3DMark Vantage Loves The Cores: QCs FTW !!

[Calmatory]

Looking on Q6600 vs Q9300 and others I think there was some synthetic memory bench there

This bench is another joke like 3Dmark when compared to real life.

Define real life. Then show a bench which isn't a joke when compared to - so called - "real life".

There can't be any bench which could accurately show "real life" performance of a CPU/GPU. Since CPU and GPU are build in different ways, with different parts, and then there are alot of different programs which work in very different way, no one can never make a benchmark which could alone show if something is better than other "in real life".

[The0men]

If 3d mark 06 is anything to go buy, which became CPU limited, then this should be a good way to bench the CPU's.

I too don't see how this is showing that AMD has anything even close the intel.

But it is showing that theyre getting better. Now if they can get this into a 45nm without errata and a lower consumption, they would make very fast HT or business computers.

[Shintai]

Define real life. Then show a bench which isn't a joke when compared to - so called - "real life".

There can't be any bench which could accurately show "real life" performance of a CPU/GPU. Since CPU and GPU are build in different ways, with different parts, and then there are alot of different programs which work in very different way, no one can never make a benchmark which could alone show if something is better than other "in real life".

Well would be nice if they didnt absolutely made benches that you couldnt compare at all to real world. Again..3Dmark

[The0men]

Well would be nice if they didnt absolutely made benches that you couldnt compare at all to real world. Again..3Dmark

Easy to say this..

But to me its pretty obvious which CPU will be faster when playing games, one that scores 8,000 or one that scores 10,000.

Game or synthetic, it still runs the CPU full steam, and gives you a good idea of real life preformance.

I'm not saying at all that its perfect, or that the FPS you get in a bench match that of a game.

What I'm saying is, mostly, if you bench a CPU and its clearly the winner of a bench, it will more than likely be faster at anything else using similar instructions.

[Epsilon84]

Yeah, you are right, some X3's are not that worth. But the lower end's X3 is not bad.

Of course, for us where overclocking is a big concern, Intel has a way bigger advantage.

Yeah, I was looking more at the O/C performance, if you count stock performance the X3 8450 does pretty well in the bang for buck stakes, but the E7200 for $133 would be my pick for the sub $150 bracket.

Anyhow, I'm not sure what the fuss is over this benchmark, I never particularly paid much attention to 3DMark scores at all, real world gaming performance is where its at for me.

[RaZz!]

mh, i have to agree with most of the posts here.

i mean, how exactly is this chart supposed to show phenom's competitiveness? sadly, it shows exactly the same picture/impression we had from phenom before: it's slower than competitive products from intel.

it's not bad either, but we already knew that from the like 23985629386293865 benchmarks we saw before

[hecktic]

one cpu seems to be missing... well two...

q9550
e8400

and why is a sempron in there lol ??

[RaZz!]

one cpu seems to be missing... well two...

q9550
e8400

and why is a sempron in there lol ??

i'd say they benched everything they had

[xlink]

I don't see how this is good at all.

highest score is on a q9770 @ 4GHz

benchers will be pushing those near 6 GHz.
and phenom still looses clock per clock and scales worse.

[Jakko]

Obviously 3dmark vantage is optimized for intel processors through sse or something. In most games phenom is not as fast as the competition, but the difference is usually not this big.

[RPGWiZaRD]

If you want a good CPU benchmark use PCSX2 PS2 emulator, performance scales linearly with overclocked CPU amount. However it only supports up to dual core, quads give no benefit at all.

[perkam]

First post updated.

For those who believe this thread is useless, threads gain momentum and support from posts, which move them up to the beginning of the list. Therefore, if one does not like a thread, they are welcome not to post.

It is a trade off. You post to give your opinion, which may be negative, but for that opinion to matter, it needs to be heard, and when it is, the thread gains value, as does the topic. Vice Versa, not posting in a thread you feel is wrong means the thread has less traffic but the content of the thread remains consistent with the topic of the first post, so it doesnt make any sense to post in a thread for its closure, its counter intuitive

The point of the thread was to point out that the Phenoms are closer to Intel offerings than they were in 3dMark 06. No one expected Phenoms to beat Intel in 3D Mark Vantage, and it would be a sham of a benchmark if that was the case. Everyone expected the Phenoms to be slower, but it turns out benching vantage with a Phenom vs an Intel qc will have a smaller impact this time around, giving AMD a much needed boost, as stated in the first post.

Perkam

[The0men]

First post updated.

For those who believe this thread is useless, threads gain momentum and support from posts, which move them up to the beginning of the list. Therefore, if one does not like a thread, they are welcome not to post.

It is a trade off. You post to give your opinion, which may be negative, but for that opinion to matter, it needs to be heard, and when it is, the thread gains value, as does the topic. Vice Versa, not posting in a thread you feel is wrong means the thread has less traffic but the content of the thread remains consistent with the topic of the first post, so it doesnt make any sense to post in a thread for its closure, its counter intuitive

Perkam

Don't get the wrong idea Perkam, the thread is far from useless.

I just feel unless AMD do something about the phenom (especially with the great deals on a Q6600 etc) these days, then its the phenom thats useless..

Well not useless, but being pushed behind because now even the prices are not so great when the lower end intel parts that have similar if not more preformance, can sometimes cost the same or less, but come with a smaller die, less heat, more OC potential, and less consumption.

I hope AMD hits a nice 45nm soon I really do.

[Mungri]

HOW on earth is this putting Phenom in a positive light?

Just to try and give an answer to everyone with this same question ...

The Phenom still 'sucks'. Theres no doubt that a quad core Phenom from AMD still cant compete with Intels Quad cores.

However, up untill now, in just about every benchmark and game so far, the etire Phenom range was even surpassed by the core 2 duo. It wasnt surprising to see reviews where the Intel E8400 was outperforming everysingle AMD processor available today. However, none of these earlier benchmarks and games were optimised for quad core proccessors.

Vantage however sees a very nice boost between dual and quad core processors, and for probably the first time yet, we can see the Phenom outperforming the Core 2 duos.

Although this performance will never be seen currently in non optimised games, and I still wouldnt ever buy any current AMD CPU, we do at least see that the phenom can outperform the core 2 duo at least.

But if you want a quad core, you should definately be buying the Q6600, and not the Phenom. They are still quite sucky in comparison.

[perkam]

Don't get the wrong idea Perkam, the thread is far from useless.

I just feel unless AMD do something about the phenom (especially with the great deals on a Q6600 etc) these days, then its the phenom thats useless..

Well not useless, but being pushed behind because now even the prices are not so great when the lower end intel parts that have similar if not more preformance, can sometimes cost the same or less, but come with a smaller die, less heat, more OC potential, and less consumption.

I hope AMD hits a nice 45nm soon I really do.

Its about envelopes right now.

AMD cant clock Phenoms higher without pushing them into higher wattage envelopes which take away from its "energy efficient" and "low power" monikers. So yes, 45nm should help.

Perkam

[mAJORD]

just look at the results, vantage is very cache friendly....

a phenom +0.1 speed increase provides roughly + 250/280 score depending on x3 or x4
same for the q9300-q9450 comparison + 278 points

but going from 7300-8300 @ same speed you increase more then 600 points
so increasing from q6600-q9300 gives you 0.1 speed advantage + 2x cache increase so that explains the big jump.

But a Q9300 has LESS cache than a Q6600. So it aint that (3mb vs 4Mb)

It does have larger associativity though, and less latency iirc.

also the FSB is higher.. a possibility

[Shintai]

Its about envelopes right now.

AMD cant clock Phenoms higher without pushing them into higher wattage envelopes which take away from its "energy efficient" and "low power" monikers. So yes, 45nm should help.

Perkam

125W+ aint energy efficient or low power

[NickS]

125W+ aint energy efficient or low power

Sad but true.

[perkam]

Sad but true.

So you finally decided to post your Dad's Intel rig as your own Acting all cool and such

Perkam

[Epsilon84]

The point of the thread was to point out that the Phenoms are closer to Intel offerings than they were in 3dMark 06. No one expected Phenoms to beat Intel in 3D Mark Vantage, and it would be a sham of a benchmark if that was the case. Everyone expected the Phenoms to be slower, but it turns out benching vantage with a Phenom vs an Intel qc will have a smaller impact this time around, giving AMD a much needed boost, as stated in the first post.

What are you talking about? The relative gap between C2Q and Phenom has actually increased with 3DMark Vantage compared to 3DMark 06, this is especially apparent with the 45nm C2Qs.

3DMark06 CPU Test numbers courtesy of THG CPU Charts 2007:
C2Q QX9650 @ 3GHz - 4357
C2Q Q6600 @ 2.4GHz - 3507
Phenom 9700 @ 2.4GHz - 3347

Extrapolating from the Phenom 9700 numbers, a Phenom 9850 would score approx 3500 points, effectively tieing with the Q6600 and about ~25% behind a QX9650.

Compare that to 3DMark Vantage:
C2Q QX9650 @ 3GHz - 12407
C2Q Q6600 @ 2.4GHz - 8598
Phenom 9850 @ 2.5GHz - 8340

As you can see, a QX9650 is now ~49% faster than a Phenom 9850, up from ~25% in 3DMark06. A Q6600 is now 3% faster than a Phenom 9850, whereas they would be virtually tied in 3DMark06. In all cases, the performance gap between C2Q and Phenom grows, slightly in the case of Kentsfield, significantly in the case of Yorkfield. I don't know how much clearer I can make this, your interpretation of the data is simply wrong.

Sorry perkam, your theory is busted.

[perkam]

What are you talking about? The relative gap between C2Q and Phenom has actually increased with 3DMark Vantage compared to 3DMark 06, this is especially apparent with the 45nm C2Qs.

3DMark06 CPU Test numbers courtesy of THG CPU Charts 2007:
C2Q QX9650 @ 3GHz - 4357
C2Q Q6600 @ 2.4GHz - 3507
Phenom 9700 @ 2.4GHz - 3347

Extrapolating from the Phenom 9700 numbers, a Phenom 9850 would score approx 3500 points, effectively tieing with the Q6600 and about ~25% behind a QX9650.

Compare that to 3DMark Vantage:
C2Q QX9650 @ 3GHz - 12407
C2Q Q6600 @ 2.4GHz - 8598
Phenom 9850 @ 2.5GHz - 8340

As you can see, a QX9650 is now ~49% faster than a Phenom 9850, up from ~25% in 3DMark06. A Q6600 is now 3% faster than a Phenom 9850, whereas they would be virtually tied in 3DMark06. In all cases, the performance gap between C2Q and Phenom grows, slightly in the case of Kentsfield, significantly in the case of Yorkfield. I don't know how much clearer I can make this, your interpretation of the data is simply wrong.

Sorry perkam, your theory is busted.

Well, I'm not here to go back and forth, but based on YOUR OWN numbers, the Q6600 was ahead by 7% in 3dMark 06 and is now only ahead by 2.5% in 3DMark Vantage. Ty for helping my argument out . (Remember that a 200 point difference to a 3000 pt score is 7% of the score, while a 200 point different to an 8000 point score is much smaller )

Though, I'm flattered that you went through the (if a tad incoherent) analysis to try to disprove me, but read my first post. The "EFFECT" of the final cpu score is so low in 3dmark vantage that, even with a 4ghz QX9650 and a 3ghz Phenom 9850, one can manage to get very similar scores, which was not possible before.

I would like to see you try to debunk my statement on Tri-cores in the first post

Perkam

[Epsilon84]

I'm flattered that you went through the (if a tad incoherent) analysis to try to disprove me, but read my first post.

The "EFFECT" of the final cpu score is so low in 3dmark vantage that, even with a 4ghz QX9650 and a 3ghz Phenom 9850, one can manage to get very similar scores, which was not possible before.

I would like to see you try to debunk my statement on Tri-cores in the first post

Perkam

Umm, your original post linked to the CPU score of 3DMark Vantage, not the overall score. You never mentioned the EFFECT it had on the overall score, heck, your OP (before the edit) was just a 2 sentence job saying how awesome Phenom was.

If all along your intent was to show how the weighting of the CPU score has changed from 3DMark 06 then you took a while to actually get your point across. I haven't actually examined the weighting system much (I'm not a hardcore 3Dmark bencher ), if you have such data please provide it.

With regard to X3, this is what I've found:
http://www.legitreviews.com/article/697/11/
3DMark06 CPU score:
C2D E8500 - 2941
X3 8750 - 2733
E8500 is 7.6% faster than X3 8750

3DMark Vantage CPU score:
C2D E8500 - 6251
X3 8750 - 5942
E8500 is 5.2% faster than X3 8750

So X3 makes a ~2.5% gain on C2D from 3DMark06 to 3DMark Vantage, are we supposed to get excited?

[perkam]

Umm, your original post linked to the CPU score of 3DMark Vantage, not the overall score. You never mentioned the EFFECT it had on the overall score, heck, your OP (before the edit) was just a 2 sentence job saying how awesome Phenom was.

If all along your intent was to show how the weighting of the CPU score has changed from 3DMark 06 then you took a while to actually get your point across. I haven't actually examined the weighting system much (I'm not a hardcore 3Dmark bencher ), if you have such data please provide it.

Yep, I really like edits, it allows me to point out the following which you fail to mention in your post

based on YOUR OWN numbers, the Q6600 was ahead by 7% in 3dMark 06 and is now only ahead by 2.5% in 3DMark Vantage. Ty for helping my argument out . (Remember that a 200 point difference to a 3000 pt score is 7% of the score, while a 200 point different to an 8000 point score is much smaller)

Be back in a few.

Perkam

[Epsilon84]

Yep, I really like edits, it allows me to point out the following which you fail to mention in your post

Be back in a few.

Perkam

Again, I'll remind you that your OP mentioned nothing regarding the overall score, or the weight system in 3DMark Vantage. You simply linked to a chart that showed Phenom getting dominated pretty badly by C2Qs. You are now completely changing the scope of the discussion with your edit.

I do agree that it does appear that the CPU score is less important this time around in determining the 'overall' score, thus making 3DMark Vantage a prodominantly GPU bound benchmark if you exclude the CPU portion of the tests.

[gOJDO]

3D Mark(any version) means nothing to real world performance. Phenom is as fast as it is, and one pointless synthetic benchmark(for e-penis measurement purposes) can't change the facts!