Q4/08 AMD desktop roadmap

[Cookie Monster]

Link

Discuss.

 

[JavaMomma]

Just noticed in the new arrival section at NCIX ...
http://www.ncix.com/products/i...0DOBOX&manufacture=AMD
http://www.ncix.com/products/i...CGHBOX&manufacture=AMD

"AMD Athlon 64 X2 6500 Black Edition Dual Core Processor Socket AM2 2.3GHZ" is not on the road map...

 

[Cookie Monster]

That was released like this month. Q3 i believe it was, along with 125W 9950s and other tri cores.

That roadmap i linked to is only for Q4 of this year.

 

[SlowSpyder]

I thought Denab was to be January of 09? I guess either way Dec/Jan is pretty close. They have some nice low power products coming, but my guess is nothing that is going to get anyone excited here until Denab launches.

 

[Andrew1990]

That 22w Athlon X2 looks pretty good. Only 1.5GHz, but would be perfect for a HTPC.

 

[Cookie Monster]

Looks like 45nm wasnt enough to reduce the TDP of the new quads (due to the increase in clocks + x3 the L3 cache). 125W is quite a high number and it will probably outpaced by similarly clocked nehalems.

 

[Idontcare]

Originally posted by: Cookie Monster
Looks like 45nm wasnt enough to reduce the TDP of the new quads (due to the increase in clocks + x3 the L3 cache). 125W is quite a high number and it will probably outpaced by similarly clocked nehalems.

Since TDP does not determine the ASP of these desktop SKU's as much as clockspeed and benchmark performance we can't really be surprised they designed for higher clocks and more L3$ to chase the dollars. It's what we reward them for doing.

Remember though the 45nm process they are using is more of a dimension shrink and less of a materials change when taken in consideration to Intel's 65nm->45nm transition. Had Intel not changed to HK/MG then we'd have seen similiarly less spectacular TDP vs GHz gains.

The question that remains to be answered is whether AMD is seriously committed to their 45nm Version 2.0 which they publicly claimed would roll HK/MG and ultra-low-k dielectrics at 45nm. That has real opportunity to improve the GHz/TDP ratings (as it did for Intel).

But AMD does a fair amount of toothless biting (gumming, is it called?) these past years now. Not a bad thing if its the service you paid for

 

[GundamF91]

Should AMD be counted out for the foreseeable future? It looks like they're just trailing Intel with no hope of catching up. So they could only maintain marketshare by cut throat pricing, which means they wont' have as much cashflow to do more R&D, and then the circle keeps on going. It's just a bad thing. Maybe the cash from GPU could help kick start AMD's processor development.

 

[Idontcare]

Originally posted by: GundamF91
Should AMD be counted out for the foreseeable future? It looks like they're just trailing Intel with no hope of catching up. So they could only maintain marketshare by cut throat pricing, which means they wont' have as much cashflow to do more R&D, and then the circle keeps on going. It's just a bad thing. Maybe the cash from GPU could help kick start AMD's processor development.

This is the life-cycle of any business attempting to compete with a near-monopoly in their respective industry.

My props go out to AMD for keeping the rate of growth in the process technology and architecture gaps as low as they have been. But the truth now is the same as it was back at 90nm...AMD's chances of success in the CPU industry hinges critically on (1) near flawless AMD execution to a hopelessly over-aggressive architecture and node timeline, and (2) simultaneous disaster occurring at Intel in both architecture and process technology choices.

Think about the X2 vs. Prescott. Was AMD so bad-ass because of the X2 or because of Prescott? The answer is both, of course. So what is it going to take for AMD to do it again? It is going to take more than just releasing Bulldozer on 32nm in fall of 2009 (which is absurdly impossible, but necessary for a chance at success against Westmere)...it will also require Intel's 32nm to somehow enable a Prescott part 2 along with Westmere itself doing something which causes 180W TDP parts or some such.

So yeah, barring miracles like this it is pretty much written on the wall how this will end. Having 4x the market share and 4x the R&D revenue pretty much gives Intel absolutely no excuse for not dominating AMD. Even if AMD wins the anti-trust lawsuit all that will give them is maybe 4-5 yrs to do something with $11B in order to create a position in the industry for themselves which will allow them to sustainably compete against Intel.

I think the world will see consolidation of SUN/IBM/AMD versus Intel in the coming 5 yrs. There is no way the three will be able push thru to 22nm and 16nm without combining R&D resources with zero-degrees of business separation.