SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?


To: Hans de Vries who wrote (196760)5/15/2006 2:04:52 AM
From: eracerRead Replies (1) | Respond to of 275872
 
Re: The 2.4 GHz for 65nm parts (presumably no K8L yet) is most
likely conservative (The key to flawless execution is to not
promise extrapolations but to promise what you know you can
do today). Even AMD does not know now what it can do in
September or so when it starts mass production.


AMD rarely over-delivers in regards to launch frequencies so I don't know why 2.4GHz would be considered conservative, more like realistic. Even if AMD delivers at 2.6GHz it won't offer anything 90-nm doesn't except for lower power consumption.

...So these parts may become competitive in the notebook sector in early 2007.

I guess that depends on the definition of competitive. It doesn't look like the 65-nm mobiles will even be launched until late Q1/early Q2. That doesn't give 65-nm AMD mobiles much exposure during the first half of 2007. I doubt 65-nm launch frequencies will be above 2.4GHz. That should be competitive with a 2.13GHz Merom in terms of performance. Turion X2 may have better power consumption at 65-nm.

There is a similar advantage later on for Quad cores.
Kentsfield will be both power and bandwidth restricted, Twice
the 3Ghz Woodcrest TPD (80W) plus four times FBDIMM (+5W)
gives an unlikely 180W while 800 MHz DDR2 provides 50% more
bandwidth as Kentfields 1066 MHz FSB.


No one would expect a quad-core launch at 3GHz from AMD or Intel early next year. The Inquirer reported that Intel was expected to produce 2.33GHz/1333MHz FSB LV Woodcrest CPUs with 40W TDP. My guess is Intel would use these chips at 1066MHz FSB to get an 80-100W quad-core Kentsfield to 2.4GHz for launch. The 1066MHz FSB will be a performance bottleneck in some sitations. In other situations it might look attractive having 8 cores on a cheaper 2S board than using a more expensive 4S board with dual-core CPUs.



To: Hans de Vries who wrote (196760)5/15/2006 5:40:09 AM
From: RinkRead Replies (1) | Respond to of 275872
 
Hans, Could you please consider a future article on NGA vs. K8L once sufficient K8L info is available? Your articles on on Yamhill vs. K8 a couple of years back were rather valuable to a lot of us here, including me.

Regards,

Rink



To: Hans de Vries who wrote (196760)5/15/2006 10:36:47 AM
From: dougSF30Respond to of 275872
 
Could be. I guess we'll see. But even if that's 3 speedgrades worth of conservatism, you have a ~20% performance shortfall in DC parts.

As far as Kentsfield, I'd think that would run a couple speedgrades below Woodcrest, much like a QC Opteron would run slower than DC Opterons-- for power considerations alone. But there's no doubt that QC (as well as 4+ socket systems) will be Intel's weakness.