i'm kind of disappointed...
don't get me wrong the review is great
but the really interesting thing (imho) would be
the power consumptions of the new phenom(s)
when installed in AM2+ motherboards !!!
http://www.silentpcreview.com/790gx-gig-vs-msi

i missed such a comparison in the first AM3 review
http://www.silentpcreview.com/asus_m4a78t-e
but thought/hoped it would follow in another context...


bottom

Great review, with all applicable aspects covered and considered! (Except for use with AM2+ mobo. (Added after reading Bottom's comment.))

One thought though. How do you define "game [mother] board"?
I just did a quick survey of the Swedish market, and the mobo used in your review is the cheapest AM3 board available (at ~1,500 SEK).
With requirements reduced to socket 775 and DDR3 support, there are a broad range of motherboards available at about half of that cost.
To make it even worse for AMD, the Q9550 is available at a price 15 SEK below the new Phenom II.

Cheers
Olle

I think those motherboard prices are too new in the Sweden to be representative.
The prices in Germany seems more balanced.

AM3: starts at €73

775 DDR3: starts at €57

S775 seems cheaper over there too but there's a smaller difference, even if you compare P45 to 790X for instance and don't look the low end S775 chipsets.

Expect the 955 to go down in price very soon, the model is just launched and will drop in price even if AMD doesn't lower it. The Q9550 have been around for ages, nothing will change the price unless Intel says so..

Why does Anandtech show significantly different results?

"The Phenom II X4 955 draws slightly more power than the Core 2 Quad Q9550, but not by a significant amount. Based on its performance, the 955 takes the energy efficiency crown away from the Q9550 in the majority of our tests."

Those power numbers are awful...From all the reviews I'd read it seemed the 9xx series Phenom II's were a lot closer to Intel in power draw. Anand shows this,


For intel they used an x48/x58 board, and for AMD 790gx/Fx, they don't seem to be specific which boards were used where.

Techreport


-Asus M4A79T Deluxe

hothardware

-Asus M4A79T Deluxe

TomsHardware

-Asus M4A79T Deluxe
-not as fair since it's a q9550s it's up against...

Hexus
They use stupid flash charts
-ASUS M4A79T
-load was wprime 1024

Guru3d gives thess numbers for idle/load,

Phenom II X4 940 89 189
Phenom II X4 945 126 180
Phenom II X4 955BE 129 182


Just thought I'd gather this information, since it's predominantly what we care about here.

edit:since techreport's peak power doesn't tell the whole story, I added another one from their review.

re -- game board pricing -- the offending line was removed, I think you're right, AM3 doesn't have any price advantage.

re the power measurements. Keep in mind that our range of tests are more narrowly CPU-focused than many of the others, which engage 3D video / gaming.

I trust SPCR for power tests more than other sites. But when most of the other sites show a lower comparable power, it makes me wonder. Something seems a bit off. Maybe you guys got a bad sample or something?

I'd also like to reiterate my call for an AM3 idle undervolt test. In the Asus M4A78T-E review, the X3 720 clocked it at 50W when running at 1.6Ghz and 0.9125V compared to 54W when running at 800Mhz and the default 1.025V. I'd like to know how low these things can do at their minimum 800Mhz and if it can get down to around 40W idle in a basic system.

Or maybe our QX9650 is particularly good?

That's another possibility

I trust articles here far more than any other place. Given that, the 'Net is often more a source of disinformation than it is information. I used to trust XbitLabs until the recent X3 X4 power consumption numbers. They are not comparable to their earlier Intel findings because only the recent AMD setups had an ATI 4870 installed, which is going to totally mess up the idle numbers. You can't tell if they are simply not thinking or adjusting the hardware to follow their conclusions, all I know for sure is such apples to oranges comparisions do more harm than good.

Full load tests should in fact fully load CPUs. Video card consumption should be handled and reported separately. Measuring power consumption with the CPU pegged at 100% while running Crysis is not a good indicator of CPU consumption, especially with an SLI or Cross Fire setup.

At least for the 65nm processors and on the 3 PCs I had folding I can state that for a given watt Intel did more work and scaled better when OCing. AMD scaled so badly I put those processors back to stock speeds and they now are idle and only support GPU folding. That kind of behavior may indicate that AMD may be more suitable for UVing than OCing. But then when Iganu was here he UVed a Q6600 that folded and was using 85 watts, a 65nm quad under load, so I can't say for sure.

Freshen up on SPCR Forum Rule 14.

hhmm...never have I seen anyone give me any grief about that...and I see plenty of images hotlinked in other threads and not a single mention about those either. I can take them down and post them on photobucket or something...is that what mods want?

The Tom's Hardware number for the X3 720 BE is a typo, right?

And it is quite remarkable that the 9650 beats the 720 on power consumption!