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

[Jakko]

http://www.xcpus.com/forums/motherbo...e.html?garpg=5

Very strange result, some mentioned PCI Express bandwidth may have something to do with it. I was wondering whether this effect can be reproduced with different titles...

Since most cpu gaming benchmarks focus on low res and no AA testing, could it be that this "bandwidth benefit" over intel processors has been missed by all other website/reviewers?

Jumpingjack did an excellent job with this article, but since he is the first ever to come with benchmarks in which the phenom beats intel clock for clock, I am wondering what's going on.

[Oldguy932]

Maybe its just that at that point there's just so much data getting shoved to the gpu that the fsb isn't fast/wide enough to get it to the gpu as quick as the ht link is.

[Jakko]

I combined the cpu intensive benchmark (cave) with the gpu intensive benchmark (snow) and that leads to the following picture:


The upper red and blue squares show the gpu limited performance(snow), and the lower squares show the cpu limited performance(cave).
A few things can be concluded.

- At resolution 1024*768 and lower, the phenom is cpu limited in a gpu intensive level
- At all resolutions, all processors are cpu limited in a cpu intensive level, phenom performing about 20% less than intel
- At resolutions above 1024*768, where the limiting factor is the gpu, the phenom outperforms intel with about 15% at the highest resolution.

[demonkevy666]

Maybe its just that at that point there's just so much data getting shoved to the gpu that the fsb isn't fast/wide enough to get it to the gpu as quick as the ht link is.

also memory bandwidth plays a role.

[BenchZowner]

What makes you think that running a QX9650 @ 2.5GHz and a Phenom @ 2.5GHz along with a decent graphics card ( 8800GTS 512MB for example ) and the following real-life gaming conditions ( 1280x1024 4xAA 16xAF, 1680x1050 4xAA 8xAF, 1920x1440 2xAA 8xAF ) with various games will show any difference between the two systems, and also give the AMD phenom based system a lead ?

It's not going to happen, and it's called GPU Limited scenario ( as real-life gaming is ).
You may test 100 titles...maybe there will be one title where AMD will be 1fps faster...but on most of them it will be lagging behind by a tad ( max 2fps lower )

[Jakko]

What makes you think that running a QX9650 @ 2.5GHz and a Phenom @ 2.5GHz along with a decent graphics card ( 8800GTS 512MB for example ) and the following real-life gaming conditions ( 1280x1024 4xAA 16xAF, 1680x1050 4xAA 8xAF, 1920x1440 2xAA 8xAF ) with various games will show any difference between the two systems, and also give the AMD phenom based system a lead ?

It's not going to happen, and it's called GPU Limited scenario ( as real-life gaming is ).
You may test 100 titles...maybe there will be one title where AMD will be 1fps faster...but on most of them it will be lagging behind by a tad ( max 2fps lower )

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!

[jas420221]

I want to see this test repeated by someone else, no offense to JJack. If its a truely empircal test, it should be repeatable by someone else.

Bench's post is actually spot on for the topic. Stop being an antagonist! You condescending tone towards a lot of people really take away any merit you *may* have in your threads.

[Jakko]

Bench's post is actually spot on for the topic. Stop being an antagonist! You condescending tone towards a lot of people really take away any merit you *may* have in your threads.

This thread is not about whether or not the benchmark is flawed. It's about why the benchmarks show what they show. Obviously some people will be so upset by a discussion that involves AMD outperforming intel that they will feel the need to question the article on which the discussion is based, which is fine, but they can start their own thread for that.

At this point there is no reason to question JJ's benchmarks.

[jas420221]

This thread is not about whether or not the benchmark is flawed. It's about why the benchmarks show what they show. Obviously some people will be so upset by a discussion that involves AMD outperforming intel that they will feel the need to question the article on which the discussion is based, which is fine, but they can start their own thread for that.

At this point there is no reason to question JJ's benchmarks.

Im not questioning the benchmarks, I just want to see the results repeated by someone else. Im the type of person that if you tell me the sky is blue, I will need to go outside and check it out for myself. Since I didnt buy that stupid arse game, I cant do it myself (nor do I have a Phenom )

Personally, I could care less who shows a minor lead at a specific resolution in one game in a couple levels of that game.

Anamoly perhaps?

[BenchZowner]

Hi Bench! Welcome to this topic!

Ironically you fail to impress me with your "tone".
Since I'm not welcome here, I'll just make my point clear with a short text response.

Since you are having hard time reading anything except the thread title, here's some help:

If this is a attempt to spice up things, you've failed.Sorry.
I take it it's your way to make humor.Bad sense of humor, but who cares.

However... I did read your post, twice.

1. this thread is about an article made by jumpingjack

Very nice.

2. in the article phenom outperforms intel quad clock for clock in gpu intensive levels of lost planet at high resolutions

Excuse me, but to come to a conclusion you need to have a clear image of what happened, how it happened & why.
Unfortunately you didn't pay the attention needed to comprehend the test.
Let me help you out:

The Lost Planet game was set to minimum settings on all graphical options except effect quality (which was determined to have significant CPU component) and filtering, these two options were set to high, a screen shot of the graphics settings page is shown below.

No anti-aliasing. All game settings set to minimum apart from the effect quality setting.
This is making this test CPU limited at any resolution up to 1920x1440.

3. this thread is about how this is possible

Explained above.

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.[/quote]

I'm never in denial. I'm not a lawyer neither a manufacturer's rep.
Unfortunately you don't have the right to tell us if we can or cannot judge the article's testing methodology and results.
Once again, the explanation is shown 2 "quotes" above.

Have a nice day!

I will.

[Jakko]

No anti-aliasing. All game settings set to minimum apart from the effect quality setting.
This is making this test CPU limited at any resolution up to 1920x1440.

As you can see in the snow benchmark, the benchmark is not cpu limited looking at the intel performance, and only partly cpu limited (1024 * 768 and lower) for the AMD cpu.
Where did you get the idea the test was cpu limited?

And even if it was cpu limited, that still wouldn't explain why the phenom beats the intel consistantly at resolutions higher than 1024 * 768.

[BenchZowner]

As you can see in the snow benchmark, the benchmark is not cpu limited looking at the intel performance, and only partly cpu limited (1024 * 768 and lower) for the AMD cpu.

Without AA and details set to minimum again, it is kinda CPU Limited.
Not as much as the other test but still is.
There are details that I have to look at more, and maybe some stuff that aren't mentioned at all in the article.
But I have more serious things to do over this and to be honest, be it faster or not, I don't really care about it.

So...let's just say that AMD is faster in Lost Planet.
1 game out of... hundreds.
Shall we go out and celebrate about it like we've found the love of our lives ?

Where did you get the idea the test was cpu limited?

Just run the same test with varying CPU Frequencies by adjusting the CPU multiplier, and you'll see.

And even if it was cpu limited, that still wouldn't explain why the phenom beats the intel consistantly at resolutions higher than 1024 * 768.

Once again, I haven't checked some details... for example the AMD configuration might have better RAM bandwidth and latency, which helps in performance in CPU Limited situations.

Off to work now.

[Jakko]

Without AA and details set to minimum again, it is kinda CPU Limited.
Not as much as the other test but still is.

I think you are confusing the two tests. "The other test" being the cave test, is completely cpu limited on all resolutions on all processors because of the insane amount of animation and physics going on in that level. (many enemnies on screen I think it was)

There are details that I have to look at more, and maybe some stuff that aren't mentioned at all in the article.
But I have more serious things to do over this and to be honest, be it faster or not, I don't really care about it.

So you are saying you didn't really look at the details, and even if amd proves to be faster you don't really care about it.
Why are you posting in this thread?

So...let's just say that AMD is faster in Lost Planet.
1 game out of... hundreds.
Shall we go out and celebrate about it like we've found the love of our lives ?

Lol, what is wrong with you? Just trying to understand some surprising gaming benchmarks, you're acting weird for someone who "doesn't really care".

Once again, I haven't checked some details... for example the AMD configuration might have better RAM bandwidth and latency, which helps in performance in CPU Limited situations.

Off to work now.

I do not expect JJ to test like that tbh.

[biohead]

Once again, I haven't checked some details... for example the AMD configuration might have better RAM bandwidth and latency, which helps in performance in CPU Limited situations.

If that's the case, isn't that contradictory to the point you're trying to make?

[AlabamaCajun]

Didn't take long for this to become a "mines is bigger'n yours" thread.

[biohead]

I will remain a child.

[Jakko]

Oh comon biohead and alabama, derailing is part of forums, deal with it.
Instead of further filling the thread up with crap, you could also make a post that has something to do with the topic.

[jas420221]

You are trying to ascertain facts off something that cannot be proven one way or anthoer. JJack even said so himself. We can all speculate as to why this is...which is the point to this thread, but in the end, its all speculation. Like that tag says, Another Jakko thread!

....further experiments would be needed to reach any form of conclusion.

Invariably comparisons will be made between these two CPUs, and using gaming data to ascertain the architectural strength is sketchy at best.

But hey, who am I to try to stop you from squeezing lemon juice from an orange!

[Jakko]

You are trying to ascertain facts off something that cannot be proven one way or anthoer.

Why not?
Further testing could prove a lot. Someone with a phenom and an intel system could first try to get the same results. Or try a different game with similar settings and see if results differ.

There are a hundred ways to get more information really.
And I don't mind the tags, as if anyone takes that stuff seriously.

And I am really not determining architectural strength by using gaming data, I am trying to determine gaming strength by using gaming data.

[Devil's Prophet]

It's about why the benchmarks show what they show.

"Have you taken yourself and/or the system into the equation then?" is what Jas and BenchZowner are asking you imho.

[Jakko]

Myself into the equation?
How would such a thing work?

And the systems, sure I have taken those into the equation.
That is the first thing I looked at, thinking it was a memory issue.

http://www.xcpus.com/forums/motherbo...d.html?garpg=7

Both systems use the exact same memory with the same settings so that can't be it.

[DIE-TEMP]

I own both a Phenom 9500 rig and a Q6600. My AMD is clocked to 2800mhz for 24/7 use with 2GB of DDR2-1000.
My INTEL is clocked to 3400mhz for 24/7 use again with 2GB of DDR2-1000.

both rigs have an 8800GT 512MB and run off 250GB sata HDD's.

When it comes to 3D MARK 06, the INTEL scores 17k the AMD scores 14.5k.

When it comes to Super Pi the INTEL computes 1M in 13.2secs whilst the AMD computes 1M in 23-24secs.

No brainer picking the benchmark winner there.

For multi-tasking with resource hungry apps the AMD rig runs smoother no doubt.

There's an easy experiment which users can try which will show what I mean.

Download an app call WALLMASTER. It's designed to change your wallpaper automatically every 1-100minutes (user selectable).
Install this tiny app and select 10 wallpapers from your fav backgrounds and set the app to switch desktops every 1minute. Once you have setup wallmaster and have it running, fire up your fav 3D game.

INTEL users will experience severe stuttering ingame each and everytime wallmaster loads a different wallpaper. AMD users will experience no such effect.

IMO both cpu's have their advantages over each other. I do believe however that AMD have a "smoother" feel to them as a result of decreased latency and an IMC.

[BenchZowner]

I think you are confusing the two tests. "The other test" being the cave test, is completely cpu limited on all resolutions on all processors because of the insane amount of animation and physics going on in that level. (many enemnies on screen I think it was)

It's not only the scene and what's happening that makes it CPU or GPU limited.
A CPU Limited scene can easily become GPU Limited.
A GPU Limited scene can also become CPU Limited.
It's just how you set up things.

So you are saying you didn't really look at the details

Don't try to twist things & posts.
I said I didn't check ALL the details.
And the charts could easily contain more information, so people don't have to cruise all over the article to find what they're looking for ( especially if they're in a rush )

and even if amd proves to be faster you don't really care about it.

1) Because I don't game at LOW settings and without AA.
2) Even if AMD dominated this game at any settings, I wouldn't care as well.
Why ? Because it's just 1 game. I'm not using my CPUs to play games only, neither play this game only. I do several stuff with them, and AMD's current desktop CPUs lineup lags behind Intel's offerings at the moment in most tests/apps/usage scenarios.

Why are you posting in this thread?

1) To annoy you... NOPE.
2) Because I can ? NOPE.
3) Because I'm seeing things and note errors or people misinterpreting stuff.
Or just want to express my opinion, thoughts, ideas, etc.

[Jakko]

Ok bench, that is cool and all.
I think the cpu/gpu limitations are as I explained in the post with the editted picture. I could be wrong though, if so feel free to explain why.


Intel puppet thanks for the info.
Do you think you could clock the intel down to AMD's speed and do some high resolution gaming benchmarks at some point?

Would be cool to see if the strange behavior can also be observed with different games and GPU's.

[perkam]

Didn't take long for this to become a "mines is bigger'n yours" thread.

Well, at least we get Intel users to post in the AMD section this way

Perkam