• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

PPU's help 3Dmark 06?

Status
Not open for further replies.

colbyman

Limp Gawd
Joined
Jan 28, 2007
Messages
200
Does any one know if having a PPU boosts your 3D mark 06 score or not? Or is the difference marginally small?
 
Does any one know if having a PPU boosts your 3D mark 06 score or not? Or is the difference marginally small?

As far as I am aware, the PPU is totally unutilized in 3DM06. It shouldn't hurt you at all because the card will sit there dormant. But it won't help either. Maybe they'll start including support in the next 3DM release but who knows?
 
3dmark 06 doesnt bench or even use the ppu at all...
 
May I ask why? How is supporting/optimizing for the PPU any different from supporting/optimizing for multi-core CPUs?

Excellent comment here. Obviously this guy has something totally against the PPU. Actually if 3dmark07 has PPU support I can see the PPU becoming much more popular. There is literally thousands& thousands of people that "play" 3dmark if you know what I mean. Being #1 or top 10 is a milestone for many people online. What this would force is even the guys that are anti PPU would run out and buy one just to have the higher score. That would make the card a bit more used and in the end be good for agiea!
 
Yeah, as sad as it is, I can definitely see a couple of hundred people around here forking out $200 for an extra thousand 3DMarks...
 
Of course! More 3DMarks would be the biggest incentive for me to buy a PPU so far :D
 
3dmark 06 doesnt bench or even use the ppu at all...

After starting this thread i have thought about it and in the 3d tests there is actually quite a bit of physics involved. the sparks in the first test? and i just noticed that when the test is loading, it has the AGEIA logo on the bottom along with the Microsoft DX 9.0 and others. what other reason would the logo be on there if AGEIA had nothing to do with it?
 
oh i have one... im just to lazy to run 3dmard without it and see if there is a difference...:D
 
The PhysX engine (once called novodex before it was bought by Ageia) has been around for a while, long since before the PPU. Early versions of the engine (even after it was renamed PhysX) didn’t support the PPU. I believe the first version of the engine with PPU support was 2.3.0. The version of PhysX used in 3dmark06 is older, and does not support the PPU.
 
Woo! finally somebody that knows the facts! thank Gotnorice? love the name.

that make total sense. and ageia just put their name on there for the advertising.
 
Maybe if there was some 3dMark support ATI/Nvidia would actually get around to allowing us to use old video cards as PPU's like they've been talking about. And along with that, maybe we'd actually see games that have meaningful PPU support.
 
That would be enice. but im not to sure how well an old card would do for newer physics. i mean... the archtecture in PPU's is completely different then GPU's
 
unified shaders should work very sweetly. i reckon a 8600gt/s would work v.well.
 
May I ask why? How is supporting/optimizing for the PPU any different from supporting/optimizing for multi-core CPUs?

Excellent comment here. Obviously this guy has something totally against the PPU. Actually if 3dmark07 has PPU support I can see the PPU becoming much more popular. There is literally thousands& thousands of people that "play" 3dmark if you know what I mean. Being #1 or top 10 is a milestone for many people online. What this would force is even the guys that are anti PPU would run out and buy one just to have the higher score. That would make the card a bit more used and in the end be good for agiea!

Why do I see it from this point of view? There are a few reasons actually. I've got NOTHING against the PPU, In fact I think it's a wonderful (yet poorly executed) idea. But, 3dmark06 is a 3d performance benchmark. The PPU is not currently designated as a 3D accelerator... It's a physics accelerator. It adds realism to games, at the EXPENSE of 3D performance. It adds MORE stress to the GPU, while offloading some CPU calculations.

If Throwing in a 150$ card, will gain you 2000 points, yet said card is used in less than 20 games or so... 3Dmark becomes less and less of a benchmark, and more and more of an e-penis extension. You said it yourself! We'd all run out to grab a card that gets us a few more points... Adding a PPU category will literally ruin the ORB concept too... Anyone with Any system can score an extra few points, yet perform worse in game (with more physically correct elements) . This detracts from the worth of this benchmark IMO.

Most importantly IMO... There's only ONE PPU chipset... The point of a benchmark is to test the difference between DIFFERENT configurations. With one PPU, everyone that owns it would gain a set amount of points. If the card nets an extra 2000pts, then everyone with the card will gain 2000 points. Why doesn't Futuremark just add 5000 points to everyone's score? Adding random points, or adding points for owning a PPU are similar, as they probably don't help the game churn out any more FPS. These points don't really MEAN anything. Until the PPU can be overclocked, or newer revisions of it are released... Well, what's the point?

A CPU is a REQUIRED piece of equipment... You can't run without one. A GPU is also required. PPU's are NOT required to play games right now. They add Content. Until they are REQUIRED for 3D performance, I don't think a 3D benchmark should include them in the main score.

3Dmark's status as a bench is tentative enough as it is... Adding PPU support is, IMO, a major detriment from their already tarnished reputation. I do forsee some sort of physics acceleration incorporated into future revisions of 3Dmark, but when this happens, it will become useless to me. I'm not going to add a card that gives me a few points, yet adds no performance in-game. That's like PAYING for gold in WoW...
 
Why do I see it from this point of view? There are a few reasons actually. I've got NOTHING against the PPU, In fact I think it's a wonderful (yet poorly executed) idea. But, 3dmark06 is a 3d performance benchmark. The PPU is not currently designated as a 3D accelerator... It's a physics accelerator. It adds realism to games, at the EXPENSE of 3D performance. It adds MORE stress to the GPU, while offloading some CPU calculations.

When was the last time you ran 3dmark? Some of the benchmarks it runs are obviously tailored toward benchmarking other aspects of games besides 3d performance. Tech such as multi-core is utilized in the benchmark yet that does not see very widespread use in games and certainly didn’t when 3dmark06 actually came out.

Also your claim that a PPU adds physics at the expense of 3d performance, what are you basing that off of, GRAW? That game used the HavokFX engine for most of the physics with PhysX sprinkled on top as sort of a last minute after-thought. It is an incredibly poor implementation. For there to be a 3d performance hit when using a PPU, they would have to actually specify that there be additional debris, etc when a PhysX card is detected. Most games will likely have identical physics content with the only variable being whether or not that is calculated using the CPU or the PPU – hard to imagine that kind of situation causing any kind of 3d performance decrease.

If Throwing in a 150$ card, will gain you 2000 points, yet said card is used in less than 20 games or so... 3Dmark becomes less and less of a benchmark, and more and more of an e-penis extension. You said it yourself! We'd all run out to grab a card that gets us a few more points...

I don’t disagree with much of what you are saying but you seem to be touching more on the problems with benchmarks in general rather than anything having to do with PhysX or Ageia. Benchmarks have always favored certain hardware, supported certain features that not all cards had or that hardly any games used, etc. That’s why reviewers don’t just use one benchmark, and why it’s important to take everything into its proper context.

Most importantly IMO... There's only ONE PPU chipset... The point of a benchmark is to test the difference between DIFFERENT configurations. With one PPU, everyone that owns it would gain a set amount of points. If the card nets an extra 2000pts, then everyone with the card will gain 2000 points.

Really? I’d think you would get a certain amount of points based on, you know, how much it improved your performance in the benchmark :rolleyes:

Why doesn't Futuremark just add 5000 points to everyone's score? Adding random points, or adding points for owning a PPU are similar, as they probably don't help the game churn out any more FPS. These points don't really MEAN anything. Until the PPU can be overclocked, or newer revisions of it are released... Well, what's the point?

Why does there have to be comparable hardware to do comparisons? PPU vs. CPU isn’t a valid comparison? How about Single Core + PPU vs. Dual Core or even Single Core +PPU vs. Quad Core? Maybe even some day PPU vs. GPU Physics. These are all benchmarks I’d love to see. There likely isn’t going to be another PPU to directly compete against Ageia because so far every company has a drastically different opinion about how to do physics. That certainly does not mean that there aren’t comparisons to be made. Remember, PhysX isn’t just about the PPU, in software mode the engine is remarkable multi-threaded, and should be able to take advantage of modern processors quite well.

A CPU is a REQUIRED piece of equipment... You can't run without one. A GPU is also required. PPU's are NOT required to play games right now. They add Content. Until they are REQUIRED for 3D performance, I don't think a 3D benchmark should include them in the main score.

3Dmark's status as a bench is tentative enough as it is... Adding PPU support is, IMO, a major detriment from their already tarnished reputation. I do forsee some sort of physics acceleration incorporated into future revisions of 3Dmark, but when this happens, it will become useless to me. I'm not going to add a card that gives me a few points, yet adds no performance in-game. That's like PAYING for gold in WoW...

You keep saying that it doesn’t increase FPS as if it’s some kind of established fact. What are you basing that off of, Graw again? If you play a game like Cellfactor that actually makes extensive use of the engine, there is a definite performance increase with the PPU. I’m not even talking about cloth or liquids; stuff that requires REAL physics horsepower, but just while blowing up boxes and other mundane objects that are present even in software mode. When the shit hits the fan, it chugs for a sec whereas the PPU just tears through it.

Not all PhysX games are going to automatically add tons of random stuff that wasn’t there before just because you have a PPU, nor does the engine exist solely as a compliment to the PPU. PhysX has actually had the most success in the console market where PPU’s don’t even exist, as there are now quite a large number of titles out there that make extensive use of the engine. My point is that the PhysX engine is a stand alone Physics engine that is being used in many upcoming games, not because it supports the PPU, but simply because it’s a damn good Physics engine. The fact that you can toss in a relatively cheap card and improve your performance is just icing on the cake.
 
.If Throwing in a 150$ card said:
Well truth is 3dmark has ALWAYS been an "e-penis" extension ever since its inception back in 2000/2001. It has always been about the score. Do you realize that 90% of the scores in the orb data base are from stressed out systems that are OC'd well past stable and safe speeds? I bet most of those high scores you see are barely making it through the test. The point is everyone needs to realize that 3dmark is a game to a lot of gamers. Not a game in the traditional sense but a game like the quarter mile for racing a car. That is what it truly is all about. And having said that, anything that can give the gamers a few extra points is a plus. Running 3dmark is no more real world then looking at a review site who benched in 1600x1200 resolution with 8X FSAA and 16X AF playing a game I never played or care to play.

It is what it is and people especially review sites need to understand this instead of trying to make the benchmark worthless and not real world...[H] is really bad for this as they try to sya what is real world and what is canned. It is ALL a game so why not enjoy it?
 
Mal and Rice:

I definitely agree with you both in some things. Most of what you've stated I completely agree with, and are part of my views on 3Dmark. I just chose not to express or explain these points, as I like to keep my posts a bit shorter... Especially during exam time.

Personally, I just feel that adding in an OPTIONAL device, one that gives someone advantage, detracts more and more from the initial goals of the 3Dmark suite.

You can't deny that adding PPU support, right now, would drive many individuals to purchase one for the score extension... Both of you said so yourselves. I think that having the ability to arbitrarily increase your score detracts from the notion of a GOOD benchmark. Why don't they include a sound card in the score? It removes some burden from the CPU, just as the PPU does. You've got a Killer NIC? Great, add on some extra points for fps gained through reduced latency.

I have VERY little respect for the 3Dmark06 suite, I think it's a terrible indication of 3D performance. All I initially said was that this addition would be the last straw so to speak... The last thing required for me to completely disregard the benchmark. The final deviation away from strict 3D testing into a more general PC-wide benchmark. Another variable to confuse newer GPU tests, a variable that confounds the entire reason I ever considered 3Dmark in the first place.
 
Excellent comment here. Obviously this guy has something totally against the PPU. Actually if 3dmark07 has PPU support I can see the PPU becoming much more popular. There is literally thousands& thousands of people that "play" 3dmark if you know what I mean. Being #1 or top 10 is a milestone for many people online. What this would force is even the guys that are anti PPU would run out and buy one just to have the higher score. That would make the card a bit more used and in the end be good for agiea!

PPU's are a useless component that just takes up space and cut airflow to graphics cards. With Multi-Cores or GPU physics.. who needs an Ageia card?!

They're too complicated to code for and there's nothing they can do that a dedicated Core or GPU can't do for less $$$.

Wheter you like it or not.. Ageia's PPU will never make it. Their software might, but the hardware will die off.
 
They're too complicated to code for and there's nothing they can do that a dedicated Core or GPU can't do for less $$$.

disregard everything else you said. how do you know they're "too dificult to code for"? where has this been stated publicly? have you writtent code for them?

I _have_ writtent code for them and it is a BREEZE to code for the PPU. It's actually easier to code for the PPU than it is to code for either Direct3D or OpenGL. The ball demo that AGEIA has published could easily be written in half a day (or less) and be entirely feature complete.
 
PPU's are a useless component that just takes up space and cut airflow to graphics cards. With Multi-Cores or GPU physics.. who needs an Ageia card?!

They're too complicated to code for and there's nothing they can do that a dedicated Core or GPU can't do for less $$$.

Wheter you like it or not.. Ageia's PPU will never make it. Their software might, but the hardware will die off.

Multiple cores won't do jack and shit. So please spare me that excuse. Also the GPU won't be doing it either, it has enough on its plate. Back to multiple cores there is a better chance the PPU will have more support then multiple cores ever will..
 
Multiple cores won't do jack and shit. So please spare me that excuse. Also the GPU won't be doing it either, it has enough on its plate. Back to multiple cores there is a better chance the PPU will have more support then multiple cores ever will..

while your tone is combative, I understand why. I just wish _more_ people understood this perspective. People just seem to think "more == better" when it comes to the physics processing. It's obvious that most people don't understand the unique nature of physics calculations and how brute force physics processing on CPU or GPU is a losing battle.
 
I don't see many pure 3D cards (e.g. the Voodoo 2) around these days as they were swiftly integrated into the existing 2D cards (e.g. the Voodoo 3) - I expect the same to happen with PhysX.
 
I don't see many pure 3D cards (e.g. the Voodoo 2) around these days as they were swiftly integrated into the existing 2D cards (e.g. the Voodoo 3) - I expect the same to happen with PhysX.


3dFX put it's own 2d soltution with 3D in one chip.

What would Ageia Merge with PPU.
AIPU ? HDaudio? Killernic-PU? a GPU made by Ageia?

Make no sense. nV and ATI GPU don't need ageia tech they already have it's unified shaders. So to merge with GPU it must make one itself.
Chipsets could be? will iNTel or AMD share a licence to do so. not this or next year.

2D merging with 3D is so obviouse PPU Physics with what? Whats close to Physics. CPU is?
CPU setsup PPU workload and CPU recompute result to render next scene to GPU.

There is a strong CPU link.
 
while your tone is combative, I understand why. I just wish _more_ people understood this perspective. People just seem to think "more == better" when it comes to the physics processing. It's obvious that most people don't understand the unique nature of physics calculations and how brute force physics processing on CPU or GPU is a losing battle.

Combative? This always cracks me up when I read this.
 
Combative? This always cracks me up when I read this. No combative is when you post a personal opinion about hardware and 45 guys jump on your ass. THAT is combative.

One thing I have noticed about this forum and that is everyone takes debating hardware way too seriously. If I said dual core sucks 49 guys would jump on my ass. If I said the PPU is good technology I would have 875 guys all take turns flaming me for my opinion. You think I really care if I sound combative? If everyone is that sensitive maby its time to leave the forum?

wow. I was agreeing with you and you flamed me for it. okay, moving on now. note to self, don't have _anyone's_ back on the [H].
 
You can bet 3DMark07 is going to compensate for PPU's.
 
Multiple cores won't do jack and shit. So please spare me that excuse. Also the GPU won't be doing it either, it has enough on its plate. Back to multiple cores there is a better chance the PPU will have more support then multiple cores ever will..

i am a big supporter of hardware accelerated physics, (hardware accelerated everything really).

but i do disagree with you about the GPU not being up to the task of doing physics calculations.

an ageia card currently costs £120 and has a mere 128MB of memory on a PCI bus

an 8600GT with 32 unified SPU's costs £80 and has 256MB of memory on a PCIe bus (and probably clocks twice as fast)

maybe a second cheap graphics card is the way forward?
 
I am still thinking it is a great idea to have a PPU, but ...

1. It is new and not many apps support it

2. Will it become part of the video card?

3. Just think how much more powerful your system would be with multi cpu cores doing other alogorithms and your GPU/PPU combined would be.

4. I believe that the Ageia will be bought once the demand becomes bigger by a graphics chipset manufacturer and will be incorporated into cards, just like 2D and 3D did.
 
if i was you, id wait and see what nvidia and ati come up with for physx.

the question is will they ever >actually< do anything with physics that is meaningful.

this is exactly nVidia & ATI's game, to keep punters guessing long enough to sink ageia.

at least PhysX is hear, now, and implementable in hardware and applications immediately!
 
i am a big supporter of hardware accelerated physics, (hardware accelerated everything really).

but i do disagree with you about the GPU not being up to the task of doing physics calculations.

an ageia card currently costs £120 and has a mere 128MB of memory on a PCI bus

an 8600GT with 32 unified SPU's costs £80 and has 256MB of memory on a PCIe bus (and probably clocks twice as fast)

maybe a second cheap graphics card is the way forward?

i still think the architecture is still unique in the PhysX cards and a standard GPU wont be the same... Even a high end one...
 
well if an 8800gtx can do a certain amount of physx with its unified shaders, surely a dedicated unifiied shader card could do physx? eg, on my striker, 2* 8800gtx sli and 8600gt for physx.
 
well if an 8800gtx can do a certain amount of physx with its unified shaders, surely a dedicated unifiied shader card could do physx? eg, on my striker, 2* 8800gtx sli and 8600gt for physx.

agreed.

while the 8600GT is a less optimised design, engineering wise it is light years ahead.

how many transistors does a PPU have?
what is the core clock?
what lithography process is it made with? (0.13u?)
what is the price compared to an 8600GT?

>if< they can make enough to afford to keep the engineering on a par with a graphics card, then yes, a ground up PPU will be the better choice.
 
Status
Not open for further replies.
Back
Top