Cartography" which could be translated to roadmap of processors to come or the layout/mapping of the microcircuits.

Now for the goose-bumps!
What really got interesting is the pictures of the dies with one Xed out and a new larger one presented. The Xed out version is clearly a pictue of actual silicon possibly from an ES. The newer verion looks like it is from the drawing board/layout software. Several changes were made to the micro instruction prefetch and execute sections and a big change to the FPU. This I'm sure IMO, is because of the Conro. What I suspect is that these were changes beyound rev G possibly H. You normally want to take it a step at a time but in this case we need the holy grail to put Intel back in it's place. FPUs do more for Desktops with folding and running PI and MP encoding of which I give a carp. Games are more into just sheer processor power and make the difference in the ram and video access. The conroe only has help in the processing power. Consider the HTT our gateway to the PCIE buss, on servers it's also the MP gateway. If AMD steps up it's micro core, FPU, Ram and HTT then this will be our conroe! Slap on DDR3 with AM3 then watch out.

The smaller core size can also cause pressure to move things around, as heat sources might be closer to each other. In other words, just because you can see changes in the layout, it does not mean they have made huge changes to the logic of the cpu.