Core 2 Extreme: 10FPS for $800 more

[BaronMatrix]

I know this is going to sound like an anti-Intel post but I don't like the difference in 4400+ and FX62 either as compared to price.

Anyway, the subject of the post:

I was over on Anand (heaven forbid) and they posted an update tot he PhysX tests using Core 2 Extreme. It was a very interesting review in that it showed a marked increase in PhysX scores, but also because there was a section that showed the difference with and without the card using an underclocked Extreme.
The chip was clocked at 1.86GHz and scored 16FPS(min) at max settings (the game is considered CPU bound) while the 2.93GHz clock only scored 20FPS(min)

The average framerate for 1.86GHz was 30FPS and the 2.93GHz made an astounding 39.1FPS.

This means that for a CPU bound game and a 7950GX2 $800 for the EExtreme will get you 9.1 EXTRA FPS. I expected CPU bound games to show a much larger increase with an aded 1.07GHz.

I'm going to look around for the tests of 3800+ and FX62 to see the difference and hopefully it will be similar or it points to a lack of real scaling with Core 2.

 

[ElMoIsEviL]

I know this is going to sound like an anti-Intel post but I don't like the difference in 4400+ and FX62 either as compared to price.

Anyway, the subject of the post:

I was over on Anand (heaven forbid) and they posted an update tot he PhysX tests using Core 2 Extreme. It was a very interesting review in that it showed a marked increase in PhysX scores, but also because there was a section that showed the difference with and without the card using an underclocked Extreme.
The chip was clocked at 1.86GHz and scored 16FPS(min) at max settings (the game is considered CPU bound) while the 2.93GHz clock only scored 20FPS(min)

The average framerate for 1.86GHz was 30FPS and the 2.93GHz made an astounding 39.1FPS.

This means that for a CPU bound game and a 7950GX2 $800 for the EExtreme will get you 9.1 EXTRA FPS. I expected CPU bound games to show a much larger increase with an aded 1.07GHz.

I'm going to look around for the tests of 3800+ and FX62 to see the difference and hopefully it will be similar or it points to a lack of real scaling with Core 2.

Ummm not it doesn't... what game was it? With such low minimum frames per second it would indicate it's GPU bound rather then CPU bound.

But of course i'll wait for you to link us.

Oh BTW..

Whatever the case, after further testing, it appears our initial assumptions are proving more and more correct, at least with the current generation of PhysX games. There is a bottleneck in the system somewhere near and dear to the PPU. Whether this bottleneck is in the game code, the AGEIA driver, the PCI bus, or on the PhysX card itself, we just can't say at this point. The fact that a driver release did improve the framerates a little implies that at least some of the bottleneck is in the driver.

Taken from the article. There's a bottleneck wether on an FX an Opteron or a Core 2 Duo doesn't matter according to there results. It would seem likely that the PCI bus or the driver is the bottleneck.

 

[m0rk]

I know this is going to sound like an anti-Intel post but I don't like the difference in 4400+ and FX62 either as compared to price.

Anyway, the subject of the post:

I was over on Anand (heaven forbid) and they posted an update tot he PhysX tests using Core 2 Extreme. It was a very interesting review in that it showed a marked increase in PhysX scores, but also because there was a section that showed the difference with and without the card using an underclocked Extreme.
The chip was clocked at 1.86GHz and scored 16FPS(min) at max settings (the game is considered CPU bound) while the 2.93GHz clock only scored 20FPS(min)

The average framerate for 1.86GHz was 30FPS and the 2.93GHz made an astounding 39.1FPS.

This means that for a CPU bound game and a 7950GX2 $800 for the EExtreme will get you 9.1 EXTRA FPS. I expected CPU bound games to show a much larger increase with an aded 1.07GHz.

I'm going to look around for the tests of 3800+ and FX62 to see the difference and hopefully it will be similar or it points to a lack of real scaling with Core 2.

What teh hell are you rabbering about...

Since reducing the clockspeed of our CPU doesn't directly reduce the performance of the PhysX card, we find that our earlier theory is correct and that the PhysX card adds a hearty 50% with two cores functioning, and 73% with one core. Both results are greater than what we saw with a 2.93 GHz CPU. However, adding the second core also has a larger effect, this time at about 85% faster than a single core, although the dual-core setup is now only 6% faster than a single core with hardware PhysX.

An important note, that I hope you allready know. When playing games at these resolutions CPU has very little to do with the result.. Its all up to the graphiccard and in this case also the physiccard

 

[ElMoIsEviL]

Wow!!! we shut this one down pretty quick.

As Anandtech stated, it's using both cores when running in software mode but there's a bottleneck in the system (PCI Bus, driver, ?) that does not allow the Physics processing to function as it should. (in the case of software rendering it would point to a driver bottleneck seeing as the driver is stull utilised even when not using the Physix hardware).

Therefore it has nothing to do with the Core 2 Duo, Athlon64 FX or AMD Opteron not scalling properly.

 

[m0rk]

Wow!!! we shut this one down pretty quick.

As Anandtech stated, it's using both cores when running in software mode but there's a bottleneck in the system (PCI Bus, driver, ?) that does not allow the Physics processing to function as it should. (in the case of software rendering it would point to a driver bottleneck seeing as the driver is stull utilised even when not using the Physix hardware).

Therefore it has nothing to do with the Core 2 Duo, Athlon64 FX or AMD Opteron not scalling properly.

Agreed, seeing as the Aegia card is still traped on old PCI 32 is probly severaly bandwith restricted..

 

[npilier]

Wow!!! we shut this one down pretty quick.

As Anandtech stated, it's using both cores when running in software mode but there's a bottleneck in the system (PCI Bus, driver, ?) that does not allow the Physics processing to function as it should. (in the case of software rendering it would point to a driver bottleneck seeing as the driver is stull utilised even when not using the Physix hardware).

Therefore it has nothing to do with the Core 2 Duo, Athlon64 FX or AMD Opteron not scalling properly.

Yeap; I guess this thread is heading no where. No further comments.

 

[BaronMatrix]

I know this is going to sound like an anti-Intel post but I don't like the difference in 4400+ and FX62 either as compared to price.

Anyway, the subject of the post:

I was over on Anand (heaven forbid) and they posted an update tot he PhysX tests using Core 2 Extreme. It was a very interesting review in that it showed a marked increase in PhysX scores, but also because there was a section that showed the difference with and without the card using an underclocked Extreme.
The chip was clocked at 1.86GHz and scored 16FPS(min) at max settings (the game is considered CPU bound) while the 2.93GHz clock only scored 20FPS(min)

The average framerate for 1.86GHz was 30FPS and the 2.93GHz made an astounding 39.1FPS.

This means that for a CPU bound game and a 7950GX2 $800 for the EExtreme will get you 9.1 EXTRA FPS. I expected CPU bound games to show a much larger increase with an aded 1.07GHz.

I'm going to look around for the tests of 3800+ and FX62 to see the difference and hopefully it will be similar or it points to a lack of real scaling with Core 2.

Ummm not it doesn't... what game was it? With such low minimum frames per second it would indicate it's GPU bound rather then CPU bound.

But of course i'll wait for you to link us.

Oh BTW..

Whatever the case, after further testing, it appears our initial assumptions are proving more and more correct, at least with the current generation of PhysX games. There is a bottleneck in the system somewhere near and dear to the PPU. Whether this bottleneck is in the game code, the AGEIA driver, the PCI bus, or on the PhysX card itself, we just can't say at this point. The fact that a driver release did improve the framerates a little implies that at least some of the bottleneck is in the driver.

Taken from the article. There's a bottleneck wether on an FX an Opteron or a Core 2 Duo doesn't matter according to there results. It would seem likely that the PCI bus or the driver is the bottleneck.


it's the ifrst review on AnandTech, hence the mention of Anand.

 

[BaronMatrix]

For all. here is the link. It shows AS I SAID marked improvements in the PhysX system. Now there is always an INCREASE in frame rates. I hope they do the same thing with FX62.


Linkage!


Again, it's not a knock against Intel but a knock against ANY $1000 CPU. I really expected better scaling. They used a 7950GX2 @ 1600x1200. That will not be GPU bound with that card. CPU bound for SLI continues til around 1920 and then games become GPU bound (IIRC).

 

[cyborg_ninja-117]

For all. here is the link. It shows AS I SAID marked improvements in the PhysX system. Now there is always an INCREASE in frame rates. I hope they do the same thing with FX62.


Linkage!


Again, it's not a knock against Intel but a knock against ANY $1000 CPU. I really expected better scaling. They used a 7950GX2 @ 1600x1200. That will not be GPU bound with that card. CPU bound for SLI continues til around 1920 and then games become GPU bound (IIRC).

Fail

 

[gOJDO]

I know this is going to sound like an anti-Intel.

everything comming from you sounds anti-Intel.
You must feel very owned to open such thread. As Action_Man sugested to you, stay away from razor blades and everything will be fine for you and bad for us here.

 

[ElMoIsEviL]

For all. here is the link. It shows AS I SAID marked improvements in the PhysX system. Now there is always an INCREASE in frame rates. I hope they do the same thing with FX62.


Linkage!


Again, it's not a knock against Intel but a knock against ANY $1000 CPU. I really expected better scaling. They used a 7950GX2 @ 1600x1200. That will not be GPU bound with that card. CPU bound for SLI continues til around 1920 and then games become GPU bound (IIRC).

Whatever the case, after further testing, it appears our initial assumptions are proving more and more correct, at least with the current generation of PhysX games. There is a bottleneck in the system somewhere near and dear to the PPU. Whether this bottleneck is in the game code, the AGEIA driver, the PCI bus, or on the PhysX card itself, we just can't say at this point. The fact that a driver release did improve the framerates a little implies that at least some of the bottleneck is in the driver.

It's self explanatory. You first need to know how the Ageia software works. The game talkes to an Ageia driver which is installed once you install the game (regardless of wether or not you have a PPU installed). The driver then talks to the PPU or in the case of no PPU being found.. the CPU.

So it's quite simple. If the game is not GPU bound, is definatly not CPU bound... then of course.. there's something limiting performance. Anandtech is thinking that it may be the driver or the game code... as those are important steps in the process of executing Physics commands. It makes sense given the fact that a simply AGEIA driver update helped achieve a HUGE performance difference in another AGEIA title (GRAW).

 

[BaronMatrix]

For all. here is the link. It shows AS I SAID marked improvements in the PhysX system. Now there is always an INCREASE in frame rates. I hope they do the same thing with FX62.


Linkage!


Again, it's not a knock against Intel but a knock against ANY $1000 CPU. I really expected better scaling. They used a 7950GX2 @ 1600x1200. That will not be GPU bound with that card. CPU bound for SLI continues til around 1920 and then games become GPU bound (IIRC).

Whatever the case, after further testing, it appears our initial assumptions are proving more and more correct, at least with the current generation of PhysX games. There is a bottleneck in the system somewhere near and dear to the PPU. Whether this bottleneck is in the game code, the AGEIA driver, the PCI bus, or on the PhysX card itself, we just can't say at this point. The fact that a driver release did improve the framerates a little implies that at least some of the bottleneck is in the driver.

It's self explanatory. You first need to know how the Ageia software works. The game talkes to an Ageia driver which is installed once you install the game (regardless of wether or not you have a PPU installed). The driver then talks to the PPU or in the case of no PPU being found.. the CPU.

So it's quite simple. If the game is not GPU bound, is definatly not CPU bound... then of course.. there's something limiting performance. Anandtech is thinking that it may be the driver or the game code... as those are important steps in the process of executing Physics commands. It makes sense given the fact that a simply AGEIA driver update helped achieve a HUGE performance difference in another AGEIA title (GRAW).

Wow, you're amazing. If everything is the same except for clockspeed then it can't be the driver. Unles of course their driver is sensitive to clockspeed, but then that's a pretty crappy driver.

But since the card SHOWS INCREASES AT EVERY SPEED and every rating(max,min,avg) this shows what seems like a scaling problem. Not that Core 2 is slow but that it doesn't seem to scale with all things the same.

This test shows that if you buy the same system with a 6800 vs a 6300, you will get (at least in this game @1600 with a 7950GX2) a whopping 9 more FPS.

Again, I'd like to see the same test with an FX62 or 5000+.

 

[evilr00t]

Again, I'd like to see the same test with an FX62 or 5000+.

I bet you'll be in for a disappointment.

 

[MSantos]

I don't think it's a good idea to measure the increase in absolute terms. If that 9 fps increase was from 1000 to 1009 fps, I would agree with you and say it's a waste of money.
But going from 30 to 39 fps is a 30% increase, which is pretty good IMO (although still not worth 800$ to me).

 

[ElMoIsEviL]

For all. here is the link. It shows AS I SAID marked improvements in the PhysX system. Now there is always an INCREASE in frame rates. I hope they do the same thing with FX62.


Linkage!


Again, it's not a knock against Intel but a knock against ANY $1000 CPU. I really expected better scaling. They used a 7950GX2 @ 1600x1200. That will not be GPU bound with that card. CPU bound for SLI continues til around 1920 and then games become GPU bound (IIRC).

Whatever the case, after further testing, it appears our initial assumptions are proving more and more correct, at least with the current generation of PhysX games. There is a bottleneck in the system somewhere near and dear to the PPU. Whether this bottleneck is in the game code, the AGEIA driver, the PCI bus, or on the PhysX card itself, we just can't say at this point. The fact that a driver release did improve the framerates a little implies that at least some of the bottleneck is in the driver.

It's self explanatory. You first need to know how the Ageia software works. The game talkes to an Ageia driver which is installed once you install the game (regardless of wether or not you have a PPU installed). The driver then talks to the PPU or in the case of no PPU being found.. the CPU.

So it's quite simple. If the game is not GPU bound, is definatly not CPU bound... then of course.. there's something limiting performance. Anandtech is thinking that it may be the driver or the game code... as those are important steps in the process of executing Physics commands. It makes sense given the fact that a simply AGEIA driver update helped achieve a HUGE performance difference in another AGEIA title (GRAW).

Wow, you're amazing. If everything is the same except for clockspeed then it can't be the driver. Unles of course their driver is sensitive to clockspeed, but then that's a pretty crappy driver.

But since the card SHOWS INCREASES AT EVERY SPEED and every rating(max,min,avg) this shows what seems like a scaling problem. Not that Core 2 is slow but that it doesn't seem to scale with all things the same.

This test shows that if you buy the same system with a 6800 vs a 6300, you will get (at least in this game @1600 with a 7950GX2) a whopping 9 more FPS.

Again, I'd like to see the same test with an FX62 or 5000+.

I was unaware that you didn't know answers to such simple questions.

When you have a bottleneck (wether it's a CPU, Video card bottleneck) performance still scales, just not in the same way. Instead of giant leaps you get questionable results. All too often those results are within an error of margin of +/- 5FPS.

You have a game that is GPU bottlenecked. Your Core 2 Extreme x6800 will still post higher Frames Per Second then a Core 2 Duo E6300... just the performance different may only be 1-10FPS and nothing like the usual performance advantages you get without a bottleneck.

It's clear that there's a bottleneck somewhere. And given the fact that the bottleneck also occurs in software mode, then it must be the driver in this case. Especially considering the improvements seen once a new driver was released by AGEIA for GRAW. You're just looking to post FUD.. as you always do...

Just another un-educated simple minded fanboi :roll:

PS.. Performance per MHz is higher on the Core 2 processor then on the K8 processor. Therefore if you increase the clockspeed, you get an increase in performance higher then you would with the same increase in frequency on an AMD K8 based processor.
So as for scaling.. there certainly aren't any issues with Core 2 on that front.

 

[m0rk]

Elmo, elmo, elmo... stop feeding the troll. Its far better to starve them instead

 

[caamsa]

ElMoIsEviL did you read the article or is your reply post off the cuff bs :?:

 

[NightlySputnik]

Bla Bla Bla...

The thing is about having a well balance system. You should know ( I know you do) that you can't compare something that is limited by something elses. For exemple if you own a 19 inches LCD display with a resolution of 1280*1024 you're cpu will be more important than your VPU in most area. That just happen to be what a lot of poeple own right now.

At that resolution (not 2560*1600 or so they do test on most site), your cpu will make quite a big difference. Add to this video encoding and you can see why poeple will pay more for a fast cpu. If you don't do any of that crap... welll, go get an XBox360. You'll be well deserve for a better price if you own an HDTV. 8)

It's funny how I remember AMD fanboy posting about how the PentiumD was crap compare to A64X2 when it came to gaming, which was absolutely true by the way. It's only that now that AMD has lost the upper hand some of them seem be trying to prove that a cpu isn't that important for gaming. Funny how things change. :twisted:

Before I go! You should also consider that the 10 or so fps more represent an increase of +30% compare to the original result. Not quite bad I should say. :twisted:

 

[CannedTurkey]

City of Villains is a really odd gaming engine. I'd have to say that with the e6300 I'm using, it is no longer CPU bound. I routinely run the game on one monitor and have performance monitor on the second. CPU usage will generally hover around 80% on both cores. It may occasionaly have a very short lived (ie, split second) spike to 100%. The Sempron 2800+ I was using before would routinely sit pegged at 100%.

Also, GPU's seem to have very little impact on overall performance, with the exception that it generally runs better on NVidia cards compared to ATI. A lot of people with mediocre setups can achieve ~60fps, but even with the hottest hardware, and lowered settings, no-one (that I'm aware of) is achieving better than 70.

It is a good game, but from what I can see in the review, and my experience with the game, is that with that hardware, it isn't CPU or GPU bound. It's just a game that scales very poorly.

 

[ElMoIsEviL]

ElMoIsEviL did you read the article or is your reply post off the cuff bs :?:

I posted a quote from the Anandtech article which goes further then anything posted so far as to explain the results being seen.

Since you seem to be of the misleading/speading mis-information type who like to spew FUD I have quite convincingly posted how AGEIA's driver works (software or hardware mode) and how a simple driver update made a HUGE difference in performance in GRAW.

Do you happen to know how Ageia's driver works? If not then you haven't got a single right to even think you can argue any points with me.

The game's physics use AGEIA's proprietary Physics coding language. They need to go through a driver to be decoded when in software mode or if running in hardware mode (PPU) they go through the PPU.

How does the driver decode in software mode? Using the processor of course. But there seems to be bugs in AGEIA's driver and it seems to still be bottlenecking performance. We've seen this once GRAW was released it performed poorly, then AGEIA released a driver update which increased performance dramatically. In other words work can still be done to improve there driver (something that may not be in AGEIA's best interest when running in Software mode if you get my drift).

I firmly believe that AGEIA's driver needs more work. And that's what I posted above... and Anandtech seems to believe the same as they mentioned the driver as being a possible bottleneck for performance.

How is that BS?

BS is claiming the Core 2 doesn't scale properly.. now that's BS.

 

[caamsa]

ElMoIsEviL gosh you are such a clever young man? For you to insult others posts and belittle them tells me a lot about you. Also what are your qualifications regarding this article? Many clever little boys can read magazines and articles on the internet and post verbatim what they are reading about a certain subject. So no you really don't impress me. Why not post your resume and credit hours on the forumz and maybe I might consider you as a semi-know-it-all. How many articles have you written on the subject? EXAMPLE: I have a friend that knows almost everything about wine and his knowledge is quite impressive..........does that make him an expert.....no....does that mean I should believe everything he says about wine.......no....his knowledge is from reading wine magazines. People with opinions are a dime a dozen. And what you posted is your opinion about what you think is happening and not based on fact> you have run no tests or conducted any kind of experiment yourself. A normal respectable person would make his opinion in a profesional and polite manner. You on the other had are not very polite. Do you talk to people on the street like you post your responses? I doubt it.

 

[IcY18]

^ Got to be the worst post on TG Forumz yet

 

[ElMoIsEviL]

ElMoIsEviL gosh you are such a clever young man? For you to insult others posts and belittle them tells me a lot about you. Also what are your qualifications regarding this article? Many clever little boys can read magazines and articles on the internet and post verbatim what they are reading about a certain subject. So no you really don't impress me. Why not post your resume and credit hours on the forumz and maybe I might consider you as a semi-know-it-all. How many articles have you written on the subject? EXAMPLE: I have a friend that knows almost everything about wine and his knowledge is quite impressive..........does that make him an expert.....no....does that mean I should believe everything he says about wine.......no....his knowledge is from reading wine magazines. People with opinions are a dime a dozen. And what you posted is your opinion about what you think is happening and not based on fact> you have run no tests or conducted any kind of experiment yourself. A normal respectable person would make his opinion in a profesional and polite manner. You on the other had are not very polite. Do you talk to people on the street like you post your responses? I doubt it.

Umm no, I'm quite a respectable individual. But you fanbois with little to no knowledge are really running down my patience.
As for who I am well, I'm not going to post personal private information regarding myself (meaning a resume or things of the sort) but I will tell you that I've graduated University with a degree in IT engineering.. I've worked for several big companies (naming them may give people ammunition to accuse me of a Bias therefore it's best to leave that as is), I've run my own company as well and then when teh It industry was hard hit in Canada I went into financial advising. I now work for Scotiabank.


Here read about the PhysX driver. I know from first hand experience seeing as I play GRAW and have toyed with the driver in the HardOCP forums with my Asus PhysX card (worst investment i've ever made BTW).

Ageia PhysX


Ageia is a company which has produced the new PhysX Card, a plugin card similar to a graphics card, but dedicated to enhancing your PC's processing power so that it can provide advanced physics effects. Ghost Recon: Advanced Warfighter is one of the first games to provide support for the PhysX card, and to see a visual comparison of normal physics effects in the game versus those provided by PhysX, see this video comparison. As you can see, PhysX effects provide that extra bit of realism, however based on initial feedback from users such as this one, enabling PhysX effects may actually reduce your framerate noticeably. The reason is that despite the PhysX card handling the extra physics processing, the extra effects add more to your graphics card's load and hence reduce FPS. Hopefully this situation will improve.



When installing GRAW, you will notice it installs a PhysX driver even if your system doesn't have a PhysX card. It also adds a system tray item which reloads in the background each time you restart Windows. This is unnecessary in my opinion, and you can easily remove this driver by going to your Control Panel>Add/Remove Programs and selecting the 'Ageia PhysX' item and uninstalling it. I noticed no impact on my FPS whatsoever by removing the driver, however it does remove a background driver which is not needed.



Of course if you do have a system with a PhysX card, then make sure you download and install the latest Official PhysX Driver as this piece of hardware is still maturing and no doubt compatibility and performance issues are continually being improved by Ageia. GRAW will automatically detect your PhysX card and enable the additional physics effects in the game.

The Driver is necessary in performing the workaround to enabling full software physx in games. If you have a PPU it will detect it and use it.. if you don't well usually you cannot use the new features in the game... but there's a workaround which is what Anandtech did.

It seems that there's a bottleneck in the system. Because improving your processor makes VERY little difference. And the New driver for GRAW does help a bit but not much. So what i've come to believe is there's something having to do with the way the driver interacts with the game. Something in there is hindering performance. I don't believe it's the PCI bus seeing as the performance issues are also present when running in software mode.

Those are observations.. not opinions. Anandtech pretty much see's things the same way.

Whatever the case, after further testing, it appears our initial assumptions are proving more and more correct, at least with the current generation of PhysX games. There is a bottleneck in the system somewhere near and dear to the PPU. Whether this bottleneck is in the game code, the AGEIA driver, the PCI bus, or on the PhysX card itself, we just can't say at this point. The fact that a driver release did improve the framerates a little implies that at least some of the bottleneck is in the driver.

Even in software mode the games supporting AGEIA's physx don't perform well given the relative "simplicity" of there graphics engines (when compared to other more demanding titles).

Those are my observations from first hand experience with GRAW.. of course I cannot comment on City of Villains but it does appear from results that it's not a GPU bottleneck and not really a CPU bottleneck but something else hindering performance.

You see CPU scaling from an Opteron 144 to an FX-57 is only 10FPS when the frequency difference is 1.2Ghz. This does not make relative sense. To me, it seems there's a bottleneck elsewhere (this is using CPU software mode BTW). It's quite easy to prove that the CPU is not the bottleneck when you look at the same test with hardware physx enabled and again only a 10FPS difference can be seen between the Opteron 144 and FX-57.

Do you see what I mean by stating that it's not a scaling issue but rather some bottleneck issue somewhere else?

Makes sense to me.

 

[evilr00t]

Learn to use the RETURN key.

EXAMPLE: Right here.

 

[sithscout80]

Considering linear scaling in performance when increasing frequency is difficult. If you look at the numbers, it is actually looking pretty decent.

% Increase of Frequency for the processor

(2.93GHz - 1.86GHz)
-------------------------- *100 = 36.52 % increase of frequency
(2.93GHz)


(39.1fps - 30fps)
--------------------- *100 = 23.27 % increase of frames per second
(39.1fps)

This also makes the assumption that the minimal increase in performance is due to the processor when it could be the game isn't threaded perfectly to take full advantage of multiple processors.