Can any 45nm AMD chip compete with Intel E7200 in energy efficiency?

[jrichrds]

I was a fan of the 45W TDP AMD X2 Brisbane chips for quite some time...not because they're any different than the regular X2 Brisbanes, but because they typically had higher undervolting/overclocking potential.

That was, until I tried the 45nm Intel E7200. Despite being a TDP 65W processor, it used less energy than even my lowest end "Energy efficient" X2 Brisbane while being significantly faster. So I've built multiple systems around the E7200 since then.

Now, I find myself in the situation that I much prefer AMD-based motherboards than the Intel-based offerings. Can any of AMD's new 45nm offerings match the E7200 in performance per watt?

 

[Denithor]

Here's what you do: consider the recent Athlon II X2 & Phenom II X2 article on the main AT page. Divide the x264 first pass encode FPS for each chip by the load power consumption from the same benchmark. Here's what you get.

PII 550 = 39.5/147.1 = 0.27 FPS/W
AII 250 = 38.1/154.3 = 0.25 FPS/W
e6300 = 34.1/162 = 0.21 FPS/W
e5300 = 33/148 = 0.22 FPS/W

Any other questions?

In all seriousness, I wish AT had included more power consumption numbers in the charts. They had the e7500 from Intel and the X3 710 and X4 940 from AMD in the x264 benching. Would have been interesting to see how much more efficient the triple & quad core chips perform.

 

[hans007]

I had an E7200 when it came out, and that was an amazing chip power wise @ load and idle.

I don't think intel has really topped that since the new e5200 / e6300s and such are the same chip. THe x2 250 might be close though since its a faster chip at load, but it seems to use more power at idle and load, but overall you might be able to beat the e7200 since the e7200 .

When I had an e7200 I used the g31 with it though. It would be tough to beat an e7200 with say a G41 or geforce 9400 chipset as their total system power would be even lower. But I think AMD is competitive at this point. the E series brisbanes really were not too competitive, though @ least you could get to competitive idle power if you use it with a 740G (55nm and the smallest chipset AMD makes). I think a II X2 250 paired with a 740G would be competitive, though you'd probably want a 780G just because its a better chipset (well it is if you use IGP and want hypertransport 3)

 

[deputc26]

Correct me if I'm wrong (Idc should know) but I believe that because GF's 45nm process is SOI and Intel's is HKMG AMDwill have lower power at low frequencies but Intel will be better at high freq's. I believe the line where Intel's HKMG becomes more efficient than GF's SOI is around (very loosely) 3 Ghz. However, unless you have both chips fully undervolted it will be hard to know. I hate to say it but I believe Core 2 is still a slightly more efficient uArchitecture than PhII. That said I'd still go with the 250 and then max undervolt it and possibly underclock to E7200 level of performance.

 

[Accord99]

Silentpcreview says no:

http://www.silentpcreview.com/article949-page4.html

 

[deputc26]

Thanks Accord, hadn't seen that. Looks like Intel wins, I still have hope for AMD at lower clocks though.

Edit: Actually on further reviewing those numbers I am suspicious about the 550 beating the 250 by 40% and 28% on the first two tests (page 4) that does not seem consistent with previous reviews I've read or with the mere 100mhz clock disparity and (not so mere) 4mb cache disparity (though perhaps there is some penalty for the lack of cache coherency due to the missing L3?)

Also I am suspicious of mobo pwr consumption throwing things off as PhII typically has lower idle power but higher load power in comparison to core 2. I do think you're right (core 2 wins) but I'm really trying to give AMD something, not a huge fan just don't want to see them belly up.

 

[geokilla]

If I remember correctly, the TDP ratings for Intel and AMD CPUs are different. Even though both CPUs may have a TDP rating of 65W, one will actually consume more power than the other due to how the companies rate their TDP.

 

[imported_Scoop]

Originally posted by: geokilla
If I remember correctly, the TDP ratings for Intel and AMD CPUs are different. Even though both CPUs may have a TDP rating of 65W, one will actually consume more power than the other due to how the companies rate their TDP.

From the aforementioned SPCR article:

"By AMD standards, the E7200 is a 45W chip."

 

[hans007]

Originally posted by: geokilla
If I remember correctly, the TDP ratings for Intel and AMD CPUs are different. Even though both CPUs may have a TDP rating of 65W, one will actually consume more power than the other due to how the companies rate their TDP.

AMD has 2 standards for the opteron chips, but i think for normal desktop ones the TDP is the same.


TDP is the maximum output value, intel and AMD both put chips into basically heat output classes so consumers will know what type of heatsink will definitely work.

It doesnt mean the CPU actually puts out that much heat at load (I find reviewers who fail to get this idiots by the way, i.e. some will say something like, the e7200 and e7500 both put out 65watts because the TDP is 65W). Its just means that it falls under that number "somewhere".

I think the opteron standard for AMD is acp and TDP, and the acp number is lower. TDP has to do with every single functional unit on the cpu being loaded at once and it doesn't happen except in very rare cases so AMD made up some other number that has to do with the average load or something like that.

 

[Viditor]

Originally posted by: hans007

Originally posted by: geokilla
If I remember correctly, the TDP ratings for Intel and AMD CPUs are different. Even though both CPUs may have a TDP rating of 65W, one will actually consume more power than the other due to how the companies rate their TDP.

AMD has 2 standards for the opteron chips, but i think for normal desktop ones the TDP is the same.


TDP is the maximum output value, intel and AMD both put chips into basically heat output classes so consumers will know what type of heatsink will definitely work.

No, TDP is a guideline for OEM's design. For Intel it's calculated as the highest load expected, and for AMD it's the highest theoretical load. Subtle differences that probably won't effect most people, but then TDP is a poor way of comparing chips anyway.

It doesnt mean the CPU actually puts out that much heat at load (I find reviewers who fail to get this idiots by the way, i.e. some will say something like, the e7200 and e7500 both put out 65watts because the TDP is 65W). Its just means that it falls under that number "somewhere".

Very well said...:thumbsup:

I think the opteron standard for AMD is acp and TDP, and the acp number is lower. TDP has to do with every single functional unit on the cpu being loaded at once and it doesn't happen except in very rare cases so AMD made up some other number that has to do with the average load or something like that.

see above...ACP is AMD's supposed attempt to replicate Intel's version of TDP, but it misses the mark by quite a bit.