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

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.

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.

I combined the cpu intensive benchmark (cave) with the gpu intensive benchmark (snow) and that leads to the following picture:
http://www.music-in-games.com/pictures/phenomcsqx.jpg

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.

Quote:

---

Originally Posted by 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.

---

also memory bandwidth plays a role.

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 )

Quote:

---

Originally Posted by 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 )

---

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!

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.

Quote:

---

Originally Posted by jas420221

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.

Quote:

---

Originally Posted by Jakko

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. :up: Since I didnt buy that stupid arse game, I cant do it myself (nor do I have a Phenom :ROTF:)

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?

Quote:

---

Originally Posted by Jakko

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.

Quote:

---

Originally Posted by Jakko

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.

Quote:

---

Originally Posted by Jakko

1. this thread is about an article made by jumpingjack

---

Very nice.

Quote:

---

Originally Posted by Jakko

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:

Quote:

---

Originally Posted by JumpingJack's review

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.

Quote:

---

Originally Posted by Jakko

3. this thread is about how this is possible

---

Explained above.

Quote:

---

Originally Posted by Jakko

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.

Quote:

---

Originally Posted by Jakko

Have a nice day!

---

I will.

Quote:

---

Originally Posted by BenchZowner

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.

Quote:

---

Originally Posted by Jakko

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 ?

Quote:

---

Originally Posted by Jakko

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.

Quote:

---

Originally Posted by Jakko

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.

Quote:

---

Originally Posted by BenchZowner

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)

Quote:

---

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?

Quote:

---

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".

Quote:

---

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.

Quote:

---

Originally Posted by BenchZowner

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?

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

I will remain a child.

http://www.newsinferno.com/wp-includ...n-Derail-3.jpg

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.

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! :up:

Quote:

---

....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!

Quote:

---

Originally Posted by jas420221

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.

Quote:

---

Originally Posted by Jakko

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.

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.

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.

Quote:

---

Originally Posted by Jakko

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.

Quote:

---

Originally Posted by Jakko

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 )

Quote:

---

Originally Posted by Jakko

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.

Quote:

---

Originally Posted by Jakko

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.

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.

Quote:

---

Originally Posted by AlabamaCajun

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 :rofl:

Perkam

Since you guys want more on it.
I'll run some ( only Core 2 Quad at the moment though ).

Quote:

---

Originally Posted by demonkevy666

also memory bandwidth plays a role.

---

Not as big a role as you would think.... I did AMD at DDR2-667, 800, 1066 and the runs varied by about 5-8 FPS.

The QX9650 actually lost 2 FPS at DDR2-1067

Jack

Quote:

---

Originally Posted by intel puppet

IMO both cpu's have their advantages over each other.

---

The way I see it is that both cpu's have their disadvantages over each other likewise.

I just want a perfect rig, damnit. :(

I like brute power, but I also like multicore smoothness.

For the most part all I've seen is Phenom running a little smoother in a number of gamed and some intels doing most things faster. In most cases the intel showed more power.

Quote:

---

Originally Posted by Devil's Prophet

T
I just want a perfect rig, damnit. :(

---

NO! Perfection is not achievable else we can't build a better-one tomorrow :D

Jako :p

At least we are reading and showing out presence while trying to lighten the atmosphere a little. In your defence, most of the time we would be more interested in results but most of it turns into arguments backed by shaded internet facts. The train wreck was cool.

Quote:

---

Originally Posted by JumpingJack

Not as big a role as you would think.... I did AMD at DDR2-667, 800, 1066 and the runs varied by about 5-8 FPS.

The QX9650 actually lost 2 FPS at DDR2-1067

Jack

---

Thanks for that info, that makes it even stranger.
Basicly the only area in which the phenom had a distinct advantage was the mem BW...

I guess it could it even be the 790FX chipset vs the X48.
But then, they both offer PCIE *16, so that shouldn't matter.
:shrug:

Quote:

---

Originally Posted by Jakko

Thanks for that info, that makes it even stranger.
Basicly the only area in which the phenom had a distinct advantage was the mem BW...

I guess it could it even be the 790FX chipset vs the X48.
But then, they both offer PCIE *16, so that shouldn't matter.
:shrug:

---

More strangeness .... I re-ran a 1280x1024 run using an HTT multiplier to set the HTT link to 800 Mhz (that's down from 2000 Mhz):

Here is the data for both runs:

Phenom 9850 2.5 GHz 2.0 GHz NB 2.0 Ghz HTT and DDR2-1066
http://www.xcpus.com/gallery/d/6553-...80x1024_r1.JPG

Phenom 9850 2.5 GHz 2.0 GHz NB 800 MHz HTT and DDR2-800
http://www.xcpus.com/gallery/d/6556-...80x1024_r1.JPG

The data at 10280x1024 in the article is 2.5 GHz Core 2.0 GHz NB 2.0 GHz HTT and DDR2-800

Neither snow or cave show much of a huge difference, Snow gets about 8 FPS (EDIT at low 'CPU limited regime' resoluionts) and generally better with DDR2-1066 but the HTT at 800 has no effect.

jack

Quote:

---

Originally Posted by JumpingJack

More strangeness .... I re-ran a 1280x1024 run using an HTT multiplier to set the HTT link to 800 Mhz (that's down from 2000 Mhz):

Here is the data for both runs:

Phenom 9850 2.5 GHz 2.0 GHz NB 2.0 Ghz HTT and DDR2-1066
http://www.xcpus.com/gallery/d/6553-...80x1024_r1.JPG

Phenom 9850 2.5 GHz 2.0 GHz NB 800 MHz HTT and DDR2-800
http://www.xcpus.com/gallery/d/6556-...80x1024_r1.JPG

The data at 10280x1024 in the article is 2.5 GHz Core 2.0 GHz NB 2.0 GHz HTT and DDR2-800

Neither snow or cave show much of a huge difference, Snow gets about 8 FPS (EDIT at low 'CPU limited regime' resoluionts) and generally better with DDR2-1066 but the HTT at 800 has no effect.

jack

---

i have noticed that low ht and low nb doesnt make much difefrence either.
is it that the ondie ht,nb is such low latency that it is still just as fast as the memory and still has a gerater or equal bandwidth. not sure. but i did find it interesting.

in Tonys 1000mhz thread he mentioned this once. that he noticed the cpu take off and get better clocks with lower ht and nb. less heat probly in tony's case. but i dont know im just a dumb newb :)

Quote:

---

Originally Posted by SkullCracka

i have noticed that low ht and low nb doesnt make much difefrence either.
is it that the ondie ht,nb is such low latency that it is still just as fast as the memory and still has a gerater or equal bandwidth. not sure. but i did find it interesting.

in Tonys 1000mhz thread he mentioned this once. that he noticed the cpu take off and get better clocks with lower ht and nb. less heat probly in tony's case. but i dont know im just a dumb newb :)

---

Yeah, in terms of an OC I can understand this... but from a pure computational flow of 1' and 0', it is hard to figure out why Phenom supports higher FPS in a GPU bound mode... it is interesting nonetheless.

Of all the contemporary games out there Lost Planet is one of the 'neater' ones to fool around with from a bench perspective. Between LP and UT3, they do the best job of utilizing quads.

EDIT: PS -- I have only ever met one person in the enthusiast forums I would catagorize as 'dumb', you are not that person :)
Jack

Ok, it's either the phenom or the 790FX.


The snow benchmarks at resolutions around 1600*1200 are mostly gpu limited right?
So this means it shouldn't matter much, going from a phenom 2.5 Ghz, to a dual core K8 at say 3.2 Ghz.
If the AMD system then still benches higher than the intel quad core system, it has to be a chipset > GPU connection thing, giving the AMD setup the advantage.

If this test scores lower than the intel system, it has to be a phenom architecture thing.

Quote:

---

Originally Posted by Jakko

Ok, it's either the phenom or the 790FX.


The snow benchmarks at resolutions around 1600*1200 are mostly gpu limited right?
So this means it shouldn't matter much, going from a phenom 2.5 Ghz, to a dual core K8 at say 3.2 Ghz.
If the AMD system then still benches higher than the intel quad core system, it has to be a chipset > GPU connection thing, giving the AMD setup the advantage.

If this test scores lower than the intel system, it has to be a phenom architecture thing.

---

That is reasonable....

Alright, let's see what we've got here:

System Specs:
CPU: Intel Core 2 Extreme QX9650 ( always at 333MHz FSB ) [ QDR1333 ]
Mobo: Asus Striker II Extreme 790i Ultra SLI [ BIOS 0704 ]
RAM: CellShock DDR3-1800 2GB Kit
VGA: Single GeForce 9800GTX 512MB Ref. Card ( Normal Stock Clocks )

Normal RAM Timings: A-A-8-7-6-21-1T-7-35-10-25-35-9-7-74
Tweaked RAM Timings: E-E-7-7-6-21-1T-1-1-5-21-1-1-1-74

Game Settings:
Set 1 = JumpingJack's settings + Default Drivers Settings ( matched with JJs ) @ 1280x960
Set 2 = Real-life gaming settings, Full Details ( All settings maxxed along with 4xAA 8xAF and Transp. Anti-Aliasing ) + Modified Drivers Settings ( see attached pics ) @ 1280x960

CPU Frequency Tests - JJ's Settings - CPU Multiplier adjustment method
http://i28.tinypic.com/el65va.jpg
Complimentary VGA Overclock Chart
http://i29.tinypic.com/331kcht.jpg
As you can see the Snow test is GPU Limited at these settings.
The Cave test is totally CPU Limited.

Memory Performance Chart - "Normal RAM" vs "Tweaked RAM" - JJ's Settings
http://i31.tinypic.com/rhmm8y.jpg
As expected, the results in the first test ( Snow ) remain totally unchanged since it's totally bound to our GPU's performance, while the performance in the more CPU/System Limited Cave test increased.

Real-Life Gaming, Full Game Details, Drivers set for max image quality, 1280x960 4xAA 8xAF Transp. AA On
CPU Frequency Tests - Full Game Details, Drivers Max IQ, 1280x960 4xAA 8xAF Transp AA On - CPU Multiplier adjustment method
http://i25.tinypic.com/4h8som.jpg

Memory Performance Chart - "Normal RAM" vs "Tweaked RAM" - Full Detail Settings, Drivers IQ, 1280x960 4xAA 8xAF Transp AA Settings
http://i25.tinypic.com/e04ykw.jpg

The game is purely GPU Limited with these settings like I told ya ;)

Quote:

---

Originally Posted by BenchZowner

The game is purely GPU Limited with these settings like I told ya ;)

---

Nice data... as it should be... the GPU is the determinant when playing at 'contemporary' resolutions with the eye candy up.

DX10 data is nice to see for this to compare.

How about sticking a lowly X2 into the 790i to see if this anomaly still persists?

Maybe you could redo this test but with the Intel CPU at 1600 FSB.

Quote:

---

Originally Posted by IvanAndreevich

How about sticking a lowly X2 into the 790i to see if this anomaly still persists?

---

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

Quote:

---

Originally Posted by thebluemeanie1

Maybe you could redo this test but with the Intel CPU at 1600 FSB.

---

Harder to do... would require a complete re-vamp of the test. at 400 Mhz, 2.4 or 2.6 is the closes I could get to 2.5... but, what I could do (and did) was go to 200 Mhz on the system clock -- use a multi of 12.5 and try again... within a few FPS, no change ... Snow at 1280x1024 came to 106, Cave was about 87 ... so overall that is not a huge modulator.

Quote:

---

Originally Posted by BenchZowner

The game is purely GPU Limited with these settings like I told ya ;)

---

At both JJ's settings, and your settings, the snow level is gpu limited.
I don't think your settings are "real life settings" though, anyone would turn up the eye candy, but not at the cost of having smooth gameplay.

It is weird how the phenom system performs better in some gpu limited scenarios.
This means that the same issue could arise on a quad crossfire system that uses settings closer to your high quality settings.

Quote:

---

Originally Posted by BenchZowner

The game is purely GPU Limited with these settings like I told ya ;)

---

Excellent work!

But most of the games nowadays is GPU limited. Enable all settings and playing with high-resolutions, CPU power becames useless. Maybe bandwith or latency (HTT or FSB) does make small diference.

In this scenario GPU limited, having an Intel or AMD CPU doesn't matter in game performance.

Quote:

---

Originally Posted by Jakko

I don't think your settings are "real life settings" though, anyone would turn up the eye candy, but not at the cost of having smooth gameplay.

---

1) Most people don't like playing games without AA & AF at all ( especially in resolutions lower than 1920x1440 ).
2) I buy high end graphics cards to play games, so yes, I like max details @ game, and highest AA/AF possible ( adjusting depending on performance ).
3) Coming in a bit...another chart.

Quote:

---

Originally Posted by Jakko

It is weird how the phenom system performs better in some gpu limited scenarios.

---

Since it's only 1 game, and 1 test, and JumpingJack himself is puzzled and can't say something about it, and he is unsure, as I am as well, it's not right to say something especially considering it as a fact.

Quote:

---

Originally Posted by BenchZowner

1) Most people don't like playing games without AA & AF at all ( especially in resolutions lower than 1920x1440 ).
2) I buy high end graphics cards to play games, so yes, I like max details @ game, and highest AA/AF possible ( adjusting depending on performance ).
3) Coming in a bit...another chart.

---

True, but most people also don't like to play anything with lower fps than 30.

Quote:

---

Since it's only 1 game, and 1 test, and JumpingJack himself is puzzled and can't say something about it, and he is unsure, as I am as well, it's not right to say something especially considering it as a fact.

---

Phenom performing better in some gpu limited cases, is a fact. What did I say that wasn't right?
Also it wasn't one test, it was several tests at several resolutions.

Quote:

---

Originally Posted by Jakko

True, but most people also don't like to play anything with lower fps than 30.

---

I've got SLI for this game ( if you're talking about the low FPS @ my charts ).
I also have better single cards.
Even if I couldn't make it run smooth with any of those ( even if one of those cards that's faster than any current dual-GPU card or SLI/CF combination ) I could easily drop the AA to 2x and game fine ;)

I wouldn't drop my details however.
Would you play Crysis @ Low ?
It looks like FarCry...maybe even worse :D

Quote:

---

Originally Posted by Jakko

Phenom performing better in some gpu limited cases, is a fact. What did I say that wasn't right?
Also it wasn't one test, it was several tests at several resolutions.

---

Where ?
And please do not tell me about low to medium game settings without AA/AF, because most people do not play games at low, and surely not me.

Hold on now please, more charts coming soon.

Here's my last piece for now.
Let's see how our FSB, memory, and cores count affect our gameplay quality with "gaming" settings ( Full Details, + AA/AF + Transp. AA )

http://i26.tinypic.com/2vvua1d.jpg
As you can clearly see, Lost Planet uses about 2 cores.
Having a single-core CPU will result in lower FPS in some occasions ( not all the time ).
Lower FSB or higher FSB doesn't make a difference ( add that in the lower FSB tests here we also had the memory running at a lower frequency than in the higher FSB tests ) [ 200MHz FSB & RAM @ DDR3-800 8-7-6-21 vs 333MHz FSB & RAM @ DDR3-1333 8-7-6-21 ]
Our memory's performance also does pretty much nothing in real-life gaming settings.
Same settings DDR3-800 8-7-6-21 relaxed sub-timings vs DDR3-2000 7-7-6-21 fully tweaked = exact same performance.

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...

Quote:

---

Originally Posted by BenchZowner

And please do not tell me about low to medium game settings without AA/AF, because most people do not play games at low, and surely not me.

---

I do, because my Intel GMA900 is too weak to play at anything better than lowest settings. At least in more recent games like HL2. I mostly enjoy my oldies at max res/settings, but's because they are REALLY old.

Maybe next year I'll be doing some gaming on my main rig again instead on my weak but very cool laptop.

Quote:

---

Originally Posted by Devil's Prophet

I do, because my Intel GMA900 is too weak to play at anything better than lowest settings. At least in more recent games like HL2. I mostly enjoy my oldies at max res/settings, but's because they are REALLY old.

Maybe next year I'll be doing some gaming on my main rig again instead on my weak but very cool laptop.

---

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

Quote:

---

Originally Posted by BenchZowner

Here's my last piece for now.
Let's see how our FSB, memory, and cores count affect our gameplay quality with "gaming" settings ( Full Details, + AA/AF + Transp. AA )

As you can clearly see, Lost Planet uses about 2 cores.
Having a single-core CPU will result in lower FPS in some occasions ( not all the time ).
Lower FSB or higher FSB doesn't make a difference ( add that in the lower FSB tests here we also had the memory running at a lower frequency than in the higher FSB tests ) [ 200MHz FSB & RAM @ DDR3-800 8-7-6-21 vs 333MHz FSB & RAM @ DDR3-1333 8-7-6-21 ]
Our memory's performance also does pretty much nothing in real-life gaming settings.
Same settings DDR3-800 8-7-6-21 relaxed sub-timings vs DDR3-2000 7-7-6-21 fully tweaked = exact same performance.

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...

---

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.

Quote:

---

Originally Posted by BenchZowner

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....

Quote:

---

Originally Posted by JohnMike

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.

---

Quote:

---

Originally Posted by jas420221

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

---

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.

Quote:

---

Originally Posted by JohnMike

As i said before....

---

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

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................

Quote:

---

Originally Posted by Daveburt714

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

---

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

Quote:

---

Originally Posted by JumpingJack

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 :D

Quote:

---

Originally Posted by 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................

---

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".

Quote:

---

Originally Posted by Devil's Prophet

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. :D

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).

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:
http://i32.tinypic.com/2s5ze35.jpg

Quote:

---

Originally Posted by 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:
http://i32.tinypic.com/2s5ze35.jpg

---

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

jack

now it 's interesting.:coffee:

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:

http://www.xcpus.com/gallery/d/6597-1/CoreScaling1.jpg

Plotted up a slightly different way:
http://www.xcpus.com/gallery/d/6600-1/CoreScaling2.jpg

Finally, a table showing scaling factors normalized to a single threaded run for each seen with respect to each CPU:
http://www.xcpus.com/gallery/d/6603-1/CoreScaling3.jpg

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

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 :)

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.

interesting to see some more data; thx for that.

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.

Quote:

---

Originally Posted by 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.

---

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

Quote:

---

Originally Posted by BenchZowner

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?

Quote:

---

Originally Posted by Calmatory

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.

Quote:

---

Originally Posted by 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.

---

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.)

Quote:

---

Originally Posted by BenchZowner

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.

Quote:

---

Originally Posted by Calmatory

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.

Quote:

---

Originally Posted by 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.

---

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

Quote:

---

Originally Posted by 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 :)

---

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.

Quote:

---

Originally Posted by Jakko

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 :ROTF:

Well said bro:toast:


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

So BenchZowner, you've come here all drenched in gasoline to piss in AMD territory. I guess you asked to be flamed.

First of all, very impressive confidence you have!
Althought you knew you werent welcome here you still thought Jakko was trying to impress you, as if you thought people were looking up to you.
While Jakko nicely told you that he didnt want an other Intel vs. AMD flame wars, you imedietly start to insult him and flame AMD. Very good start for a nice topic.

You state several times that you dont care yet still you are trying to convince us that we are wrong and you are right.
However thats where you lost me. You claim that it doesnt matter since the game is GPU limited but its wrong to say AMD is just as good as Intel.
As if the 'GPU limited' argument is only valid when AMD got the upper hand and thus making the 'win' nullified. While for Intel it counts as a 'win'.

All your posts are 98% off topic because this topic was about WHY PHENOM WAS IN A LEAD AT SOME POINT IN THIS BENCHMARK!
Not whatever the benchmark is GPU limited or not, the Phenom is still in the lead at some point in the bench and so why is that?

With the results of your actions meny people propably have come to the conclusion that you are simply a less intelligent troll, hence why they call you stupid.
Althought some of your post where not insulting or flaming and showed understanding of a nice discussion. You still seem to have a long way to go.

Quote:

---

Originally Posted by BenchZowner

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?"

---

He didn't say that. He said that in any given competition, even if one contender seemingly wins all the time, there will still be occasions where the other contender will triumph given the right circumstances.

He didn't say one was better than the other, he didn't even try to predict how one contender will do. He simply attempted to answer the question posed in this thread: Why is this contender beating the other contender in this paticular area? The answer being, of course, that given the right circumstances any contender will win.

When coming into a discussion, please try to remain neutral and leave your own personal world view at the door. Otherwise it'll twist what you read and then make people think you're stupid because you seemingly can't read.

Quote:

---

Originally Posted by JumpingJack

...

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

---

Scaling doesn't matter.

This is a must read : :ROTF:

http://www.realworldtech.com/forums/...79593&roomid=2
http://www.realworldtech.com/forums/...79593&roomid=2
http://www.realworldtech.com/forums/...79593&roomid=2

and the masterpiece :
http://www.realworldtech.com/forums/...79593&roomid=2

Don't know if you've came across this thread ; but it's worth reading . :yepp:

Hi, i want ONE amd system and i am thinking of a Foxcon motherboard,
any help which is he best Foxcon MB and which is hte best chip.

regards
steve

Quote:

---

Originally Posted by savantu

Scaling doesn't matter.

This is a must read : :ROTF:

---

Scaling does matter I think.
A quote from the article you posted:

Quote:

---

It's not that Intel "scales badly", it's that Intel does
so much better on single-core benchmarks thanks to
the bigger usable cache.

---

What this person forgets is that it doesn't really matter why something scales or doesn't scale.
Yes intel uses a cache intensive system that improves single core performance dramaticly, and doesn't do that much for adding more cores, but really what does this have to do with anything?

The final result is still good or bad core scaling.

Quote:

---

Next time I hear somebody talk about "bad scaling", when
it's just that the performance is higher to start with,
I'll find a chainsaw and do unspeakable acts to the
idiot. Scaling is meaningless.

---

Next time I hear somebody talk about AMD's poor single core performance, when it's just that the scaling is better in the end, I'll find a chainsaw and do unspeakable acts to the idiot. Single core performance is meaningless

Sounds ridiculous eh?
And it is.

Quote:

---

Originally Posted by savantu

Scaling doesn't matter.

This is a must read : :ROTF:

http://www.realworldtech.com/forums/...79593&roomid=2
http://www.realworldtech.com/forums/...79593&roomid=2
http://www.realworldtech.com/forums/...79593&roomid=2

and the masterpiece :
http://www.realworldtech.com/forums/...79593&roomid=2

Don't know if you've came across this thread ; but it's worth reading . :yepp:

---

Scaling is relative in the context that I state. The Phenom scales over the threads better, this is what the data says... no arguing that.

However, if single threaded performance has a 20% lead, and the Phenom only scales 5% better utilizing 4 ... the overall absolute performance is still overwhelmingly lopsided.

The only thing the scaling for each shows is that the native approach yields better scaling... for absolute performance, it is not near enough to catch up.

Whether it is important or not -- it is only important to me as a learning experience to help understand how CPUs work, that is the short of it. Both AMD and Intel are bound by more constraints than 'let's make the faster CPU'... the factors going into design decisions include -- time to market, die size, costs, yield, competitive forces, etc. etc. While AMD is techologically more 'elegant' (I will not argue against that), it is Intel that has the technologically important characteristic -- good raw absolute performance, quad core 1 to 1.5 years early, etc. etc. and the effects are (unfortunately) showing up in AMD's bottom line.

Considering this is an AMD section though, this is not about -- Intel trouncing AMD ... not at all, the Phenom is a fine CPU, good performance, and AMD has priced it appropriately such that it is a good purchase for anyone who want to purchase AMD for the sake of purchasing AMD -- or for what ever reasons -- I would not criticize someone for choosing/taking this processor for example.

Jack

Oh and thanks Jack for wanting to do that test in the future.
I do have a 790X board with a brisbane 5000+ BE, and similar ram to what you used, but sadly I only have a 3850 at the moment, probably not enough to make the snow level gpu limited in the way the gpu was the limiting factor in your benchmarks.

What I find strangest is that

1. at very low resolutions (completely cpu limited) intel wins
2. at a middle ground (high resolution, standard details, gpu limited) amd wins
3. at the very highest details and resolutions, it's a tie.

I could understand how, when the gpu has to work very hard(#3), bandwidth requirements between the gpu and cpu are lower than a scenario in which the gpu can pump out much more FPS at a fairly high resolution, due to limited detail level.(#2)

But it stops making sense when intel wins at those low resolutions. Isn't that the situation in which the bandwidth/architecture advantage seen at #2 should show as well?

Quote:

---

Originally Posted by Jakko

Oh and thanks Jack for wanting to do that test in the future.
I do have a 790X board with a brisbane 5000+ BE, and similar ram to what you used, but sadly I only have a 3850 at the moment, probably not enough to make the snow level gpu limited in the way the gpu was the limiting factor in your benchmarks.

What I find strangest is that

1. at very low resolutions (completely cpu limited) intel wins
2. at a middle ground (high resolution, standard details, gpu limited) amd wins
3. at the very highest details and resolutions, it's a tie.

I could understand how, when the gpu has to work very hard(#3), bandwidth requirements between the gpu and cpu are lower than a scenario in which the gpu can pump out much more FPS at a fairly high resolution, due to limited detail level.(#2)

But it stops making sense when intel wins at those low resolutions. Isn't that the situation in which the bandwidth/architecture advantage seen at #2 should show as well?

---

Well, it is hard to draw generalized conclusions ... this is one game, with two different scripted scenes ... one thing to take away from this data set is that a gaming experience is really dictated by a complex set of factors (I am diverging here from a CPU/architectural discussion true).

In any given benchmark, at any given set of settings the GPU may be important, the CPU may be important ... overall, it is the GPU that determines the experience for the most part -- choosing the CPU should be done carefully though, either CPU in today's gaming environment will handle any game (only possible exception is Crysis). It is hard to declare any CPU a 'winner' with such a very narrow subset of conditions/scenes/rendering conditions.... the best way (impractical way) would be to determine through the entire story line of a game, how often either CPU limits the gaming out put to below the playable FPS.... this is simply not possible, i.e. you can never guard against the 'worst case condition' as none of the review sites nor any of the built in benchmarks can possibly account for worst case in actual game play.

Back on topic -- I think you are right, ultimately the BW argument will ultimately be the appropraite explanation, be it at the HT (analogous FSB) point, or a the PCIe point of connection.... if we were to generalize a statement, it would fall out very similar to what has been said time and again -- CPU limited work loads, Intel has a stronger core -- throughput limited workloads AMD's interconnect shows it's advantage.

jack

Quote:

---

Originally Posted by savantu

Scaling doesn't matter.

This is a must read : :ROTF:

http://www.realworldtech.com/forums/...79593&roomid=2
http://www.realworldtech.com/forums/...79593&roomid=2
http://www.realworldtech.com/forums/...79593&roomid=2

and the masterpiece :
http://www.realworldtech.com/forums/...79593&roomid=2

Don't know if you've came across this thread ; but it's worth reading . :yepp:

---

I dont own an amd system atm, but i always come to visit this forum to see how things are going. the posts you linked to are very enlightening to me (I respect LT's views), but looking at the setup, I don't think its a scaling advantage. with the test being completly GPU dependent, I would think that the performance advantage has to be attributed to the motherboard/ chipset in some way. it could be the amd system getting a boost somehow, or the intel system being hindered in some way.

perhaps running the intel system on a different chipset to see if there is a similar falloff at higher levels, and run the amd system on a different chipset to see if the performance curve is the same.

Quote:

---

Originally Posted by JumpingJack

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.

---

Mem: Ballistix 1066 @ tbd timings, maybe do some runs on both systems @ diff timigns

Mobo: Intel: GB EP35-DS3L , AMD: GB 790FX

Vid: 3850 512mb

OS: Vista, edit SP 1 + no additional updates, should be better
I'm installing clean OS on both systems, exactly the same. Defragging between prog installs.

Quote:

---

Originally Posted by savantu

Scaling doesn't matter.

This is a must read : :ROTF:

http://www.realworldtech.com/forums/...79593&roomid=2
http://www.realworldtech.com/forums/...79593&roomid=2
http://www.realworldtech.com/forums/...79593&roomid=2

and the masterpiece :
http://www.realworldtech.com/forums/...79593&roomid=2

Don't know if you've came across this thread ; but it's worth reading . :yepp:

---

I don't agree with his conclusions whatsoever, he's clearly talking in a server context (and even on a desktop I still don't agree). He's saying if you add more raw power to a system, it shouldn't scale linearly. That's not very cost-effective is it?

Additionally, he's saying that the CPUs should be matched to the infrastructure. By that logic, we should be back with 386s, since there's no chance a HD will be able to saturate a faster CPU. Even a 386 might be too fast.

I know he's getting at Intel being faster in single-threaded apps is a good thing, and that scaling beyond that, it's still faster than AMD. That's fine, and most definitely true. But trying to pretend that scaling isn't important with multi-CPU server/super-computer situations is a joke. Incidentally in servers Phenom beats Intel handedly, seems that scaling suddenly matters.

Quote:

---

Originally Posted by boogle

I don't agree with his conclusions whatsoever, he's clearly talking in a server context (and even on a desktop I still don't agree). He's saying if you add more raw power to a system, it shouldn't scale linearly. That's not very cost-effective is it?

---

More power as in more cores? More parallel power in situations that need serial power? Or what are you saying that isn't right?

Quote:

---

Originally Posted by boogle

Additionally, he's saying that the CPUs should be matched to the infrastructure. By that logic, we should be back with 386s, since there's no chance a HD will be able to saturate a faster CPU. Even a 386 might be too fast.

---

That's why they invented RAM and even RAMdisks. Now solidstate is also getting closer.

Quote:

---

Originally Posted by boogle

I know he's getting at Intel being faster in single-threaded apps is a good thing, and that scaling beyond that, it's still faster than AMD. That's fine, and most definitely true. But trying to pretend that scaling isn't important with multi-CPU server/super-computer situations is a joke. Incidentally in servers Phenom beats Intel handedly, seems that scaling suddenly matters.

---

Doesn't he mean the absolute numbers are more important than the scaling numbers on their own? Just like you say in the above of this quotation.

Anyway, are you sure you're not pulling his words out of context in a charade of sarcasm? Because in my experience that happens A LOT in here.

Quote:

---

Originally Posted by Devil's Prophet

More power as in more cores? More parallel power in situations that need serial power? Or what are you saying that isn't right?

Doesn't he mean the absolute numbers are more important than the scaling numbers on their own? Just like you say in the above of this quotation.

Anyway, are you sure you're not pulling his words out of context in a charade of sarcasm? Because in my experience that happens A LOT in here.

---

From my understanding he's saying that you should have loads of performance with whatever the software was initially designed for, and the more nodes you add the less of an improvement you should get (diminishing returns).

Being a software developer he should know rule #1 with server apps: The software should scale with an infinite number of CPUs.

I agree with his comments about absolute power, especially with Intel favouring single-threaded apps - this is a good plan from Intel. However, when he starts talking about scaling shouldn't scale evenly, he's lost me. Completely. If you don't scale evenly, there's a bottleneck somewhere and that bottleneck needs to be resolved. Its NOT a good thing!

Not sure if it applies to this thread, but I'm going to be getting the parts in on wednesday to build a Q9450/X48 system to compare it and the 9850 at crossfire and crossfire alone.

Intel system is...
Q9450 cooled by TRUE with Scythe Slipstream 1900rpm 110cfm
Asus Rampage Formula
Kingston 2x1gb DDR2-1066 5-5-5-15 1.8v
2x Sapphire 3870's in crossfire
Seagate Barracuda 250gb
Windows Vista Ultimate x86

AMD system is...
9850 BE (my week 811) cooled by TRUE with same fan
Asus M3A32-MVP Deluxe Wifi, bios 1101
Kingston 2x1gb DDR2-1066 5-5-5-15 1.8v
2x Sapphire 3870's in crossfire
Western Digital 250gb with perpendicular recording
Windows Vista Ultimate x86

Only differences will be the case, hd, and cd rom other than the necessary cpu/mobo in order to minimize variables. I'll just do tests at 2.66 and 3ghz because my board doesn't let me take the 9850 higher than 3ghz stable even though it does it at 1.325v.

AMD performs better on high res
Intel Q9450 vs Phenom 9850

http://www.overclockersclub.com/revi...l_q9450/14.htm
http://www.overclockersclub.com/revi...l_q9450/13.htm
http://www.overclockersclub.com/revi...el_q9450/8.htm

AMD scales better using more cores
Processor performance list

as far as games go cpus really don't have that great of an effect on fps. the difference might be by just a few frames. mostly the difference is something that the computer can't tell you. its something you will have to see for yourself. now you can get an intel cpu clocked 4ghz and get a phenom clocked at 3ghz. the intel will have a higher fps but does it mean its better? if you actually look at the game instead of the fps you will notice that the amd system runs a lot smoother even tho the fps is lower. benchmarks and results can't be the only proof of why something is better. in this case the benchmark favors one but you can look at it and clearly tell the other is better.

Quote:

---

Originally Posted by roofsniper

now you can get an intel cpu clocked 4ghz and get a phenom clocked at 3ghz. the intel will have a higher fps but does it mean its better? if you actually look at the game instead of the fps you will notice that the amd system runs a lot smoother even tho the fps is lower.

---

Is fps not what makes a game smooth? I can see the argument in saying windows is smoother etc but in a game does fps not define smoothness?

Quote:

---

Originally Posted by gallag

Is fps not what makes a game smooth? I can see the argument in saying windows is smoother etc but in a game does fps not define smoothness?

---

Remember microstuttering with SLI/Crossfire setups? Maybe this is something equivalent.

Holy thread-digging batman :D

As for the "lower FPS but smoother"... please, don't kill me ( that's definitely funnier than this phrase ).

When you have:

Lower minimum FPS
Lower average FPS
Lower maximum FPS

And probably bigger drops from average FPS to minimum FPS...how can you feel that the game runs smoother than the other setup with higher min FPS, avg FPS and max FPS and less fluctuation ?

Stop beating on a dead horse

and... like we said... in GPU Limited scenarios ( most of the games [ if not all ] when run under "normal gaming" settings ) the processor plays little to none ( none most of the time ) [ if you're running anything better than a A64 3500+ / P4 3.4GHz ] role.

Quote:

---

Originally Posted by gallag

Is fps not what makes a game smooth? I can see the argument in saying windows is smoother etc but in a game does fps not define smoothness?

---

How smooth the game is depends on how evenly frames are displayed. If you have 100 FPS but 90 of those is displayed on the first half of the second, 10 the second half. You have another computer that has 30 FPS but those are evenly displayed. Then the 30 FPS computer will be smoother than the 100 FPS computer.

Intel has one schizophrenic behavior. It is very fast when data is in the cache, it has high frequencies. But when it needs to communicate with other hardware it isn’t as fast, the latency goes up. On AMD the processor isn’t as fast but the processor is built to scale. Cache, hypertransport, IMC etc. Areas that will enable AMD to scale evenly.
You could say that AMD has worked on their weak parts. AMD don’t have any bottlenecks. Intel has worked on their strong parts but the bottlenecks still exists.

Quote:

---

Originally Posted by gosh

How smooth the game is depends on how evenly frames are displayed. If you have 100 FPS but 90 of those is displayed on the first half of the second, 10 the second half. You have another computer that has 30 FPS but those are evenly displayed. Then the 30 FPS computer will be smoother than the 100 FPS computer.

Intel has one schizophrenic behavior. It is very fast when data is in the cache, it has high frequencies. But when it needs to communicate with other hardware it isn’t as fast, the latency goes up. On AMD the processor isn’t as fast but the processor is built to scale. Cache, hypertransport, IMC etc. Areas that will enable AMD to scale evenly.
You could say that AMD has worked on their weak parts. AMD don’t have any bottlenecks. Intel has worked on their strong parts but the bottlenecks still exists.

---

I agree, If a game was running at 90fps then 10fps it would be very bad, But Intel cpus very ofter have a higher min fps so how could it seem less smooth?

Quote:

---

Originally Posted by gallag

I agree, If a game was running at 90fps then 10fps it would be very bad, But Intel cpus very ofter have a higher min fps so how could it seem less smooth?

---

I don’t think that you could see this effect on games that isn’t very new (are using threads and is heavy on the graphics). There might be situations on C2Q and if the game has more than one thread. If one thread is moved to another core (on another C2D, C2Q = 2 x C2D) and cache is invalidated then performance will go down because data used would need to be fetched from memory again. If the other thread is sending and retrieving data also then this will be slower, Intel is using the FSB for memory and I/O, more traffic and there is greaer risk for conflicts.

There could be situations where you have other programs running on the computer, these programs doesn’t use that much power (then are mostly resting) but sometimes they might and that will hurt Intel more.

I talk to a few tier 1 reviewers who do a ton of testing, one reported right after Phenom was launched games average frame rates were slower but the games ran smoother, now I don't take that much notice while testing so i just ask more reviewers to see what they feel....and they all say the same.

Now I thinks it not only CPU, I think its total platform and AMD have been working really hard on this for the past 18months. 4XXX GPU's have suddenly come good, ATI's driver dept look to be getting a move on to take the cards to new levels. I actually feel the driver writers were being held back by the 3XXX series cards and are now starting to get it together as they have more to work with.

So, AMD are just building a platform that plays games well, they dropped the need to compete in super pi etc and just concentrate where they think end users feel the system should perform.
I think it will get even better with the new GX chipset, all we need is someone to test it for us vs the older FX chipsets.

Quote:

---

Originally Posted by gallag

Is fps not what makes a game smooth? I can see the argument in saying windows is smoother etc but in a game does fps not define smoothness?

---

there are many things besides fps that make a game smooth. everyone here is familiar with tearing right? where you can even have TOO many fps and the game will not be smooth. just like gosh said its more because of the features amd has. no they may not have higher clocks and be running slower but they get it coming out at a steady pace and make it smooth. think of it as a fast reckless driver compared to someone just driving along. reckless driver will most likely be braking and accelerating over and over and changing lanes while the normal driver will be just going along. the jump from high fps to lower fps will make the games not as smooth. even if its a jump from 50-70 where fps is still high. so even tho intel pumps out more frames amd has them coming out at the same rate the entire time which is all a cpu should be doing anyway :up:

I think there are Valid points on both sides,

one sais single otimized scaling, one sais 4 core scaling beyond your dreams,

let me ask you this, and i beg you to answer honestly Intel urinal stalker/AMD trouser sniffer,





how are they with teh pron? :ROTF: