AMD Phenom II 920 & 940 Full Review

[mAJORD]

Anway for all tests including synthetics, but exluding memory related tests as they're useless.

Phenom II @ 2.8 vs Q9950 @ 2.83Ghz:

Q9950 is 11.9% faster with a 1% clock advantage

take away the heavily skewed towards Core FPU julia test (50% faster ck/ck, lol) and it's 10% ..

So all up, on these tests Yorkfield is around 9% faster clock/clock across the board. which makes it around the same IPC as Kentsfield.

Still two very different beasts when it comes to performance.. Intels raw decoding and prefetching ability, Vs Denebs high mem throughput and shared L3 makes them trade blows something chronic

As allways, a carefull eye on the types of benches chose to represent a desktop PC will need to be taken when the rest of the reviews hit. Results could be skewed towards either CPU easily by hand picking benchmarks.

[PetNorth]

The funniest part is that in these load of synthethic useless benches crap Q9550 is ~17.5% average faster than Q6600, when it has ~18% more clock. So basically, Kensfield = Yorkfield

[OverClocker_gr]

The funniest part is that in these load of synthethic useless benches crap Q9550 is ~17.5% average faster than Q6600, when it has ~18% more clock. So basically, Kensfield = Yorkfield

Kentsfiled used tighten strap because the 266mhz.

Yorkfield used a much lighter strap,so as you can see @ mem latency kentsfield is faster than yorkfield

[Miss Banana]

I don't think we have enough info to be claiming performance-per-watt as a strength. I'll believe it when I see it. Performance-per-watt over Yorkfield isn't looking too promising...

We have info regarding the server parts. Those performance per watt numbers were nothing short of impressive.

[mAJORD]

also, not all benches scale with frequency.. hence why we can't compare IPC of Q6600 with PhII

Other thing to remember is the mismatch of RAM. The 775 platform is running some pretty kickarse DDR3 there, so should be taken into account if looking at CPU alone. At least until AM3 models hit..

Saying that Kudos to the reviewer for using decent DDR2 for a change, it probably makes the mismatch almost negligable depending what setting the systems were using (?????) I can't count the number of times I've seen Phenom reviews with DDR2-800, and the 775 systems running top of the range DDR-9000000

[Macadamia]

Isn't this on a beta Destroyer BIOS by the way?


Foxxconn's full P-II support hasn't come yet.

[vietthanhpro]

Anway for all tests including synthetics, but exluding memory related tests as they're useless.

Phenom II @ 2.8 vs Q9950 @ 2.83Ghz:

Q9950 is 11.9% faster with a 1% clock advantage

take away the heavily skewed towards Core FPU julia test (50% faster ck/ck, lol) and it's 10% ..

So all up, on these tests Yorkfield is around 9% faster clock/clock across the board. which makes it around the same IPC as Kentsfield.

Still two very different beasts when it comes to performance.. Intels raw decoding and prefetching ability, Vs Denebs high mem throughput and shared L3 makes them trade blows something chronic

As allways, a carefull eye on the types of benches chose to represent a desktop PC will need to be taken when the rest of the reviews hit. Results could be skewed towards either CPU easily by hand picking benchmarks.

http://www.hwbox.gr/showthread.php?t=3189

This is fake !

Phenom II @ 2.8 vs Q9950 @ 2.83Ghz:
Q9950 is 11.9% faster with a 1% clock advantage

I don't think so! Deneb is faster than Agena in same clock !---> this is fake !

[OverClocker_gr]

Isn't this on a beta Destroyer BIOS by the way?


Foxxconn's full P-II support hasn't come yet.

i have a beta bios that supports fully PII

Asus at M3A79T doesn't have any bios for fully support

[OverClocker_gr]

This is fake !

I don't think so! Deneb is faster than Agena in same clock !---> this is fake !

in this review we have Q9550 is Intel Core 2 Quad Q9550.

So please watch what you are accusing me

[vietthanhpro]

in this review we have Q9550 is Intel Core 2 Quad Q9550.

So please watch what you are accusing me

Oh ! this is Q9550 not 9950 --> Phenom II is slower !

[informal]

Oh ! this is Q9550 not 9950 --> Phenom II is slower !

You are not,by any chance ,just maybe lo squartotore?

[Glow9]

Q9550 is a drop-in upgrade too.

Funny how the goal-post keeps shifting.

um yeah if u have a intel rig

[savantu]

What can I say , congrats to AMD for finally getting where Intel was in November 2006 !

Kentsifeld users can breathe a sigh of relief ; maybe in late 2009 AMD will finally have something able to win across the board vs. Kentsfield.

[Zucker2k]

Can you please run one of the above mentioned apps and post consumption in idel and load, so we can compare once PII is available for all of us?

Per your request:

[Morais]

Someone move this thread to AMD section...Admins there wont allow this OT and bull posts

[justapost]

Per your request:

Thanks Zucker2k, but what does the power meter say? Is your system running on air or water cooling?
I'll post my results as soon as I get an PII.

EDIT: I see Amps and Watta ge in Everest but it seems everest uses the VID (1.25V) here to calculate the wattage and not the 1.38V really used atm. Beside that 16A look good for the cores but it's quite useless to compare cpu's consumption allone with all the different power sources used nowadays.

[OverClocker_gr]

Per your request:

Why do you think that these values are true?

[Eson]

Per your request:

0.66W at idle... Then why don't NASA use overclocked Core 2 processors in their Mars robot?

[Zucker2k]

I gave a disclaimer a few pages back; what you see is what you get. I'm on water ~ 72F ambients. I'm too lazy to use a multimeter right now, so software would have to do.

Thanks Zucker2k, but what does the power meter say? Is your system running on air or water cooling?
I'll post my results as soon as I get an PII.

EDIT: I see Amps and Watta ge in Everest but it seems everest uses the VID (1.25V) here to calculate the wattage and not the 1.38V really used atm. Beside that 16A look good for the cores but it's quite useless to compare cpu's consumption allone with all the different power sources used nowadays.

I thought this was a cpu to cpu comparison? I don't have the resources to do a platform comparison. See below.

System Specs.

Rampage Formula x48
Q9550 Quadcore C1 @ 8.5x471
4x 2GB G.Skill PC2-7200 @ 942mhz 4-4-4-10, 2.19v

Edit: I don't have a power meter, sorry. This does give a good indication of power consumption by the chip. It is relatively spot on.

Someone move this thread to AMD section...Admins there wont allow this OT and bull posts

Yeah let's gag this thread too.

0.66W at idle... Then why don't NASA use overclocked Core 2 processors in their Mars robot?

You've been spending too much time on the other side.

[justapost]

I gave a disclaimer a few pages back; what you see is what you get. I'm on water ~ 72F ambients. I'm too lazy to use a multimeter right now, so software would have to do.
I thought this was a cpu to cpu comparison? I don't have the resources to do a platform comparison. See below.

System Specs.

Rampage Formula x48
Q9550 Quadcore C1 @ 8.5x471
4x 2GB G.Skill PC2-7200 @ 942mhz 4-4-4-10, 2.19v

Edit: I don't have a power meter, sorry. This does give a good indication of power consumption by the chip. It is relatively spot on.

Ok, I will use the IES tool on the AsRock board. It's main usage is to power off unused phases but it's also good to monitor the current amps the cores use (without the NB phase). But it will not say much about the whole systems consumption.

[Glow9]

What can I say , congrats to AMD for finally getting where Intel was in November 2006 !

Kentsifeld users can breathe a sigh of relief ; maybe in late 2009 AMD will finally have something able to win across the board vs. Kentsfield.

Most of the tests we saw the the q9450 and q9550 get beat or were close and a few months ago people didnt even know if it would be even beat a Q9450 in anything. Like I said before if it beats a Q9450 I'd pick on up seens like a lot of you ae bashing on results when it exceeded most peoples lower expectations, Too bad this review didnt have OC benchmarks to compare.

[railer]

no phail2 4 me. amd is mainstream now not a leader in performance. i had the power like that 2 years ago. i see not a single reason to switch from core2 to amd right now, it would be just a waste of $

i7 on the other hand is really powerful architecture. For multimedia pc it would be great, conversion times will improve allot from core2,

my only hope that with the great clock phail2 seams to get on LN2 it could compete on the ORB.

[Leeghoofd]

Awaiting bulldozer or I7 die shrink... like said before this is a nice step up for current AM2(+) users, maybe AM3 will bring a few percentages more with DDR3...

Think it's time to start a price fight !! let the pricewar begin !!! as the current retail price in my humble opinion is too high for the tested PII models... let's hope they clock 3.8Ghz on air !! and +4ghz on water

Nice work overclocker.gr

[OverClocker_gr]

in about an hour or so we will add X264 enconding comparison between

Q6600,Q9450,Q9550,i7 920 , pii 920 + 940

[Glow9]

in about an hour or so we will add X264 enconding comparison between

Q6600,Q9450,Q9550,i7 920 , pii 920 + 940

how bout with all them OCed make things interesting