Silicon Investor (SI) -- The First Internet Community

STOCKTALK

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: Petz who wrote (238945)	8/16/2007 3:52:18 PM
From: wbmw	Read Replies (1) \| Respond to of 275872

Re: It's not just the FBDIMM penalty, your northbridge more than tips the scales in AMD's favor. The northbridge also has the problem that it can't go to low power mode unless neither CPU is accessing memory. To accurately judge platform power, you need to take all the board and chipset components into account. Intel integrated a lot of functionality into their server north and south bridges, and AMD platforms need to pay the same power tax either way. AMD uses PCI-Express tunnels, which need to be accounted for, while Intel's PCI-Express mostly comes from the same northbridge that includes the memory controller. Intel also integrates networking and server management into their southbridge. How much additional power do these additional discrete devices dissipate in an AMD server? Agreed that the buck does not stop at the CPU, especially while idle, but arguably the CPU under load does dissipate a large percentage of the overall platform power. Re: And you conveniently forgot that AMD and Intel TDP specs are not created equal. This was probably true at some point, but if you read the language on TDP in any of the Intel and AMD thermal datasheets they are practically indistinguishable. I had a discussion with pgerassi about this, and I linked to all the datasheets and printed the exact verbiage. I won't bother looking up this post, but if you have trouble finding it, I can probably look it up again. Basically, TDP from both AMD and Intel represents max power dissipated under normal workloads. Intel's server parts have another max power term that defines the same thing, but at max VID. Intel's processors are perhaps even more conservative when it comes to TDP, since they measure at a higher Tj than AMD, which artificially raises leakage. AMD's idle specs have traditionally been lower than Intel's, but starting with the G-stepping Core parts, Intel has closed the gap to the point where they are fairly competitive, even on idle power. Of course, during idle, Intel servers pay the penalty of 4.0GT/s FB-DIMM links that don't shut off when the platform is idle. I've said before that Intel should design their platforms to address this, but perhaps the problem is bigger than I realize.