Page 1 of 2

AMD Phenom II X2 550 BE & Athlon II X2 250

Posted: Fri Jun 05, 2009 2:11 am
by MikeC

Posted: Fri Jun 05, 2009 3:06 am
by Mats
Great review! I still think SPCR's X3 720 uses way too much power in idle, compared to what other sites have shown.
At load, the triple core X3 720 used 24W more than the X2 550. Taking into consideration the power supply's 80% (more or less) efficiency, that's a difference of about 19W DC. The X3 720 is a 95W part, so if you take that as a baseline, 80W for the X2 550 is not too far off. The X2 250 however, used about 25W DC less, which would put it closer to 70W than 65W.
I'm sorry but I have to say that this reasoning isn't very good for the readers. I can't really say it's SPCR's job to educate people, but newbies may interpret this as TDP being more of an absolute value, rather than a limit. I've tried to explain so many times how TDP works in the forums, and that 65 W should be ≤65 W. Still, people post questions, maybe every week, based on the belief that TDP is an absolute value.
By AMD standards, the E7200 is a 45W chip.
I'd say it's by any standard. I'm still not sure why Intel have done this, but it could be a way to avoid motherboards being designed for a CPU with that kind of low TDP only, or to make the mobile products appear more differentiated form the desktop counterparts.

Posted: Fri Jun 05, 2009 6:36 am
by bbzidane
do you know how the 4850e compares with the athlon II x2 250

it would nice if i was able to swap in a faster cpu without increasing power consumption by much

Posted: Fri Jun 05, 2009 6:40 am
by Mariner
Hmm. I'm slightly disappointed with the apparent undervolting ability of the Athlon II X2, especially when compared to the Brisbane Athlon X2s.

I suppose I may have been hoping for too much - after all, the core speed of this Athlon II is up to 3GHz - but it would seem that Intel's 45nm process (or possibly just their architecture?) is perhaps a little more advanced than that of AMD and may allow greater levels of undervolting.

It would, however, be interesting to see some undervolting/power draw figures with the Athlon II downclocked to, say, 2.5GHz.

Posted: Fri Jun 05, 2009 7:04 am
by Mats
Mariner wrote:I suppose I may have been hoping for too much - after all, the core speed of this Athlon II is up to 3GHz - but it would seem that Intel's 45nm process (or possibly just their architecture?) is perhaps a little more advanced than that of AMD and may allow greater levels of undervolting.
I don't expect much from neither undervolting nor overclocking. AMD have the slower CPU and have to push them to be able to compete with Intel.
I remember when my Opteron 146 overclocked 50% from 2 to 3 GHz, it was 2005 and AMD was the best. Now people think 35% is great, and that's usually what you can expect from many Phenom II CPU's. The reason is of course that they're clocked very high from the beginning. AMD's 45 nm process is still on it's first public revision (C2), so things can only get better. The six core Opterons have a D0 stepping.

Posted: Fri Jun 05, 2009 10:39 am
by colin2
I really appreciate your work on this. And yikes, that's a lot of power consumption. I'm pinning my hopes on competition around notebook chips.

Posted: Fri Jun 05, 2009 10:43 am
by ryboto
Lostcircuits tested the power consumption here, http://www.lostcircuits.com/mambo//inde ... mitstart=7, 30W before the VRM for the 250. Also, Anand tested the power consumption of the chips here, http://anandtech.com/cpuchipsets/showdo ... =3572&p=10, and show the two dual core AMDs using around a Pentium E6300 in full load, though they only tested load during an encoding run.

edit: lostcircuits also had the NB/IMC overclocked to 2.4ghz.

Posted: Fri Jun 05, 2009 12:22 pm
by Rewdoalb
some of these power figures have me confused as well. The article here at SPCR used the E7200 as their comparison Intel CPU - this CPU is no longer on the market and has been subceded by processors with higher clock speed (and higher power draw?) such as the E7400, E7500, and E7600.

Meanwhile, the article at Anandtech which ryboto refers to shows both AMD chips having overall a better efficiency compared to the new E6300, and the two AMD chips come in at a power draw almost as good as the E5200.

Basically, I'm wondering if this assumption is wrong:
If our E7200 sample is indicative of the average power consumption one can expect from all of Intel's 45nm dual core chips, any of them could easily wipe the floor with the X2 550 in an energy efficiency contest. At idle, they use about the same amount of power, but at load the difference is about 40W.
At load the X2 550 beats the E5200 by 0.9W and the E6300 by 14.9W. This is not "wiping the floor" by Intel...

If ya'll could look into this, I would appreciate it. Otherwise, I really appreciate the review, and keep up the good work it is very helpful!

Posted: Fri Jun 05, 2009 1:50 pm
by jessekopelman
It is worth noting that the new Pentium E6300 supports VT (just like the old E6300). This makes it all the more compelling for an $84 part. Apparently all > Celeron parts will have VT going forward, so this advantage for AMD is soon to be gone.

It is also worth noting that current BIOS don't support the Athlon II's Cn'Q implementation properly and this will impact performance for video playback. I'm sure it will get fixed eventually, but it is something to consider if you are looking to buy immediately.

Posted: Fri Jun 05, 2009 7:10 pm
by yukon
I also find the power usage numbers a bit confusing, but I use them when I research buying new equipment, and I'm not researching lately.

"Intel's advantage is simple: their chips perform well for the price and are more energy efficient."

This is disappointing. I was a big AMD fan back in the later Pentium days. AMD's products performed similarly, and sometimes better, for a far more efficient processor that cost perhaps a quarter of Intel's parts. My last AMD CPUs were an Athlon64 and a dualcore Opteron, but since then it's been all Intel. I've been hoping for AMD to become more competitive.

Posted: Sat Jun 06, 2009 8:55 am
by smilingcrow
Rewdoalb wrote:Meanwhile, the article at Anandtech which ryboto refers to shows both AMD chips having overall a better efficiency compared to the new E6300, and the two AMD chips come in at a power draw almost as good as the E5200.
The Anandtech load power consumption figure for Phenom II X2 is lower than for the Athlon II X2 which contradicts every other review and doesn’t make sense considering the extra circuitry it has. Therefore I dismiss those figures out of hand.
jessekopelman wrote:It is worth noting that the new Pentium E6300 supports VT (just like the old E6300). This makes it all the more compelling for an $84 part.
Yes, it does seem the best of the budget Intel range and it has a higher FSB which is advantage in certain cases – E6300.
jessekopelman wrote:It is also worth noting that current BIOS don't support the Athlon II's Cn'Q implementation properly and this will impact performance for video playback. I'm sure it will get fixed eventually, but it is something to consider if you are looking to buy immediately.
The Anandtech review speculated on whether BIOS issues affected the power consumption of the Athlon II X2.
It’s a shame to see niggling BIOS issues with so basic a feature as CnQ but overall it’s good to see AMD raising their game so much these last 3 months or so.

Posted: Sat Jun 06, 2009 9:41 am
by MikeC
There's little doubt the lostcircuits measurements of DC power draw at the 2x12V Aux12v socket are quite precise. One issue with the lostcircuits comparison tables is that the C2D processors represented are first gen. There are only the early 6000 series chips, which had higher TDP. No 7000, 8000 or 5000 series.

It's also possible that the Intel board we used has lower power demand than the AMD boards. Since we're measuring total system power at the AC socket and every other component -- RAM, PSU, HDD -- remains identical (the same ones, not just same models, are used), the one difference other than CPU is the motherboard.

Right now, our Aux12V power measurement tool is not operational, but when it's repaired, we'll try some DC measurements to see if they correlate with our total system power measurements.

Edit: It's also been suggested that the onboard video on the AMD 790GX boards may continue to draw power even when it's been disabled. To be verified.

Posted: Sat Jun 06, 2009 6:31 pm
by frisky
Is the idle P-state undervolted as well?

Typically P2 undervolt to below 0.8V at idle, if that is not done, their figures at idle would be higher than they should be.

Posted: Sat Jun 06, 2009 8:03 pm
by MikeC
A counterpoint to the lostcircuits findings is at http://www.xbitlabs.com/articles/cpu/di ... html#sect0 -- they obtained similar results as we did, and they were using an FX (no onboard video) board for AMD.

Posted: Sun Jun 07, 2009 1:03 am
by dvdmonster
Is it me, or does this article seem a bit.. ehh, rushed?
Why test against a discontinued Intel processor?
Why not run tests with integrated gfx on/off to see if it's acually off.

I would like to see how the Athlon II X2 250 preforms against the 5050e/4850e/BE2400. This is pretty easy drop in replacement, and very interesting for all those with low power 780G systems.

Let's see some updated number please 8)

Posted: Sun Jun 07, 2009 8:03 am
by winguy
I would love to SPCR review the new Intel E6300. Hey it has Intel VT!

Posted: Sun Jun 07, 2009 6:15 pm
by colin2
Could you cool the Athlon II X2 250 passively? That is, a Ninja or equivalent plus a case fan?

Posted: Sun Jun 07, 2009 6:28 pm
by Vibrator
colin2 wrote:Could you cool the Athlon II X2 250 passively? That is, a Ninja or equivalent plus a case fan?
Yes

It's a 65W TDP processor

The Ninja 2, HR-01 Plus, Orochi, and the likes can cool it passively no problem

Posted: Tue Jun 09, 2009 4:49 pm
by vext
Interesting article and I always take SPCR articles seriously. However I wonder about this statement:

"As expected, the average power consumption for both the new AMD chips during our timed tests was much higher than the E7200..."

Contrast that with Lost Circuit's review which found the Athlon II to have the lowest idle power draw of any desktop cpu they have tested. Lost Circuits spent some time and isolated cpu power draw from system power draw whereas SPCR measured system draw. But if the Athlon II is truly as efficient as Lost Circuits article indicates then why did SPCR get such high system power draw? ATI chipsets are not power hogs, partly because the memory controller is on the cpu. Odd.

Unfortunately the SPCR article implies the new A II and P II dual core cpus need hefty cooling and may be difficult to cool passively, therefore not suited for quiet systems.

What we need is some comparative data on cpu temperatures instead of relying totally on system power draw. I have a Phenom II 955, equipped with an older Thermalright XP-90 aluminum cooler, and a Noctua 90mm fan. The cooling was inherited from my previous A64 Manchester 2.2Ghz single core setup. The 955 idles at the same temperatures as the Manchester did. That confirms Lost Circuit's chart which shows 955 and Manchester idle power draw is identical.

If Lost Circuits is right, then the A II and P II should be very exciting cpus for experimenting with passive cooling.

Posted: Tue Jun 09, 2009 5:23 pm
by MikeC
vext -- we're not the only ones w/ these findings. Go back up a few posts to the xbit labs link.

One thing worthy of note: We're always trying to push the CPUs to their max load, using whatever utilities work best. For AMD processors, CPUBurn has produced the highest power draw numbers. This used to be true of Intel P4s -- but changed w/ C2D -- Prime95 produced significantly higher power draw w/ C2D (and later Intel processors) than COUBrun or any other utility. So we switched to Prime95 for Intel and stayed with CPUBurn for AMD.

Lost Circuits uses the same stress loads for all the CPUs and say they get the same results w/ all -- "Caligari 5.1 rendering with all cores, 4 instances of Prime95 etc are all within ~0.5 W difference from each other."

It's possible that CPUBurn somehow pushes AMD processors harder than Prime95 does Intel chips.

Posted: Wed Jun 10, 2009 2:02 am
by Mats
MikeC: What about Intel burn test, I thought it was optimized for newer Intel CPU's?

Posted: Wed Jun 10, 2009 4:20 am
by loimlo
http://www.techreport.com/articles.x/17023/12
http://www.techreport.com/articles.x/17023/2

Techreport has compared several CPUs' power consumption, including comparisons between DDR2 and DDR3 with Phenom II X4 810 on different motherboards. Techreport's power draw is a far cry from SPCR's result. I guess ASUS DDR2 motherboard needs to take responsibility that draws more power in SPCR review?

Posted: Wed Jun 10, 2009 4:42 am
by Ch0z3n
MikeC, I think you start doing tons of consulting work; help some of these companies figure what the hell people (read: us) want. IIRC you said you helped Antec with the P180 & 182; they and the solo are pretty much the goto cases on this site. Maybe help Antec make some good mATX and ITX cases, get AMD to make some ultra-low power (TDP < 30w) processors that are still serviceable for HTPC and productivity use.

It seems like more often than not products come out that *could* be awesome but end up having some fatal flaws. The only flaws I can see with the P182 is the power button under the door and it is large enough to sleep in. 8)

Posted: Wed Jun 10, 2009 5:16 am
by Mats
A combo board is the only useable motherboard for a DDR2 vs DDR3 comparison. MSI and Jetway makes such boards.

Posted: Wed Jun 10, 2009 6:13 am
by MikeC
loimlo wrote:http://www.techreport.com/articles.x/17023/12
http://www.techreport.com/articles.x/17023/2

Techreport has compared several CPUs' power consumption, including comparisons between DDR2 and DDR3 with Phenom II X4 810 on different motherboards. Techreport's power draw is a far cry from SPCR's result. I guess ASUS DDR2 motherboard needs to take responsibility that draws more power in SPCR review?
Unless I skimmed it too fast, there's so little info in techreport's article on how they tested the AMD processors that it's difficult to understand how you came to your conclusions.

Posted: Wed Jun 10, 2009 7:11 am
by loimlo
MikeC wrote: Unless I skimmed it too fast, there's so little info in techreport's article on how they tested the AMD processors that it's difficult to understand how you came to your conclusions.
The combination of Techreport components:
Phenom II X4 810, MSI DKA790GX Platinum, Corsair WIN4X4096-8500C5DF DDR2 SDRAM 2GBx2, ATi 4870
Phenom II X4 810, Asus M4A79T Deluxe, Corsair TW3X4G1600C9DHXNV DDR3 SDRAM 2GBx2, ATi 4870

Techreport tested the same processor with different rams on different motherboards, but the power figures were very close. The power difference between AM2+ and AM3 board was 3W at idle and during peak Cinebench rendering load. 142W and 202W on AM2+ side vs 145W and 199W on AM3 side. That's how I came to my supposition.

Edit: ATi 4870 added.

Posted: Wed Jun 10, 2009 7:23 am
by MikeC
loimbo -- I didn't see that info anywhere, what page? Anyway, fair enough. We'll try another AM2+ board with the same processors in the near future.

Posted: Wed Jun 10, 2009 7:28 am
by loimlo
The info was on the page 2 or second link that I gave before. But you need to spend some time figuring it out as it's too complicated to take a glance.

Posted: Wed Jun 10, 2009 8:45 am
by Tobias
The information is really hard to spot, unless one knows what to look for. Well spotted, loimlo:)

On page 12, the Phenom II X4 810 is listed with both DDR3 and DDR2 if one looks careful in the diagram (both are dark green in colour).

On page 2 there is confirmation in the table listing the hardware used. The two rightmost columns in the table list the 45nm AMD processors and Phenom II X4 810 is listed in both columns.

Posted: Wed Jun 10, 2009 9:40 am
by loimlo
Thank you, Tobias. I just happened to notice the Techreport's article tonight. In fact, Techreport elaborated the comparison on page 2 in this article. Anyway, hope to help clarifying power consumption mystery.