And I'm waiting for the Bulldozer...No matter if it doesn't exist.

wish the regular i5 had been in the list. to compare directly.

guess i go open up another window to compare...

Good article!

I would be interested in seeing a follow-up article that compares the power consumption of 5+ different production samples of the same processor model (ex. 5 different Core i5-2500) to see if they all have the same power consumption...

The reality is that not all processors have the same effective power consomption due to small manufacturing and internal leaking variations, and thus should yield different results.

My expectations would be that the T series have less variation since these are probably hand picked from within the regular bin and relabeled T-series (for a premium $$)

The regular bins have a bigger MAX TDP value, but it doesn't mean that ALL the units will perform at the MAX TDP value (as seen in this article). It's actually very unlikely that they will...

In short, your can be lucky and end up getting get some good samples in the production bins that were not good enough to be classified as T-series, but they can be close to the threshold and delifer impressive power consumption results.

On the other hand, you can also get a 'bad' production unit in the regular production series that has more internal leaking and perform closer to its MAX TDP...

That's a process call 'binning' in the chip industry.

still too high

it means that Core i5-2500T consume 31A?

Is there a reason you didn't put anything from AMD for reference on those graphs? I'm especially curious of power consumption. I would really like to see how the 620e stacks up against Intel's low power offerings. For sure, performance per watt would favor Intel, but I like to see just by how much they beat AMD to decide if it's worth the extra dollars.

I'm just trying to get a point of reference. The measurements in this review are all of core processor power consumption, and the last in-depth review of AMD processors was displayed as total system power consumption, so it's difficult to draw a direct comparison.

http://www.xbitlabs.com/a...us-m4a88t-i-deluxe_7.html

This was a good review and I'm big fan of the site.

The one thing that was missing, considering it's about energy efficiency, was energy measurements to achieve particular tasks.

Using the numbers in the article and (big if) assuming x264 encoding is an equivalent load to CPU burn, then the most efficient processors in the test by energy consumed per x254 frame encoded are:

1. Core i5-2500T (100%)
2. Core i5-2390T (104%)
3. Core i5-2310 (110%)
4. Core i3-2100T (111%)
5. Core i3-2120 (115%)
6. Core i3-2100 (116%)
7. Pentium G850 (131%)
8. Pentium G620 (140%)
9. Pentium G620T (158%)

These numbers are really rough which is why it'd be better to have the direct measurements.

Anyway, thanks for the great site. Keep up the good work.

The discussion of power savings in this article has a critical flaw in its logic. The article fails to consider the amount of time the CPU spends at elevated power levels in order to complete its processing task. When considering this a lot of the conclusions about which CPU consumes less power turn completely upside down.

Example: A large RAR compression. Once started, all systems must process the entire tar. Most tasks, apart from gaming, are like this. (Once processing task is given, system works on it at elevated power usage until task is done.)

Pentium G620T needs 2 hours at 20W, for 40 watt-hours total.
Pentium G620 needs 1.85 hours at 26W, for 48 watt-hours total.
Core i5-2310 needs 0.99 hours at 21.3W, for 21 watt-hours total.

Obviously, if your system is going to tar some files, total power usage spent on that task by the CPU -more than doubles- if you chose one of the lowest power, lowest performing chips.