Print 80 comment(s) - last by maynardc.. on Dec 11 at 3:20 PM

Athlon 64 X2 gets the 65nm treatment

AMD today announced its long-awaited 65nm Brisbane core Athlon 64 X2 processors. This begins AMD’s transition to 65nm from 90nm. AMD expects to transition its complete product lineup to 65nm in mid-2007. Nevertheless, four Brisbane core Athlon 64 X2 processors are launching today—the 5000+, 4800+, 4400+, and 4000+. All 65nm Brisbane core processors are equipped with 2x512KB of L2 cache.

AMD Athlon 64 X2 Brisbane 
L2 Cache
TDP Pricing
5000+ 2.6 GHz 2x512KB 65W
4800+ 2.5 GHz 2x512KB 65W $271
4400+ 2.3 GHz 2x512KB 65W $214
4000+ 2.1 GHz 2x512KB 65W

With Brisbane, AMD Athlon 64 X2 dual-core processors equipped with 2x1MB of L2 cache will become a thing of the past as AMD completes its 65nm transition. While AMD has announced its Brisbane core 65nm processors, the processors only available in limited quantities. Expect mass availability in Q1’2007.

OEMs are expected to deliver Brisbane powered systems as well. These OEMs include Acer, Dell, Founder, Gateway, HP, Lenovo, Packard Bell and TongFang. Dell is expected to adopt Brisbane 65nm processors in its Optiplex and Dimension desktops while Acer is expected to have Brisbane powered Aspire systems too.

Pricing for Brisbane 65nm processors start at $169 for the 4000+ and goes up to $301 for the 5000+ in quantities of 1000. Do note that Brisbane core processors are still K8 based and not K8L like the upcoming Stars core processors.

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

Why the hate
By ajfink on 12/5/2006 10:43:24 AM , Rating: 3
Why is there so much angst over this? It's just the die shrink everyone knew was coming for the past year. Clock for clock they'll probably perform exactly the same as their 90nm counterparts. I'm curious, though, as to whether these chips fixed the annoying memory controller thing where certain speeds of RAM performed better with certain speeds of processors.

It will be interesting, also, to see how a 65nm 4000+ stacks up again a 90nm 4000+, noting the 100Mhz discrepancy in clock speed. I wouldn't expect these first generation 65nm to overclock extremely well, but the rumor is that their 65nm process turned out at the higher end of AMD's expectations in terms of quality. So, we'll just have to wait a few weeks for the benchmarks to start rolling in.

And it does take time for a process to mature, and AMD would like very much to roll out their Stars cores on a more mature 65nm than one that still has places it could be improved. Not like they're ready to produce them now, anyway.

RE: Why the hate
By mino on 12/6/2006 1:59:15 AM , Rating: 2
The "annoying memory controller thing" did NOT need a fix.
The simple reason beeing it is design a FEATURE of the K8 architecture. Not a bug.

RE: Why the hate
By mino on 12/6/2006 2:06:40 AM , Rating: 2
Actually AMD has mature yields at 65nm so their process is mature from this point of view. AMD intentionally did not pmass-produce 65nm parts until they were certain of better overall output that their _very_efficient_ 90nm process because for their problems to satisfy demand.
FYI AMD is sold out for _every_ consecutive quarter starting by Q3 2005!!!

The Stars could not be rolled out yet NOT because of the manufacturing process but because of the CPU design itself. It is still buggy and at least one more stepping is expected to be necessary to iron out most of the remaining significant bugs.

RE: Why the hate
By JumpingJack on 12/6/2006 3:06:54 AM , Rating: 2
You have been reading too much AMD propoganda.

Could you tell me, quantifiably, what mature yield really means?

RE: Why the hate
By Viditor on 12/6/2006 3:52:48 AM , Rating: 2
Could you tell me, quantifiably, what mature yield really means?

It means that they began volume production on 65nm at a percentage yield equivalent to that of their 90nm process...which of course is much higher in unit yield.
And as they announced this to their shareholders at the CC, I would suggest that it isn't propaganda (because they would be risking some serious lawsuits if it were...).

RE: Why the hate
By mino on 12/6/2006 7:12:20 AM , Rating: 2
IMHO not completely true.
I am pretty sure their percentage yield on 65 is 10+ percent lower than their "current" 90nm yields. One thing we know is the overall output on 65nm is higher that on 90nm. Otherwise it would not make bussiness sense to migrate allready.
As 12x/184 ~ 0.65 therefore their 65nm yields are _no_worse_than_ than ~0.65 of their current percentage 90nm yields.
A qualified guess 85-90% on 90nm puts 65nm to the 56%+ percentage.
As AMD said last waffers came on the better end of expected spectrum I would guess for 60-65% real yields.

2 JumpinJack: Almost in no circumstance a chip company does publicize its yields, these data are considered too sensitive.

RE: Why the hate
By slayerized on 12/7/2006 3:54:22 PM , Rating: 2
In theoretical terms, according to yield prediction models yield is a function of die size. Since the die size for a 65nm process is definitely smaller than a 90nm one one should expect yields to be higher. This is of course true only from a modeling perspective and it is true when you compare two dies from the same process but with different sizes (that is one of the reasons why intel released a quad core from 2 dies rather than one native quad core). As mino said there is no way to really say whether the process has really matured or not because yield numbers are extremely confidential and it is totally proprietary information. The only way to say if someone is having trouble ramping up is when you look at the supply chains (PS3 is an excellent example; the blue ray diode thingy is having pathetic yield and hence the bottleneck). Initial yield issues in process shift will be negated to a certain extent to the wafer throughput (dies/wafer). In the end everyone lives! So peace out guys

RE: Why the hate
By JumpingJack on 12/10/2006 2:37:17 AM , Rating: 2
2 JumpinJack: Almost in no circumstance a chip company does publicize its yields, these data are considered too sensitive.

That was exactly my point :) ... mature yield as stated by either player is absolutely meaningless to us as we do not know what 'mature' means in absolute quantity.

"People Don't Respect Confidentiality in This Industry" -- Sony Computer Entertainment of America President and CEO Jack Tretton
Related Articles
AMD Beyond "Brisbane"
November 14, 2006, 6:31 PM
AMD 65nm Product Plans Unveiled
November 14, 2006, 3:17 PM

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki