Advanced Micro Devices - Moderated (AMD) Message Board

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)

AMD 230.82

+1.3%

1:55 PM EST

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next

To: Joe NYC who wrote (169653)	8/23/2005 1:28:32 PM
From: pgerassi	Read Replies (1) of 275872

Dear Joe: Opteron already has more than three times the performance per watt of P4 Xeons. Don't go by the TDP numbers because Intel uses some liberal typical standard and AMD uses a absolute worst case family standard. By end of 2006, it will be on 65nm and may get 5 times the performance per watt. So Merom compared to the competition at that time it is shipping looks to be hot, slow and expensive. Ditto for Conroe when it ships. It is always the same with Intel. Lok at some year or two down the road product goals against a real world shipping CPU of today. Use our pie in the sky numbers (which we rarely meet) versus our competitor's current real world numbers. A few questions need to be answered. How do they measure performance? Doing what? How do they measure power? Are the standards like AMD's? Doing integer logic operations at battery power clock speeds (lowest speed they go) on software (programs and data) that fits entirely in the L1 caches will have a far larger performance per watt numbers than a CPU running heavy floating point code at top speed on huge datasets. The former can use the lowest voltages and the smallest currents, thus have low power use and have ok performance but, have high performance numbers than the exact same CPU, running a application that uses 8GB of memory for its working set, doing lots of DP FP operations at top speed. The many stalls due to memory latency, branch heavy code and non optimal binaries and the high supply voltage, cause the power used to be high and the performance to be low making the performance per watt be very low. I take Intel's future numbers with a huge pile of salt. There is just too much wiggle room and history to have any confidence that any of the goals will be met. Pete

Report TOU Violation

Share This Post

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next