Replies to post #36012 on Intel Corp (INTC)

[wbmw]

Klaus, whoever made that slide should be fired.

It starts by saying that TDP is NOT a measure of real energy consumption, and then AMD goes on to rate these processors as far as Processor X TDP at idle.

Does that not seem inconsistent to you?

Well, foolish slideware aside, other reviews have confirmed that Core 2 processors dissipate more in idle. Since we already know that Core 2 has more transistors (more than double the number of power hungry logic transistors), so this is not all that surprising. The other aspects to consider are HALT state voltage, inclusion of various sleep states, and process differences.

If AMD were as forthcoming as Intel, we could look up at least the first two in the product documentation. Unfortunately, AMD does not publish these things without NDA. As for process differences, I would hardly expect AMD's 65nm process to be lower leakage than Intel's, let alone AMD's 90nm, which is what is being compared here.

My guess is that with >2x more logic transistors (which gave Intel the edge in performance), and no sleep states enabled on the desktop line, Intel simply ended up with higher idle power. So now all of a sudden AMD marketing tries convincing the world that they need to base their PC purchasing decision on a metric that no one has ever needed to use in the past.

In fact, while AMD can claim 73.3% less idle power than Core 2 in their 35W TDP line (!!!), it still amounts to only 10W.

Now you tell me: assuming the absolute WORST CASE scenario where a PC is on 24 hours a day, 7 days a week, running no significant work at all during this time, how much do you lose by going with a 10W higher TDP part?

Well, assuming the going rate of $0.06 per KWh, and 8,760 hours in a year, then at 10W (0.01 KW) you lose approximately $5.25 per year with the Core 2 processor (WORST CASE).

Now good luck convincing anyone that this is worth coming in 30% behind when performance/watt in real workloads is actually important.