Not enough. :(

I thought that tri-cores should run at higher frequencies compared to quad-cores? Either this information, or information about >2.6GHz quad-core Phenom in Q4 is wrong...

Thought so too. Don't really see the point of this product at this frequency.

A higher freq 3C would be faster in many (desktop, limited threads) benches than a lower freq 4C, yet the 3C would have to be cheaper. For the same reason AMD (nor Intel) doesn't sell higher freq 1C than 2C CPUs.

I wouldn't have a problem (depending what I want to do with it) with paying more for a 3 GHz 3C than for a 2.6 GHz 4C, yet consumers might have big trouble understanding this. Or Marketing folk of AMD/Intel that is ;)

OTOH we might see higher clocked 3C vs. 4C at the same thermal spec some day, this way 3C would be a real win!

Now Digitimes is reporting that tri-cores will be launched in late March and there will be 45nm tri-cores in H1 2009:
http://www.digitimes.com/mobos/a20071023PD216.html

Quote:

Toliman adopts a 65nm process, supports socket AM2+ motherboards, and supports HyperTransport 3.0. The CPUs will have 2MB L3 cache and will be launched by the end of March 2008, according to the sources. The next-generation, 45nm-based CPUs will adopt socket AM3, support HyperTransport 3.0 and DDR2/DDR3 memory, and will launch in first half of 2009, noted the sources.

Taking account their usual 6 months delay on any schedule it should translate into a 2Q 2009 CPU release and actual products in costumer hands not early than 1Q 2010

jack wrote:

Now Digitimes is reporting that tri-cores will be launched in late March and there will be 45nm tri-cores in H1 2009: http://www.digitimes.com/mobos/a20071023PD216.html Quote: Toliman adopts a 65nm process, supports socket AM2+ motherboards, and supports HyperTransport 3.0. The CPUs will have 2MB L3 cache and will be launched by the end of March 2008, according to the sources. The next-generation, 45nm-based CPUs will adopt socket AM3, support HyperTransport 3.0 and DDR2/DDR3 memory, and will launch in first half of 2009, noted the sources.

Do you still believe that process technology does not make any difference?

who?

I think his comment more about the fact AMD is ALWAYS way behind with its fab upgrades. How many times did they shift and delay 65nm production?

oh, indeed sorry. Sure, I wouldn't bet on 45nm being on schedule either.

Considering all the delays and the imho fairly lacklustre performance of barcelona I wonder if it wouldn't have been easier and at least been a good back up plan to just throw an extra 1meg cache per core for the 65nm AthlonX2s and a respin or two to get the clock up.
The latest G2 steppings of the X2s almost look like that. 3.2ghz achievable(no idea on how good the quantities are though) and it seems like they have lowered the latency for the L2 cache back down(Never really believed that BS about the original latency increase for the 65nm X2s.)

Now Digitimes is reporting that tri core K10s will be available in February, however dual core K10s are pushed back to late Q2 2008.

http://www.digitimes.com/mobos/a20071129PD205.html

Perhaps AMD is waiting until they can make a 65nm K10 dual core that is faster than 90nm K8 dual core...

It seems that 2.3-2.5GHz@89W may have become 2.1-2.3GHz@95W: http://www.digitimes.com/mobos/a20071225PD219.html

According to the article, tri-cores released in March will be based on B2 stepping, if earlier rumors are correct this would mean that these CPUs would also have L3 cache disabled.

In Q2 there will be 2.1-2.3GHz tri-cores based on a new B3 stepping and also a 2.4GHz model...