Slashdot Mirror
AMD's Phenom II 965, 3.4GHz, 140 Watts, $245
Vigile writes "While AMD does not have the muscle to push around the i7, they certainly have the ability to give the older and more common Core 2 Quads a run for their money. With the release of the Phenom II X4 965, AMD further attempts to dethrone the Core 2 Quad as the premier midrange CPU offering. While it may not be a world-beater by any stretch of the imagination, it certainly is catching Intel's attention in the breadbasket of the CPU market. The X4 965 is the fastest clocked processor that AMD has ever produced, much less shipped in mass quantities. While the speed bump is appreciated, the cost in terms of power and heat will make the introduction of the X4 965 problematic for some. Many of us thought that we would never see another 140 watt processor (as the Phenom 9950 was), but unfortunately those days are back. Still, AMD offers a compelling part at a reasonable price, and their motherboard support for this new 140 watt processor is robust."
Beckton, the 8 core / 16 threads Nehalem CPU will be out in Q1 2010.
http://www.theinquirer.net/inquirer/opinion/1050976/intel-bunch-fun-cpus-moves-2010
Intel and AMD release different numbers for their CPU's power consumption. Intel gives an average and AMD gives a maximum. They're not comparible. In real world testing, the X4 965 uses slightly less power at idle and slightly more power at full load than a stock Q9550.
You can get an i7 920 for $200, so not only is it faster and lower power, it's also cheeper.
http://www.microcenter.com/single_product_results.phtml?product_id=0302727
If you're referring to Moore's law, it's not a doubling of speed every 18 months, it's a doubling of transistor counts. Clock speed has never been part of that equation, no matter what intel's late-1990s marketing department would have you believe.
I'd mod you up.
But this is the customer line they are looking for. We currently use 6 computers with Phenom II 940's onboard at this time for data analysis. When comparing these with Intel computers the cost would have been much higher. What we found repeatedly during matrix calculations was that the performance vs cost just wasn't a contest.
I am really really beginning to wonder how much calculation stress people put on their computer in this mid-range market, because I hear and see all of these benchmarks and people discussing their performance and I just don't see a cost performance comparison. I have talked with other people who do some of the same type of work and they see the same thing.
We do take some added cautions with air circulation just to make sure that we don't run into issues. We repeatedly have cranked these things for 24 hours straight with cpu's running at 70 to 95% of capacity and actually had one occasion where the systems ran over a weekend (limits weren't set correctly).
Yeah. Intel's TDP definition is different from AMD's. AMD reports the maximum wattage the CPU can burn at full load. Intel reports a "typical" number.
If this was an Intel CPU the number would be something like 90W or 95W for the very same CPU.
Anyway, everybody knows that. You can't trust vendor's numbers. Just do your own checking with a power meter.
Tech Report tells a different story: ... A story that says that you can't necessarily compare like-for-like when using different motherboards.
Which is why X-bit Labs did this:
To get a better idea of the situation, we performed a separate Phenom II X4 965 power consumption test under heavy load when none of the other system components are taken into account. To be more exact, we measured the consumption along the 12 V power line connected directly to the processor voltage regulator on the mainboard. In other words, this measurement method didnt take into account the efficiency of the voltage regulator circuitry. ... and got 84W for the Q9550 and 147.6W for the X4 965. Granted, like they said in the last sentence, this doesn't take VRM efficiency into account, and it might as well be that those on the AMD motherboard were woefully inefficient and the CPU itself uses plenty less watts than measured. Still, this is 75% extra, and the VRMs cannot be the main reason for it.
They do similar measurements for graphics cards, too, because measuring total system power consumption from the wall can only get you so far.
I'd personally be more inclined to trust the Ukrainians.
Generally speaking, CMOS power consumption is the result of charging and discharging gate capacitors. The charge required to fully charge the gate grows with the voltage; charge times frequency is current. Voltage times current is power. So, as you raise the voltage, the current consumption grows linearly, and the power consumption quadratically, at a fixed frequency. Once you reach the frequency limit of the chip without raising the voltage, further frequency increases are normally proportional to voltage. In other words, once you have to start raising the voltage, power consumption tends to rise with the cube of frequency.