AMD Phenom II 920 & 940 Full Review [UPDATED with more tests]

[BenchZowner]

BenchZowner, can you run x264 benchmark with a newer x264 version?
The included are too ancient :P (8 months old the newer).

I've adjusted the bat file (and the vbs file to launch the new bat) to run one of the latest (if not the last, I've not checked today), r1069. The only issue is that fps results don't write to the rtf file and I can't figured out why Anyways, you can see the fps from the ms-dos window

The settings are the unmodified Standalone-Blu-ray_fast profile from Megui (only adjusted the bitrate to 3959 from 8000, just like the bitrate used with original x264 benchmark, and I guess this standalone profile is the equivalent to the one used with original bench).

You have these files attached inside a zip.

And here the x264 r1069 http://rapidshare.com/files/17938606...-1069.rar.html
Thanks.

Don't know, I'll ask the guys if they have the time to do another re-run.

[TL1000S]

Your card is much faster than 2900Pro, so the CPU score is higher because of it, not because the CPU itself.
P.S. I hate you, all that have already got their Denebs

This is wrong I am surprised you even post this rubbish here?
I really thought you would know better at this site.
I bench 3M06 extensively and swaps graphics card a lot.
You get the same CPU-score regardless of GPU. (Give or take a few points).

[BenchZowner]

This is wrong I am surprised you even post this rubbish here?
I really thought you would know better at this site.
I bench 3M06 extensively and swaps graphics card a lot.
You get the same CPU-score regardless of GPU. (Give or take a few points).

If you're talking about cards from the same manufacturer ( AMD or nVIDIA or Matrox ), then in general you are correct.
But the CPU score in 3D Mark06 differs wildly between a setup with a AMD card and a nVIDIA card.

[PetNorth]

Don't know, I'll ask the guys if they have the time to do another re-run.

Thanks! I hope they have the time, because I think these days nobody uses so old x264 versions

[Stuen4y]

This is wrong I am surprised you even post this rubbish here?
I really thought you would know better at this site.
I bench 3M06 extensively and swaps graphics card a lot.
You get the same CPU-score regardless of GPU. (Give or take a few points).

If you're talking about cards from the same manufacturer ( AMD or nVIDIA or Matrox ), then in general you are correct.
But the CPU score in 3D Mark06 differs wildly between a setup with a AMD card and a nVIDIA card.

The situation also changes when you use multi GPUs.

[TL1000S]

The situation also changes when you use multi GPUs.

Hmmm.. this contradicts my tests.

Have done several benches with 3DM06 with same MB/CPU (Q6600/Q9450) swithcing from 9600GT, 8800GT, 2xHD3870 and now HD3850x2.
The CPU score stays consistent - OS=WinXP 32-bits.

If you get 300 more pts as stated above by disabling services, then I would call that a "normal" user install (lotsa crap starting up) and not a "bencher" install.

Are you sure this is with the same OS? (Preferably a tweaked "bench-OSDisk" the same for all systems).

OK. Enough off-topic from me, and I could be wrong of course. It has happened before.

[justapost]

No one postet it yet I think.

Quad Core Round Up: AMD Vs Intel Clock to Clock

Respect for the hardwork.

Few questions. Do you have a power consumption deneb vs. agena and what voltage did you need for the deneb oc.? Turbo on on i7 I assume?

[Oliverda]

No one postet it yet I think.

Quad Core Round Up: AMD Vs Intel Clock to Clock

Respect for the hardwork.

Few questions. Do you have a power consumption deneb vs. agena and what voltage did you need for the deneb oc.?

This review is a little bit confused IMHO. And what's this?





I'm still waiting for a straight review with the following CPUs:

Phenom 940 (3.0GHz)
Phenom 9950 (OCed to 3.0GHz)

Core 2 Extreme QX6850 (3GHz)
Core 2 Quad Q9650 (3GHz)
Core i7 940 (OCed to 3.0GHz)

[Stuen4y]

I'm still waiting for a straight review with the following CPUs:

Phenom 940 (3.0GHz)
Phenom 9950 (OCed to 3.0GHz)

Core 2 Extreme QX6850 (3GHz)
Core 2 Quad Q9650 (3GHz)
Core i7 940 (OCed to 3.0GHz)

And why not one E8400 to compare the difference between quad amd and dual core intel, because they've been competing against dual core with their quads during 2008 .

[BenchZowner]

This review is a little bit confused IMHO. And what's this?





I'm still waiting for a straight review with the following CPUs:

Phenom 940 (3.0GHz)
Phenom 9950 (OCed to 3.0GHz)

Core 2 Extreme QX6850 (3GHz)
Core 2 Quad Q9650 (3GHz)
Core i7 940 (OCed to 3.0GHz)

Charts explanation:

Agena = Phenom I
Deneb = Phenom II
Both processors clocked at 3GHz and compared with each other to show the differences between these two.

Deneb = Phenom II
Kentsfield = Core 2 Quad ( 65nm ) [ Q6600 ]
Yorkfield = Core 2 Quad 9 ( 45nm ) [ Q9450 ]
Bloomfield = Core i7
All of them clocked at 3.7GHz for a clock per clock comparison.

I'm having trouble finding what's wrong with this comparison to you.

[Final8ty]

Charts explanation:

Agena = Phenom I
Deneb = Phenom II
Both processors clocked at 3GHz and compared with each other to show the differences between these two.

Deneb = Phenom II
Kentsfield = Core 2 Quad ( 65nm ) [ Q6600 ]
Yorkfield = Core 2 Quad 9 ( 45nm ) [ Q9450 ]
Bloomfield = Core i7
All of them clocked at 3.7GHz for a clock per clock comparison.

I'm having trouble finding what's wrong with this comparison to you.

I think he maybe thinking that there is something wrong with the graph because the assumption is that Deneb will be faster/equal to but never slower than Agena.

[loutsos]

I think the most interesting part of the clock2clock review is the power consumption.

AMD looks pretty bad there...

[justapost]

This review is a little bit confused IMHO. And what's this?

Maybe bios problems? Many results look worse than the ES previes we saw.

I'm still waiting for a straight review with the following CPUs:

Phenom 940 (3.0GHz)
Phenom 9950 (OCed to 3.0GHz)

Core 2 Extreme QX6850 (3GHz)
Core 2 Quad Q9650 (3GHz)
Core i7 940 (OCed to 3.0GHz)

Yepp, power consumption would look abit different here.

[justapost]

I think the most interesting part of the clock2clock review is the power consumption.

AMD looks pretty bad there...

Ask for details befor you complain please.

[biohead]

This review is a little bit confused IMHO. And what's this?

probably 1.8 vs 2.0 ghz NB clock.

i'd like to see some 3 ghz NB performance. surely that's possible under cold with some volts and pineapples.

[BenchZowner]

I think both processors NB clocks were equal in the tests, but I'll have to check it out again to be sure.
It's absolutely possible though.
Being a new and supposedly faster ( and truly faster in most apps ) processor doesn't make it unbeatable, and of course doesn't mean that the older "revision" can't surpass it in a few tests.

p.s. I doubt that the NB clock would make any difference in the FPU tests, can't guarantee that, but I'm pretty sure

[Oliverda]

Maybe bios problems? Many results look worse than the ES previes we saw.

Yepp, power consumption would look abit different here.

Yes it would be a little bit different and the scores as well because the i7's NB and the Core 2 processors FSB would be at stock clocks and not overclocked.

probably 1.8 vs 2.0 ghz NB clock.

i'd like to see some 3 ghz NB performance. surely that's possible under cold with some volts and pineapples.

If I'm not mistaken the X4 940's NB/HT stock clocks are 2GHz.

[bedlamite]

Nice work! And very good conclusion.

[justapost]

Yes it would be a little bit different and the scores as well because the i7's NB and the Core 2 processors FSB would be at stock clocks and not overclocked.

I remember 7% advantage in wprime and ~10% in CB10, compared to Agena. It are 3% for wprime and ~1% for CB10 now.

If I'm not mistaken the X4 940's NB/HT stock clocks are 2GHz.

Odd why did they use 1,8GHz then? Too much work at once.

[informal]

I remember 7% advantage in wprime and ~10% in CB10, compared to Agena. It are 3% for wprime and ~1% for CB10 now.

Exactly.Oddly enough the Greek tests don't reflect that .It could be a bios issue,a testing error,OS issue or who knows what

[Hornet331]

well only a few more days till more data should be available and then we see whats up with that.

[informal]

Oh and that winrar score seems a bit low compared to what we'd expect from Phenom II.But a few more days and we'll see what's up with those numbers .

[accord99]

I remember 7% advantage in wprime and ~10% in CB10, compared to Agena. It are 3% for wprime and ~1% for CB10 now.

Er, didn't your Agena beat Coolice's Deneb in Wprime?

http://www.ocxtreme.org/forumenus/sh...03&postcount=8

And the Deneb score for 32-bit Cinebench 10 looks right, it's just the Agena score that seems higher than expected.

[informal]

Er, didn't your Agena beat Coolice's Deneb in Wprime?

http://www.ocxtreme.org/forumenus/sh...03&postcount=8

And the Deneb score for 32-bit Cinebench 10 looks right, it's just the Agena score that seems higher than expected.

As for wprime and Agena beating Deneb,the culprit is the OS used.Justapost already spoke about this IIRC.
Cinebech score for Agena is a bit high in hwbox review while the Deneb one is a bit low,so this could be the reason fo so close scores.I remember the 7-9% better scores per clock from the initial tests with C1 Denebs.

[Oliverda]

I remember 7% advantage in wprime and ~10% in CB10, compared to Agena. It are 3% for wprime and ~1% for CB10 now.

Odd why did they use 1,8GHz then? Too much work at once.

Exactly. Just take a look at this Shanghai review:





(please note that this is dual CPU configuration - 8 cores)

source


I multiplied up the Barcelona's results.

1 CPU benchmark ~24% advantage over Barcelona

Multi CPU ~8,5% advantage over Barcelona



mod: I found this CB result here in the XS few weeks ago.