I can understand nV blocking ATI + Geforce (all business of course) but blocking ATI + PPU just really rubs me the wrong way...they deserve any bad PR they get for that move. As BFG mentioned it's a good thing ATI didn't go with PhysX...who knows what stunts NV would have pulled. Hopefully MS implements some standard into DX for physics.
SunnyD
Belgian Waffler
You need to clarify this BFG: PhysX does NOT work on the 5870, period. The hack does work to enable PhysX on a dedicated Nvidia GPU in a hybrid video card setup.
It's a pretty big misnomer to say that PhysX is working better on a 5870 rather than Nvidia's native solution (in fact, in this case it still is on Nvidia's native solution: a Nvidia GPU).
I'm not trying to split hairs here, because to most of us your post is rather plainly obvious. But it should be clarified before the likes of Wreckage and Co. get ahold of your blurb there and start chest-thumping over the semantics of the "corrective beatdown" they could potentially have you on there.
That being said, it saddens me to see that PhysX, introduced by AEGIA to be an open and vendor-agnostic API, has been cut down so hard by Nvidia since their acquisition. At every turn, Nvidia said that PhysX would remain open and available, yet all they have done since is close it off trying to make it into the niche must-have product. What's worse is that they continually "break" previous functionality in manners such as this. This is most definitely not a matter of hardware testing and QA issues - PhysX has absolutely no dependency on the hardware renderer used; it never has... otherwise the PPU would have never worked. PhysX was designed to be hardware agnostic, which is the exact reason the software stack can even survive without accelerated hardware (read: can run quite happily on the CPU, albeit with lower performance).
One can't fault Nvidia for trying to preserve its market share however. Especially in the face of their market situation today, this is yet another obvious outcome of the situation. I find it rather short-sighted though, especially given Nvidia's stance on the "GPGPU" taking over computing. If that was the case, one would assume Nvidia would want to push their non-graphics solutions just as hard, if not harder than their graphics solutions. Any wins they make with the latter would become a bonus to their change in business philosophy.
This, unfortunately, is yet another example of the hypocrisy that is Nvidia in the market place today. They don't truly want to change and adapt to the market. No, they would rather be able to use their (eroding) market position to stamp out any possible competition they have, stifle any further innovation, and essentially do as little work as possible to accumulate and sit on fat piles of cash. Hopefully they will learn of their mistakes here, just as AMD and ATI have learned from their recent misfortunes. The difference here is that the latter two have done what they needed to, and have become relevant, and dare I say it: Competitive.
I'll end my little editorial with this: It would be a shame if Nvidia comes away from this not learning it's lesson, continues to be anti-competitive and anti-consumer. For all the innovation Nvidia has put towards computer graphics over the years, it would be sad to see them disappear from the landscape. Unfortunately, that's the path they seem to be taking at this point.
Ask ATi, they will not allow vid capture to work using an AIW if a nV GPU is your primary display, exactly the same thing that nVidia is doing.
Wrong.
Yes, in that game it worked without issue with the drivers they happened to be using. This is exactly the same as ATi locking out the vid capture device functionality of the AIW parts when using a nV GPU as a primary display. That it could work flawlessly under certain conditions doesn't mean it always will work.
AFAIK PPUs still function under XP and Vista, just not under Win7 which isn't supported.
Amber confirmed it for you last time we had this conversation.
Go ahead and turn PhysX on in game without a hack right now usinig an ATi part, it will run on the CPU without any additional work.
I would love to see a comparable counterpart to PhysX that was cross platform- Havok is the only other option and that can't handle the level of effects that PhysX can. It seems to me that the other companies at play here are the ones that want to sit on their cash and refuse to innovate. This is a lot like Glide was back in the day. Until DirectX was remotely competitive, I liked Glide a lot better. When a viable alternative became available my stance changed considerably.
Qbah
Diamond Member
- Oct 18, 2005
- 3,754
- 10
- 81
How is the tuner thing relevant to "How to Activate Hardware PhysX to play with an ATI Video Card " ? 
HardOCP showed that PhysX runs great in a mixed config, even better than in a pure nVidia configuration. This is completely different from what nVidia claims (so they're basically full of it). Not to mention this hack is not for "the ignorant Joe" that goes to Best Buy and picks a 9500GT over a HD4870, just because the nV card has 1GB RAM (compared to 512MB for the Radeon). You don't see mentally handicapped people running dual-card configs, it's mostly done by enthusiasts. Bottom line is, it runs great (better!) on an ATi+nV combo than it does on an nV+nV config. It's easy to configure and install. And there are no drawbacks.
And why are you grasping at the poorly worded things from BFG? Everybody with at least a bit of clue knows PhysX is running on the GTS250 in the hacked configuration. That's the whole point! We know you can run PhysX on the CPU and get single digit FPS as a result - nobody is saying the Radeon is doing PhysX calculations. So why pointing the obvious? BFG clearly knows that. People interested in this topic know that too.
EDIT: As far as the tuner thing goes, did someone hack it back in? Or was it like that from the beginning? Are you sure it's not something in your config? Plenty of people have said they've been running a GF card with the tuner card without issues.
EDIT2: Did some searching earlier and there are hardly any complaints for this not working, mostly from years ago and even then pointing at ATi drivers sucking bad. Tried to look a bit more now and didn't come up with anything new. If you have problems properly configuring your setup, I'd say start a new topic - it's completely irrelevant here as it looks it's more an issue on your end rather than ATi blocking anything
HardOCP showed that PhysX runs great in a mixed config, even better than in a pure nVidia configuration. This is completely different from what nVidia claims (so they're basically full of it). Not to mention this hack is not for "the ignorant Joe" that goes to Best Buy and picks a 9500GT over a HD4870, just because the nV card has 1GB RAM (compared to 512MB for the Radeon). You don't see mentally handicapped people running dual-card configs, it's mostly done by enthusiasts. Bottom line is, it runs great (better!) on an ATi+nV combo than it does on an nV+nV config. It's easy to configure and install. And there are no drawbacks.And why are you grasping at the poorly worded things from BFG? Everybody with at least a bit of clue knows PhysX is running on the GTS250 in the hacked configuration. That's the whole point! We know you can run PhysX on the CPU and get single digit FPS as a result - nobody is saying the Radeon is doing PhysX calculations. So why pointing the obvious? BFG clearly knows that. People interested in this topic know that too.
EDIT: As far as the tuner thing goes, did someone hack it back in? Or was it like that from the beginning? Are you sure it's not something in your config? Plenty of people have said they've been running a GF card with the tuner card without issues.
EDIT2: Did some searching earlier and there are hardly any complaints for this not working, mostly from years ago and even then pointing at ATi drivers sucking bad. Tried to look a bit more now and didn't come up with anything new. If you have problems properly configuring your setup, I'd say start a new topic - it's completely irrelevant here as it looks it's more an issue on your end rather than ATi blocking anything
Amber, is this true? Do you have proof that ATi was intentionally "locking out" video capture when an Nvidia card is present alongside an ATi All-In-Wonder card? Because that simply doesn't make any sense to me. For years I used an Nvidia GeForce3 Ti200 in conjunction with an ATi TV Wonder PCI and the video capture function performed flawlessly. And before that, I had a GeForce 256 installed that also co-existed happily alongside the TV Wonder.
If ATi were deliberately "locking out" the video capture function of an All-In-Wonder card in the presence of an Nvidia card, then why would they allow it to work with their TV Wonder card?
This sounds more like a software or hardware incompatibility between the GPU in the All-In-Wonder and the Nvidia card than a deliberate "locking out".
It has always worked fine with the tuner card, it won't work with the AIW.
ATi has it crippled at the driver level.
Then could you please post a link verifying your story, thats all that is being asked, I'd hate to recommend people to use that setup if it indeed does not work.
BFG10K
Lifer
- Aug 14, 2000
- 22,709
- 3,002
- 126
Again you need to provide evidence of this. You claim they?re disabling it on purpose so the burden of proof is with you. Simply repeating that it doesn?t work for you is not evidence since there could be a range of reasons why it doesn?t work.
What it proves is hardware acceleration on a secondary nVidia card provides better performance overall on a 5870 than a GTX285, thereby proving that nVidia?s block is simply an artificial one to cripple the competition.
But I think you know exactly what I was saying, you just love arguing semantics and splitting hairs, kind of like when you were arguing the difference between a ?CD check? and a ?disc check?.
Actually no, it?s not the same thing at all. One is a clear and artificial vendor lock that when defeated by the community, the system runs better as a whole on the 5870 than it does on the GTX285.
The other is your unfounded speculation that ATi is actively blocking features based on the vendor installed in the system, speculation that has no concrete evidence to back it up.
Are you doing this on purpose, or are you truly not aware of what has happened?
I?ll give you the benefit of the doubt and assume you simply aren?t aware of recent developments:
http://physxinfo.com/news/338/...ageia-ppu-is-affected/
The PPU was working fine before, now it stops working after a driver update if an ATi card is detected, and it starts working again if the ATi card is removed with the same driver.
She did no such thing.
Yes it will, even if you have a PPU installed, and that?s the problem. Can you describe to us in your own words what the purpose of an AIB like a discrete PPU is? Do you feel its purpose is to run PhysX on the CPU? If so that would simply make it a paperweight.
Hahah, I just had to emphasize that. Anyways, list of facts below. These should be completely non-arguable.
1. PhysX only works natively on Nvidia's GPU with their current drivers
2. Nvidia has locked out PhysX through software in systems detected with any other GPU
3. Nvidia states #2 is for performance/compatibility/business reasons
4. Hacks allow ATI + Nvidia GPUs to work in spite of #2
5. Future Nvidia patches could stop #4
6. HD5870+Nvidia PhysX card is faster than GTX285+Nvidia PhysX card
7. #6 implies that there is nothing special about the "primary" render to affect a dedicated PhysX card
8. There is nothing wrong with #2
9. As to how much business sense #2 makes, it depends on how customers perceive it
10. Most important of all, PhysX is no big deal at all, so almost all of this is moot.
11. Physics, is a big deal and the future is open
It's possible to run the entire Physx package from a game using only a CPU & ATI GPU. The reason it's possible is because there is nothing special about Nvidia hardware that makes Physx an exclusive capability. Physx's exclusivity has only to do with programming. The patch that enables Physx (in Batman for example) makes the physics calculate on the CPU in a single thread; a secondary patch tries to mulithread these calculations to make them more efficient on quad core CPUs. Nvidia has Physx highly optimized to be done on >8800 GPUS, it is very efficient; and ironically inefficient on a CPU.
The math intense calculations for rendering things like sparks, fog, flying objects, cloth, & debris are bundled into the CUDA/Physx package and allow the GPU to concentrate on rendering the main bulk of the image at an acceptable framerate. This is apparent when you enable physx using only one Nvidia GPU and get worse performance in exchange for more eyecandy.
You take a game like Batman, bundle all those special effects and make them the Physx package, and the resulting 'base' game has less eyecandy. Doesn't mean that the eyecandy isn't there, it's just hidden. Perfect example is the Batman game, and having all those great effects done on a Corei7. Supposedly all this is becoming less relevant with the adoption of OpenCl. Nvidia is no doubt good at controlling the market. But it isn't doing us any good. They just sit on their butts re-releasing and re-branding the same old crap for newbies to purchase. That doesn't say much for innovation, but says a great deal about their theory on competition. Imagine if Physx had a 25 times greater prescence in games (which is Nvidia's goal). Why would you want a 3x faster Radeon when you have to have an Nvidia card to get the special effects. But you're disappointed that it's now 10 years down the road and the fastest Nvidia card is a GT200b or a G92b.
SunnyD
Belgian Waffler
I know, seriously. It's one of those "don't say I didn't tell you so" moment. *sigh* What makes it worse is that I knew exactly that Wreckage et'al would fail to quote the first part as well.
Let the record show also that I am in no way supportive of Wreckage, the focus group, or Nvidia's tactics here. For the love of Pete, I sure hope Wreckage isn't thinking I'm on his side. I was pointing out a technicality that I simply knew he would attempt to exploit just like any other marketing shill would have.
There are compatibility reasons not to run Nvidia and ATI in the same system. Nvidia doesn't want a bunch of tech support calls because of incompatibilities between drivers. If ATI would port it to OpenCL they could then QA it from their side. A 5870+PhysX card should be faster than a 285+PhysX because the 5870 is faster. Compare the 295+PhysX to the 5870+PhysX for a more accurate comparison. And if PhysX gets a 2 year lead over other solutions it will be a big deal.
This has been the case always. And yet, if I choose to have an ATI onboard GPU and an nvidia card I can do that. I have done that. And I've had absolutely zero problems. As evidenced by this working just fine previous to the 186.5 NV drivers.
Are you for real? Who in their right mind calls a chip manufacturer after slapping together a frankenstein multi-GPU system from two vendors? As pointed out, "special" people don't build gaming hardware like this. And if they did, they'd run into problems long before being able to run a game in the first place. They'd be just as likely to call Intel or NV for reasons of not being able to find their zipper. Normal people would google for solutions, troubleshoot by removing one or the other piece of hardware, etc.
The problem is not ATI doing the port. The problem is the PhysX middleware being 100% the intellectual property of NV. One they love to control tightly, as evidenced by this lockout. ATI made the right choice in not licensing or otherwise adopting this technology. It needs to die ASAP so we can move on to a standards based solution available to all makers of GPUs, past present and future.
Originally posted by: Keysplayr
#6 question. Isn't a 5870 faster than a GTX285 to begin with?
#8 If PhysX is no big deal, we wouldn't be having this thread now would we.
Agree with pretty much everything else.
I agree with your #6 analysis. Sometimes the obvious need to be pointed out.
Heh, I think this thread has continued for more reasons than PhysX. If it were just PhysX, the thread would've went something like this "OP: Woo hoo, I got ATI and Nvidia to work with PhysX, Reply: Yay, nice stuff and good find. /thread"
Originally posted by: KIAman
If it were just PhysX, the thread would've went something like this "OP: Woo hoo, I got ATI and Nvidia to work with PhysX, Reply: Yay, nice stuff and good find. /thread"
Funnily enough...that's exactly what I said when I did it a while back:
Thread title: "WOOHOO!! Got ATI + Geforce PhysX working in Win7!!!"
http://forums.anandtech.com/me..._key=y&keyword1=woohoo
BTW, BS, yes, PPU is broken in any OS, thanks to Nvidia.
Of course, as soon as I replace the cuda.dll it works again...
NV = bunch of f@*&^% disgusting, PoS low-life liars.
Of course, as soon as I replace the cuda.dll it works again...
NV = bunch of f@*&^% disgusting, PoS low-life liars.
Ask them. It is really that simple.
Because it works under one set of circumstances, of course that proves everything would always be flawless.
Try this, take a clean install of Win7(just assuming that is something you may have around, will work with XP or Vista too) with a nVidia GPU and PPU and install the latest drivers. The PPU won't work. You need to perform a hack to get it to work at all currently with the latest drivers, if you are using nV or ATi. My comment was based on the fact that nV has dropped support entirely for Win7 for PPUs. Even using the legacy OSs atm the PPU will not work out of the box if you install the latest drivers.
You are correct, she dodged the question in that thread. Easy way, ask AMD.
Right now one of two things is happening, either there is a major bug in the PPU drivers or nV is dropping all support across the board. The PPU isn't working on nV parts atm using a clean install(but it will if you install older versions first and then update over them).
It serves no purpose at all anymore, that's why it was discontinued a long time ago.
It's broken on nV parts atm too. I'm not saying they aren't blocking it, but as of this moment it looks like they have shut it down across the board(although there are different hacks you can do to get it working for both nV and ATi).
Qbah
Diamond Member
- Oct 18, 2005
- 3,754
- 10
- 81
From the same article at HardOCP you could see there's a bug on the nV+nV combination causing a massive FPS drop at one point. It doesn't occur on the ATi+nV combination. Obviously PhysX should be blocked on nVidia's hardware until they get it solved and test it on every possible hardware configuration?
Surely you're kidding... No technology is flawless. Claiming that it should be allowed only when fully tested is silly. That's not even possible on an all-nV configuration.As for the PPU - if they're dropping support for Windows 7 in general for it, I have no problem with it. It's painfully slow for the amount of PhysX effects anyway.
I don't know that I would call running out of RAM a bug. ATi's AA uses less resources then nV's, we have known this for a long time- the nV parts are clearly running into an issue with RAM limitations in that bench(dropping down one resolution notch increase min FPS by over 700%).
No, but I don't think they should be expected to even attempt to support such a setup. The AIW thing I was talking about isn't a major deal, I use it on one of the older setups and it serves its' purpose, even if it would be nice to be able to use it in the main machine(if it was that big of a deal, I'd just buy a stand alone board). What I find absurd is that people are b!tching that nV isn't supporting a feature that improves 3D graphics when another GPU is being used to render those graphics. Sure, it may work fine right now, but the notion that they should honestly be expected to support such a setup? I fully understand why the AIW has technical issues making support for it in a multi GPU box as anything other then primary display a potential nightmare, that's the reason why I don't post a new thread whining about it every couple of weeks.
BFG10K
Lifer
- Aug 14, 2000
- 22,709
- 3,002
- 126
Uh-uh, these little games you?re playing don?t fly. You made the claim so the burden of proof is with you.
You ask them, or link to credible evidence that they?re specifically blocking features on the sole basis of another vendor being in the system.
You either provide this evidence so we can agree with you, or retract your baseless claim.
?Ask AMD? isn?t evidence.
I never claimed that, so stop trying that strawman. I simply claimed there?s widespread success and that there?s no obvious reason why nVidia is blocking, other than for artificial reasons. Again, has PhysX been tested on every single platform configuration in existence? No, but that?s not relevant any more than your statement is.
Even if that?s the case, it?s absolutely irrelevant to the article clearly demonstrating the PPU won?t work on an XP box with newer drivers unless the ATi card removed, but it works fine with the ATi card on older drivers, which is something you denied last time.
But it wasn?t discontinued, because the article clearly demonstrated it functions when the ATi card is removed.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter