Physix Cards

[bobthemoo]

So whose got one?

Any good? noticed any realy performance increase? Also, whta are your 3dmark scores?

[Burninate]

So whose got one?

Any good? noticed any realy performance increase? Also, whta are your 3dmark scores?

From the reviews I've read, they actually lower the performance somewhat. The physx cards just allow you to enable more eye candy and make explosions and stuff look better, but as a result of there being more particles on screen there is a bit of slowdown (I've heard from 1-3 FPS). And as for 3dmark I don't think the cards actually improve or help anything unless the program/game is specifically designed to take advantage of it. I'm not an expert, but thats what I've gathered from the few reviews I've read on them. They sound kind of like they're little more than another way to separate you from $250+.

[chevro1et]

Waste of money at this point... not well implemented or optimized. Very cool in concept though, something to keep an eye on for future.

[GruffiGummi]

they blow, dont waste your bones

[IWasHungry]

Aye, they're pretty expensive and don't really do much.

[CHECKERED]

Perhaps in the future games will be designed to make use of a separate physics card. Like Half Life 5…ha

The upcoming ATI RD600 based Mother boards, have 3 PCI-E slots, two for Crossfire and a third for physics, but the third one only operates at x2….a waste..

[Neural Net]

Wait a second, with the CPU industry making the move to dualcore processors, wouldn't utilizing the second core for physics be a more realistic option, seeing as its just sitting there waiting to be used?

[jivetrky]

Yeah, it's not that they blow...it's just that they aren't really useful right now. But the future would be different.


And as for them slowing down...that is because they are causing many more particles for your video card to render...slowing it down.

[JigPu]

Wait a second, with the CPU industry making the move to dualcore processors, wouldn't utilizing the second core for physics be a more realistic option, seeing as its just sitting there waiting to be used?
Yes, but by the same logic one could argue something very similar:

"With the CPU industry making the move to dualcore processors, wouldn't utilizing the second core for graphics be a more realistic option, seeing as its just sitting there waiting to be used?"

CPUs are great at being flexible, but they will never be as fast as a dedicated solution. Video cards are used because they are way faster than the CPU at creating graphics. Similarly, the PhysX cards (at least on paper) are amazing at doing physics calculations, and relieving the CPU of that task should be helpful to gameplay.

The PhysX cards currently make no real difference to gameplay because of at least two problems: they're bottlenecked by the PCI bus (almost every review has noted a short stutter in the game at the moment an explosion goes off while using the PhysX card), and games which use them inevitably don't let you use the card without also increasing the ammount of physics used (so you can't just accelerate the game--you also increase the physics workload :()

JigPu

[BossBorot]

Yes, but by the same logic one could argue something very similar:

"With the CPU industry making the move to dualcore processors, wouldn't utilizing the second core for graphics be a more realistic option, seeing as its just sitting there waiting to be used?"


Isn't that exactly what (ami/ati/daamit)'s gpu+cpu solution is intended to do?

sure it wont use a second cpu core for it but they could put a gpu on the same die.

"It's also proper to suggest, as I did in my coverage of Sun's MAJC back in 1999, that a GPU will probably wind up on the same die as a CPU at some point. Indeed, AMD's Dirk Meyer mentioned a dual-core, CPU/GPU combination with a shared L2 cache as something they're looking at. All of this being the case, AMD will need graphics expertise if they're eventually going to play that game. So die-level CPU-GPU integration is one benefit among many that the new merger affords AMD, which is pretty much exactly the way it was presented in the conference call. There's no need for grand conspiracy theories here."*

with the move towards mini cores in the near future what going to prevent cpu makers from devoting part of their CPUs to other formerly dedicated options? Most likely we wont see a move towards CPUs handleing graphics or physics for prob a two to four years, but most likely we will see it.

you can even hear nVidia's Director of Strategic Marketing, Kevin Krewell, talk about this in twit 63 with the old techtv gang starting around min 5. http://www.twit.tv/63

* http://arstechnica.com/news.ars/post/20060724-7333.html

[Neural Net]

I can't help but be pessimistic about this idea of the GPU being on the same die as the CPU. Firstly, wouldn't the GPU performance be compromised simply because it wouldn't have its own dedicated graphics RAM, and wouldn't it make overclocking much more difficult (and drastically limit it)?

My point for using a second core is, we all know the Havok physics engine in HL2 is pretty good, if it had an entire core to itself for caculations, couldn't it be taken to the next level? Such as deformable terrain, many more breakable structures? I'm sure for the moment it would be more efficient than the PPU.

[jivetrky]

I think that, at least in the beginning, integrated graphics will not be anything high end. It'll be used for cellphones and PDA's and stuff like that.

I think you're right, not having the dedicated RAM will surely impede performance.

[PhysX]

So whose got one?

Any good? noticed any realy performance increase? Also, whta are your 3dmark scores?

performance increase in physx api "with card" = 1533% gain over using cpu to emulate physx "with no card".

so yes it is better.

[jivetrky]

Wow, is this a real Ageia person? Or just a big fan?

[Sgt_Pinto]

wat about using graphics cards for physics? i heard that the 7 and x1000 series can be used for physics.

[Maviryk]

I installed the physics card runtime engine w/o the physics card, and scored 100pts higher in 3dmark06.

[PhysX]

wat about using graphics cards for physics? i heard that the 7 and x1000 series can be used for physics.

you can use the x1 serires cards from ati for physics, but not ageia physics, just havok physics :)

nvidia has their own thing but requiere you to have SLI setup and use one for physics.

[Shell]

you can use the x1 serires cards from ati for physics, but not ageia physics, just havok physics :)
nvidia has their own thing but requiere you to have SLI setup and use one for physics.
And Havok physics resides in many more titles then AGEIA physics... at the moment.

I have a PhysX question:
WHY?
For the cost of a PhysX card anyone can just buy an X1900 GPU and that has over 8x the processing power(X1600 has nearly double) and is capable of soo much more...

PhysX is useless and will be for a long time, if ever... IMO...

[metloaf]

As previously stated, the concept is interesting and does have some future possibilities. The current application however is definately not worth a $250 price tag. I considered getting one when they came out. After reading all the reviews and info out there I still think that for now they are a waste of money. :beer:

[PhysX]

Yes, but by the same logic one could argue something very similar:

"With the CPU industry making the move to dualcore processors, wouldn't utilizing the second core for graphics be a more realistic option, seeing as its just sitting there waiting to be used?"

CPUs are great at being flexible, but they will never be as fast as a dedicated solution. Video cards are used because they are way faster than the CPU at creating graphics. Similarly, the PhysX cards (at least on paper) are amazing at doing physics calculations, and relieving the CPU of that task should be helpful to gameplay.


on the other hand CPUs are usable, im not sure if you have heard of the game alan wake, but it uses the cpu to do physics for the game, one core of the cpu :) so its quite useful :)

[PhysX]

And Havok physics resides in many more titles then AGEIA physics... at the moment.

I have a PhysX question:
WHY?
For the cost of a PhysX card anyone can just buy an X1900 GPU and that has over 8x the processing power(X1600 has nearly double) and is capable of soo much more...

PhysX is useless and will be for a long time, if ever... IMO...


WHY? for the same reason as the KillerNIC hahah...but there will be so much more to PhysX in the future.

[Speciale]

i'm intreagued

[PhysX]

Documents Leak NVIDIA's Quantum Physics Engine
Kristopher Kubicki (Blog) - October 5, 2006 1:21 AM
Print article Email article 37 comment(s) - last comment scrapsma54.. on Oct 30, 2006 at 4:35 PM
Recipient E-mail Please enter a valid E-mail addressPlease enter a valid E-mail address
Sender E-mail Please enter a valid E-mail addressPlease enter a valid E-mail address
NVIDIA is ready to counter the Triple Play

With the release of the G80, NVIDIA will also release a new engine dubbed Quantum physics engine. Quantum Effects Technology is similar (at least in spirit) to NVIDIA's PureVideo Technology -- a dedicated layer on the GPU for physics calculations. A few documents alluding to this new engine appeared on public FTP mirrors late last week.

Quantum utilizes some of the shaders from NVIDIA's G80 processor specifically for physics calculations. Physics calculations on GPUs are nothing new; ATI totes similar technology for its Stream Computing initiative and for the Triple Play physics.

NVIDIA and Havok partnered up this year claiming that SLI systems would get massive performance gains by utilizing additional GeForce GPUs as physics processors. Quantum may be the fruits of that partnership, though NVIDIA documentation clearly states that Quantum will work just fine without SLI.

NVIDIA's documentation claims Quantum will specifically compete with AGEIA's PhysX, yet does not mention who is providing the middleware. Given that there are only two acts in town right now, it would be safe to say Havok has a hand in the Quantum engine.

source:
http://www.dailytech.com/Article.aspx?newsid=4444