Nvidia PhysX Software is Ancient, Slow for CPUs
PhysX for CPUs is built on x87. Not the best choice on modern day CPUs, it seems.
Nvidia's acquisition of Ageia in 2008 was a strategic move to boost the marketability of its GPU offerings. With the discontinuation of the dedicated PhyX boards, the acceleration moved to the GeForce GPU as a differentiation factor that set it apart from AMD's ATI cards.
If a PhysX game detected the presence of an Nvidia GPU, it would move the hardware physics to the video card. Without an Nvidia board, the physics would hit the CPU, which in all cases is slower than what a GPU can do.
It's expected that Nvidia would like to do everything it can to distance itself from the CPU and the GPUs of its competitors, but closer looks at the PhysX software implementation have shown that there could be some shadiness going on.
An excellent investigation by David Kanter at Real World Technologies found that Nvidia's PhysX software implementation for use by CPUs still uses x87 code, which has been deprecated by Intel in 2005 and now has been fully replaced by SSE. Intel supported SSE since 2000, and AMD implemented it in 2003.
The x87 code is slow, ugly, and remains supported on today's modern CPU solely for legacy reasons. In short, there is no technical reason for Nvidia to continue running PhysX on CPUs using such terrible software when moving to SSE would speed things considerably – unless that would make the GeForce GPGPU look less mighty compared to the CPU.
Ars Technica's Jon Stokes confronted Nvidia about deficient PhysX code and we are just as surprised as he was that Mike Skolones, product manager for PhysX, said "It's a creaky old codebase, there's no denying it."
Nvidia defends its position that much of the optimization is up to the developer, and when a game is being ported from console to PC, most of the time the PC's CPU will already run the physics better than the console counterpart. The 'already-better' performance from the port could lead developers to leave the code as-is, without pursuing further optimizations.
"It's fair to say we've got more room to improve on the CPU. But it's not fair to say, in the words of that article, that we're intentionally hobbling the CPU," Skolones said. "The game content runs better on a PC than it does on a console, and that has been good enough."
Another problem is that the current PhysX 2.7 codebase is very old; so old, in fact, that it goes back to before 2005, when x87 was deprecated. Skolones said that Nvidia is working on version 3.0 which should bring things up to date a little bit, though we don't doubt that the GPGPU functions will still be faster.
This isn't the first time that we've heard that Nvidia's PhysX software is less than well-optimized. AMD accused Nvidia of disabling multi-core support in CPU PhysX earlier this year.
- Windows 7 64-bit Creeping Up on 32-bit Installs
- Dell and Threadless Team Up for Funky Laptops
- Blizzard Backs Down on Real Names in Forums
- EA and BioWare Announce Dragon Age 2 for 2011
- Power Strip is Waterproof, Won't Electrocute You
- Any USB Peripheral is a Potential Security Threat
- 3D Displays May Be Hazardous to Young Children
- Gears of War, BioShock Added to GFWL Download
- MS Addressing Google-Exposed Flaw Next Week
- Deals for July 12: More Laptops, Plus Free Gum
- Ballmer Promises Windows 7 Slates
- Spore-Based Action/RPG Revealed at Comic-Con
- eBay, Dell, HP Get Microsoft's Azure ''Appliance''
- Surprise: China Still Welcomes Google's Business
- Comcast Subscribers Get $16 Each for P2P
- 74 Percent of IT Pros Admit to Network Snooping
- Windows 7 SP1 Beta Now Ready for Public Testing
- Intel Says That Celeron Will Continue to Live On
As much as I think Nvidia is a great marketing strategist, I think they dropped the ball on Physx...
This is why Agners CPU blog is a good read.
seriously need better standardization and a massive spring clean with the x86 code base and all extensions since, it could bring a lot of performance improvements with very little expenditure on cpu resources much like optimizing bad code.
Right. You are soo motivated to make it work well even without nVidia's mini-ovens.
GPU scene has some shady part; no doubt about it, even hardware Ageia cards back then is questionable.
Why so lazy, Nvidia? "Better than it was" and "good enough" are not real excuses for neglecting a key technology you market to developers.
I really wouldn't be surprised if AMD eventually brought about and advocated the development of a new "Open-PL" physics standard or something.
DirectCompute / OpenCL >>> PhysX
DirectCompute / OpenCL >>> PhysX
DirectCompute/OpenCL/CUDA != PhysX. They serve different purposes.
Silly to say nvidia is deliberately hobbling physx on the cpu, though; given that ALL the physx code is old (by the author's admission) it's just old code. I'm an nvidia fan but the truth is that physx is so poorly adopted among game developers it wouldn't bother me in the least to just see it go away.
There is one thing to keep in mind, though: MMX is useless for this kind of computation, and SSE was absent from:
- Pentium Pro
- Pentium II
- early K7: Athlon (up to 1200 MHz) and Duron (up to 950 MHz)
So, at the time the software-only implementation of PhysX was written (around 2005, I'd say), there were still some SSE-less machines around.
Of course, Nvidia has no excuse about not making an SSE-optimized build until now, will x87 fallback.
To be fair, it takes almost as much effort to make something run with SSE instructions as it does with GPU. "Deprecated" is a strong word. x87 instructions are still very widely used and are much more efficient today than they were with the old 8087 co-processor. The only time programmers use SIMD (Single-Instruction-Multiple-Data) instructions is when they need to speed up processing on a massive amount of floating point data.
I do agree with your premise, but not the language you use. Optimizing your code to run the SSE instructions can give you about 3x to 4x speedup. However, it takes effort. NVIDIA would rather spend the resources on the parts of PhysX that sell their GPUs than to maintain the CPU compatibility code.
This strategy may backfire, since they still make money by licensing the PhysX technology to game developers. If they cripple their physics engine for non-NVIDIA setups, they will lose revenue and market share to Havok. Developers want their games to work on as many configurations as possible. If they can't have what they want from NVIDIA, they will go to someone who will provide it.
PhysX has always been shady. This should come as no surprise to anyone. I'm not sure why it's still around. Well, other than the fact that Nvidia wants it to be around.
But with multi-core CPUs the norm now, why would I want to dedicate part of my graphics computing power for physics? Rhetorical question; I wouldn't. I'd rather have the game written to be optimized for multiple cores, and perhaps dedicate one core to physics, if it would help. Seems silly to have CPU cores idling while the GPU does double duty.
VHS-Betamax format wars,HD/DVD or BD technology all over again.
When there is one open standard it will become mainstream! Until then it is still sideline to the main game!
Nvidia could do better here by working to get their PhysX as main standard used by correct code development and open licensing with incentive!
I see PhysX going away in the long run if they continue to deal with it so close like they are now being replaced with a more open implementation.
"discontinuation of the dedicated PhyX" That was a bad move for PC gamers.
DirectCompute/OpenCL/CUDA != PhysX. They serve different purposes.
I was talking about their usefulness in game engines for physics simulations.
And DirectCompute/OpenCL/CUDA can do everything PhysX can, but not the other way around, that's why they are better than PhysX.
Also, CUDA is not as good as OpenCL and DirectCompute for games because they are excusive to nVidia cards which don't have the same performance/price ratio as ATI cards.
i wonder is enabling proper physx on cpu to use all cores and use SSE code will render the load on the physx cable video card lower and/or enhance cpu's to the point that the physx card is not required.
In my experience, I don't like physx at all. I have had much smoother gameplay experiences without it, and I don't feel like the physx are that much worse....
Physx.... or Havoc...
Havoc for me.
Why so lazy, Nvidia? "Better than it was" and "good enough" are not real excuses for neglecting a key technology you market to developers.I really wouldn't be surprised if AMD eventually brought about and advocated the development of a new "Open-PL" physics standard or something.
That Id like, but if AMD is gonna be as slow as in the past, and with the HD 6xxx series marked for last half of 2011, I dont think its gonna happen anytime soon
Who cares about all this kiddie peecee gamer crap anyways? Not me.
I was wondering when this biased stuff was going to show on Toms.
Taking the two primary points of the investigation:
1: PhysX isn't multithreaded by default
2: X87 is old and depriciated
My response:
1: DirectX, OpenGL, C++, JAVA, etc are not multithreaded by default
2: While NVIDIA's implementation isn't the best, it should be realativly simply for developers to replace the offending code with SSE instructions. In short: Implementation is up to the developer [which is the same design concept that DirectX follows]
Nothing to see here.
This is what happens when you secure your position in the market.
That Id like, but if AMD is gonna be as slow as in the past, and with the HD 6xxx series marked for last half of 2011, I dont think its gonna happen anytime soon
AMD has been pushing an open physics API of some sort, but yeah, they also promised hardware-accelerated Havok. I haven't seen that shown off in a game. At least in AMD's case, optimizing physics on both CPU and GPU is important.
It is obvious. They need to sell their video hardware. The physx software implementation must be run slower than their hardware equivalent.
Physx.... or Havoc...Havoc for me.
Havok is showing age as well and needs a major update.
moricon 07/12/2010 4:09 PM Hide -1+
"VHS-Betamax format wars,HD/DVD or BD technology all over again."
don't bother referencing HDdvd or Blueray it's like comparing stereo to dolby and was blown out of the water the moment THX showed up. comparing HD/Blue ray is more like 8 track, they were both blown out of the water the moment you could download them off the internet. doesn't matter their format it could be laser disc for all any one cares. it's obsolete the second something better came along, which it had before HD and Blueray even came to market.
the issue at hand is how much longer are we going to need a decicated graphics card with the advent of multi core cpu.in hind sight i see AMD scoring a huge win the moment they are able to take the ati chip and put 1-2 cores dedication for video onto the cpu itself like intel has done with the new i-5's. nvdia will be screwed if they don't find a way to work with cpu makers fast to incorporate their gpu chips into cpu multi core technology. intel has been pushing their own in house graphics chip for quite awhile now, i seriously doubt they'd want to waste the money buy out a spendy company like Nvidia, they would be better off buying out Havok, and screwing over Nvidia.
it's just a matter of time now, the clock is ticking, and Nvidia is going to be dust in the wind if they don't play some very serious catch up immediately. i'm just sad that 3dfx went under after basicly starting down the multi core road 10 years ago, and Nvidia never bothered to continue that work until now that it's almost too late.
very well played forward thinking strategy AMD, very well played!
Nvidia makes GPUs, to me it seems perfectly reasonable that they would spend their resources on improving the performance of their proprietary software on their proprietary hardware.
Even if Nvidia completely overhauled the Physx code for better CPU compatibility, an old (GeForce 8 series) GPU would still run it must faster.
Physx is an implementation of CUDA, CUDA is designed to run on Nvida GPUs. It isn't rational to expect Nvidia to devote much time or effort negating one of their competitive advantages.
I think it should be obvious that physics was a big real a couple of years ago but everyone is falling all over themselves right now to get 3d products to market. nVidia is definitely on the 3D bandwagon now, physx isn't where development is happening currently.
I was wondering when this biased stuff was going to show on Toms.Taking the two primary points of the investigation:1: PhysX isn't multithreaded by default2: X87 is old and depriciatedMy response:1: DirectX, OpenGL, C++, JAVA, etc are not multithreaded by default2: While NVIDIA's implementation isn't the best, it should be realativly simply for developers to replace the offending code with SSE instructions. In short: Implementation is up to the developer [which is the same design concept that DirectX follows]Nothing to see here.
Bias, indeed. You wouldn't happen to be running an Nvidia card, now, would you?
lol @ any idiots that actually bought a PhysX card back then.
So.... When they do change the code for PhysX, does that mean games of today that uses an obsolete code will not work on future machines? Just Cause 2 for example will become a KOTOR/ Vietcong problem of the future?
How does crysis 2 do its physics? Let their physics be open source and boom you got the best looking and working games on the pc. Take into mind this is a what if crysis 2 is in a very optimized situation.
laziness is not shadiness. Why would they bother to optimize for someone else's hardware? No upside. That's not shady, it's lazy.
It's not as if they took SSE code and made it x87. They bought old code and didn't ever improve it. Obviously, that part of the code was not where their interest lay...