Why does phenom perform better than an intel quad at high resolution in Lost Planet?

[JohnMike]

So a Quad-Core Core 2 45nm @ 4GHz ( 333MHz FSB ) & RAM @ DDR3-1333 7-7-6-21 fully tweaked ram has the same performance with a Dual-Core Core 2 45nm @ 2GHz & RAM @ DDR3-800 8-7-6-21 relaxed sub-timings...

As i said before....

But most of the games nowadays is GPU limited. Enable all settings and playing with high-resolutions, CPU power becames useless...... In this scenario GPU limited, having an Intel or AMD CPU doesn't matter in game performance.

[Devil's Prophet]

Yes...you would not be in "most".

I think you'd be surprised when you get all the statistics about that claim. :x

edit: I'm also including the casual gamers that just play games on their computer. Not only the max quality graphics fiends that some may see as the only true type of gamer. Anyway, most gamers just play on their or everybody's rig with the graphics card they got it with. Numbers, settings or even fluency doesn't mean as much to them as it does to most "xtreme" members here.

[BenchZowner]

As i said before....

I knew that already mate.
Just did the tests for the non-believers

[GeorgeStorm]

Hi,
BenchZowner, Im sure you do play at those settings, however you shouldnt disregard or mock other peple tests if they arent at the settings you play at, i mean, i dont play at those settings, along with many others, i dont have the money to afford a system that could do those settings, and I find it intruiging that the phenom does well considering overall its performance is somewhat lacklustre.

Just me................

[IvanAndreevich]

Pretty sure you can't put an X2 in a 790i...

The Phenom board I meant, regardless of what chipset it uses.

[BenchZowner]

Could you repeat this with 640x480 and low details, no fSAA.... to see how the core count scales when the full responsibility of the load is on the CPU. I will do the same for DX9 under XP.

It's hard for me to refuse such a kind request.
Starting the tests now.
BRB

[BenchZowner]

Hi,
BenchZowner, Im sure you do play at those settings, however you shouldnt disregard or mock other peple tests if they arent at the settings you play at, i mean, i dont play at those settings, along with many others, i dont have the money to afford a system that could do those settings, and I find it intruiging that the phenom does well considering overall its performance is somewhat lacklustre.

Just me................

I'm not mocking at anyone's tests, and surely not JumpingJack's.
In case of a very low end graphics card, you'll be gaming at really low resolutions and low to minimum settings and of course without AA/AF.
It's already been discussed in the past ( not only here on XS ) that you just need to overclock the cr*p out of your CPU.
In those settings ( low/minimum game detail settings, low resolutions with NoAA NoAF ) you'll get a tiny to none performance increase by overclocking your VGA.
But overclocking your processor will give you a nice boost ( in most cases ).

However, if you're targeting on gaming, and you have a small budget, logic says "Go for a cheap dual-core CPU, 2GB RAM, and get the best VGA you can get with what's left".

[jas420221]

I think you'd be surprised when you get all the statistics about that claim. :x

edit: I'm also including the casual gamers that just play games on their computer. Not only the max quality graphics fiends that some may see as the only true type of gamer. Anyway, most gamers just play on their or everybody's rig with the graphics card they got it with. Numbers, settings or even fluency doesn't mean as much to them as it does to most "xtreme" members here.

In this context, my point remains true. The particular audience is this case is XS. Not to mention, those same people that know that website and article exist, are usually not the people playing games on 'everybody's rig' either. Remember, I was only speaking of under the guise of the audience here and the demographic the article was shooting for...which, IMHO, was not intended for most of the HP/Dell/Compaq/Emachines/IPlayOnmyLaptopForSomeStrangeReasonWhenIHaveABette rSystemToPlayOn people. My apologies if I was unclear in what I was alluding to.

EDIT: I have to believe from Bench's posts that he is talking about the same audience unless otherwise specified (like he did in his last post talking to George).

[BenchZowner]

Here you go Jack:

System Specs:
Asus Striker II Extreme 790i Ultra SLI motherboard ( BIOS 0704 )
Intel Core 2 Extreme QX9650 C0 @ 2.5GHz - 7.5x333 ( 333MHz FSB )
2GB CellShock DDR3-1800 @ DDR3-1333 8-7-6-21 ( loose timings from my first test post in this thread ) - Linked/Synch with the FSB
Reference GeForce 9800GTX 512MB ( Driver 175.16 )

Game Settings: Your settings from your article
Driver Settings: Your settings from your article

Results:

[JumpingJack]

Here you go Jack:

System Specs:
Asus Striker II Extreme 790i Ultra SLI motherboard ( BIOS 0704 )
Intel Core 2 Extreme QX9650 C0 @ 2.5GHz - 7.5x333 ( 333MHz FSB )
2GB CellShock DDR3-1800 @ DDR3-1333 8-7-6-21 ( loose timings from my first test post in this thread ) - Linked/Synch with the FSB
Reference GeForce 9800GTX 512MB ( Driver 175.16 )

Game Settings: Your settings from your article
Driver Settings: Your settings from your article

Results:

Ohhhhh good data, many many thanks.... I will post my runs shortly.

jack

[adamsleath]

now it 's interesting.

[JumpingJack]

BenchZowner -- again, thanks for contributions. As a comparision, I did the same. To ensure no confusion, recall my data is being done on systems with 8800 GTX, Windows XP, thus DX9....

The raw data can be found here: Thread Scaling

Here is the summary info for low res/low detail thread scaling:



Plotted up a slightly different way:


Finally, a table showing scaling factors normalized to a single threaded run for each seen with respect to each CPU:


So here the native quad advantage shows up -- from single thread to 4 threads, the scaling factor of the phenom scales with respect to the single threaded performance better.

Jack

[mAJORD]

If anyones interested in an i9300 vs a9850 compo also, I have both puppies here, won't take long to inst identical Vista installs and do some testing

[Jakko]

Interesting stuff Jack!
It kinda makes you wonder how amd would compare once clockspeeds raise.

Do you think you could do that snow benchmark on a high resolution with settings you used where the phenom was faster, but this time with a brisbane cpu at a fairly high frequency?

I really wonder what would be the result.

[adamsleath]

interesting to see some more data; thx for that.

[Calmatory]

Some people don't udnerstand that even if X wins Y in 932840923840280 tests, it might be very wenn possible that Y remains clearly faster in test Z.

[BenchZowner]

Some people don't udnerstand that even if X wins Y in 932840923840280 tests, it might be very wenn possible that Y remains clearly faster in test Z.

So you're telling us to buy something that's better in 1 app than the other that's faster in all the rest ?

[Calmatory]

So you're telling us to buy something that's better in 1 app than the other that's faster in all the rest ?

Can't you read or are you being stupid on purpose?

[BenchZowner]

Can't you read or are you being stupid on purpose?

There are already a dozen reviews comparing those CPUs in several tests.
Actually quite a lot.

p.s. Thanks for calling me stupid, I feel flattered.

[keithlm]

Some people don't udnerstand that even if X wins Y in 932840923840280 tests, it might be very wenn possible that Y remains clearly faster in test Z.

You are CORRECT. There are also several other things people ignore when looking at benchmark reviews:


1. Summaries of individual reviews mean nothing.

Some people also have a hard time understanding that a "review" article is not a "distinct" point of demarcation. In other words you can't read a bunch of reviews and declare a winner or loser based on results from the review level. Saying the following means absolutely nothing: "This unit won 8 reviews and the other won 3 reviews."

This is because a single reviewer can keep adding specific benchmarks to get the outcome of the review to reflect a preferred viewpoint. Actually at the current time most available reviews provide a ton of "game" benchmarks which actually do not reveal anything. See point #2 below.


ANYWAY: to ascertain reality, you must compare the results of the same benchmark from various reviewers and analyze what is being revealed.

====================


2. People take scores of benchmarks as an exact science. They don't realize or understand that when scores are within a certain percentage: there is not really a "win" and "lose" situation but a "tie".

Some people prefer the absolute results method. They may claim: "My preferred choice won 50 games and yours only won 5". When in reality most of the results were a tie and there is not winner or loser.

I guess the problem in this particular situation is not whether this is true or not... but rather WHAT PERCENTAGE of standard deviation would be acceptable to use to determine a "tie"? (Well not totally true... you'll still get the people that want to use only exact results; they probably don't understand the concept being discussed here.)

You will also get people that don't understand that sometimes a reviewer might have completely anomalous results compared to other reviewers on a particular test. Either this variance must be explained or the results should be thrown out.

=====================


I often find on forums that people want to claim that using the above methods is "biased". In truth it is less biased... the OUTCOME can go either direction. It is only "biased" if it doesn't provide the results YOU want to see. When you analyze benchmarks reviews using the above methods it is very possible for the results revealed to be completely reverse of what most people might actually believe is true. (And of course after using these methods... I ended up with a Phenom. So now I AM biased. But actually using the methods discussed above is NOT biased.)

Oh... and the Phenom is known by many to be "smoother". (A year ago I got all but castrated for attempting to use that very word... now there are MANY people using the SAME word. Go figure.)

(Sorry... Off Topic. This thread is actually an attempt to explain an anomalous result as described above.)

[Calmatory]

There are already a dozen reviews comparing those CPUs in several tests.
Actually quite a lot.

p.s. Thanks for calling me stupid, I feel flattered.

So now you might want to show me when I told "you" to buy Phenom in my post. Until then I declare you stupid.

[BenchZowner]

So now you might want to show me when I told "you" to buy Phenom in my post. Until then I declare you stupid.

You were obviously talking about Phenom vs Core 2 Quad.
You were talking about it in a thread that some people believe that Phenom runs faster in real-life conditions a specific game.
And you say "Since the Phenom is faster in this game, what makes you think that it may not be faster in another game?"

So in other words you said "phenom might be faster in other games that we haven't tested".

With the same "way" you can easily come up with "Since the C2Q wins in 50 of the 52 tests, we can assume that it's faster in other (untested) apps/games/etc".

Heck we're once again taking things into a personal level without a single reason to do that.

I don't know what made you "insult" me, but if my first post was "confusing" or just written in a unclear way and you felt insulted, then I apologize.

[JumpingJack]

Interesting stuff Jack!
It kinda makes you wonder how amd would compare once clockspeeds raise.

Do you think you could do that snow benchmark on a high resolution with settings you used where the phenom was faster, but this time with a brisbane cpu at a fairly high frequency?

I really wonder what would be the result.

I can, and will -- unfortunately, do not expect data soon -- my plans are not to swap out the 9850 for a while.

I did drop the HTT bus down to 800 Mhz on the 9850 and retested, made just a few FPS difference, not much at all. The biggest modulator for the 9850 was memory speed, DDR2-667 to DDR2-1066 (in the increments the dividers allow) showed a total of around 10-12 FPS in either script. (I don't have the data in front of me, and have not been very systematic, I will go back and do that shortly).

There is one difference between the two setups -- all components were matched, and bought at the same time, except the 8800 GTXs, one was purchased about 8 months prior (the one in the phenom). The BIOS is different, so to be thorough I need to swap the cards between the two rigs and doublecheck.

Jack

[JumpingJack]

If anyones interested in an i9300 vs a9850 compo also, I have both puppies here, won't take long to inst identical Vista installs and do some testing

Actually, yes... I just picked up a i9300 myself, so getting someone to run benches to check reproducibility across different configs would be interesting to.

What MB, memory, OS?

EDIT: What I am finding (initial info) is that the 6 Meg (1/2 the cache) at the same clock has a bigger impact than what I would have thought.

[en4cer]

Hi Bench! Welcome to this topic!
Since you are having hard time reading anything except the thread title, here's some help:

1. this thread is about an article made by jumpingjack
2. in the article phenom outperforms intel quad clock for clock in gpu intensive levels of lost planet at high resolutions
3. this thread is about how this is possible

If you're just going to be in denial about the benchmarks shown in this article, you might as well open your own thread about this, as this thread is not about whether or not the benchmarks are right. It's about why the benchmarks show what they show.

Have a nice day!

LOL

Well said bro


this is the Intel fanboy BS that haunts these boards no matter where you look.