PHYSX NEWS PHYSX SDK
PROJECTS TABLE
GPU PHYSX
GAMES INFO
PHYSX
ARTICLES
PHYSX WIKI FORUM
РУССКИЙ ENGLISH


:: Back to news index ::

Nvidia cripples PhysX performance: True or False ?

with 19 comments

Recent news coming from NGOHQ website with statement “performance on non-Cuda cards got dropped even further (from 20 fps to 5 fps [Mirror's Edge]) in the recent PhysX software” have forced us to perform more detailed testing procedures, to discover how actually valid such claims are.

Update: From NGOHQ:

After deeper investigation, this issue has identified as a rare overflow bug. Looks like Nvidia PhysX System Software 9.10.0222 doesn’t cripple performance, but I’ll run more tests just to be sure

In following article we are comparing both GPU and CPU performance of PhysX effects, using most recent PhysX System Software 9.10.0222 (blamed for “ruining game experience”), previous 9.10.0129 version and several games – Mirror’s Edge 1.01 and Cryostasis Tech Demo.

System: C2Q 9400 @2.66 GHz CPU, GTX 275 GPU, 4GB RAM, Windows 7 Professional x64

MIRROR’S EDGE.

Standart “Flight Fly By” sequence was used (can be activated by “- FlybyFlight” parameter), in-game setting; 1680×1050, all high, 4xAA. Framerate measured by FRAPS 3.0.1.

As you may see, there is almost no difference in PhysX performance with different System Software versions. Moreover, our framerate pattern correlates with early benchmarks, published more than a year ago, like  Overclockers Club or Tech Report.

CRYOSTASIS TECH DEMO


In-game settings (click to view). Framerate measured by internal benchmark system.

Again, framerate with 9.10.0222 and 9.10.0129 drivers is equal. CPU PhysX performance is similar to other tests.

SUM: according to our tests, there is no difference in performance between 9.10.0222 and 9.10.0129 System Software, either for CPU or GPU calculations. We can not also verify any performance degradation compared to previous benchmarking results.

Thus, we see two explanations of original statement:

1) Performance impairment really does exist, but does occur only when non-Nvidia GPU is spotted in the system.

2) Mr. Eran “Regeneration” Badit has provided not very accurate information, either without malicious intent or with it.

Written by Zogrim

April 19th, 2010 at 3:07 pm

19 Responses to 'Nvidia cripples PhysX performance: True or False ?'

Subscribe to comments with RSS

  1. najz x)

      

    MicrO

    19 Apr 10 at 3:46 pm

  2. How about adding Batman AA to the benchmark list

      

    Strike

    19 Apr 10 at 5:34 pm

  3. Strike
    How about adding Batman AA to the benchmark list
    Well, since I have Steam version, and it’s not installed – it may take some time to add.

      

    Zogrim

    19 Apr 10 at 5:38 pm

  4. Sounds like the guy didn’t delete the software PhysX files from his installation of Mirrors Edge. I had the same problem with my hybrid setup then deleted:

    PhysXcore.dll
    PhysXDevice.dll
    and the folder called PhysXLocal

    It ran great after that.

      

    Jazar

    19 Apr 10 at 7:47 pm

  5. So the statement was:
    “performance on *non-Cuda* cards got dropped even further in the recent PhysX software”

    Your system: C2Q 9400 @2.66 GHz CPU, *GTX 275 GPU*, 4GB RAM

    So what’s the point of this testing?

      

    GenL

    19 Apr 10 at 11:18 pm

  6. So what’s the point of this testing?
    Strange question – point was to test CPU PhysX performance with different drivers, to verify if there is performance degradation or not. GPU PhysX scores were added as bonus.

    Or you would say that CPU PhysX with NV GPU in system and CPU PhysX with ATI GPU in system are different things ?
    If you can run a fly-through in ME and verify – yes, with 9.09.0010 and my ATI GPU I have 30 fps, but with 9.10.0222 I have 5 fps – I will be first who’ll rase the hype and blame NV for that.
    But stuff like original news (claims without any significant proof) from NG must be countered, and will be countered.

      

    Zogrim

    19 Apr 10 at 11:29 pm

  7. I think you’ve missed the point. See…

    guy1: guys, i have a radeon and i feel that my CPU-PhysX performance got crippled. And i found that a piece of nvidia software is to blame.
    guy2: guys, don’t listen guy1. I have a geforce and i can assure you – there are no performance drops.

    From my pov, guy1 has at least a chance to prove his point by installing older PhysX SS and comparing results. And guy2 can prove nothing… unless he has a radeon too.

    Yes, PhysX software doesn’t need to call a cuda card to decide if it is installed in the system. I saw some portions of code inside. I’m not going to debug it, but i can say – it is possible.
    You know, when even CORE file from Engine\v2.8.1 contains the “nvapi.dll” string, it says a lot already.

      

    GenL

    19 Apr 10 at 11:54 pm

  8. guys, don’t listen guy1. I have a geforce and i can assure you – there are no performance drops

    I’ll just quote myself
    we see two explanations of original statement:
    Performance impairment really does exist, but does occur only when non-Nvidia GPU is spotted in the system

    You see, I’m giving guy1 as much room as possible to prove his point. Task of PhysXInfo is not to promote PhysX, but provide valid and actual information about PhysX – so I’m on his side, and you missed that, actually.

      

    Zogrim

    20 Apr 10 at 12:15 am

  9. If the task is to provide valid and actual information, you should have noted that you can’t provide the actual results on GPU like Regeneration owns. Because you don’t have it, and it is crucial to the subject. No offense.

    I’ve just tested ME on my system with 8.09.04 (the oldest i have) and 9.10.0222. Didn’t find real difference in FlybyFlight, but someone else should check this too, because my system already met some nvidia card recently, and i’m not 100% sure i cleaned all drivers properly.

      

    GenL

    20 Apr 10 at 12:32 am

  10. ADD: I’ve tested it on XP. I assume others should test it on 7. Believe me, it may affect such things.

      

    GenL

    20 Apr 10 at 12:39 am

  11. you can’t provide the actual results on GPU like Regeneration owns
    Definitely. And thus I’m not saying that he is wrong, I’m saying that he may be wrong ;)

    Believe me, it may affect such things
    You know, GenL, for all 5 years I’m watching over PhysX, I saw hundreds and hundreds of people telling how enabling PhysX is overheating their cards, how System Software is crashing (or conversely – boosting fps) non-PhysX games and affecting monitor resolution, and other crap – this gave me right to be sceptical, cause I was never wrong.

    Just look at that from my position – CPU PhysX was always working regardless to GPU manufacturer, and now you’re just so easy saying that it has changed ? sorry, no way I would believe without solid proofs

      

    Zogrim

    20 Apr 10 at 1:20 am

  12. Zogrim
    You are not saying he is wrong, but you are not trying to prevent people from being misleaded. You’ve bolded “no difference” text 3 times, “non-Cuda” text – zero times.

    >5 years
    I meant that about importance of OS (7 vs XP in this case), not about PhysX overall.
    I did tests and i can say that latest PhysX runs slightly different on XP and 7. At least 1 thing is being activated only on 7 and not on XP. A timebomb.

      

    GenL

    20 Apr 10 at 2:35 am

  13. If nVidia deliberately damaged performance for ATI owners (with a “timebomb” lol), at least fanATIcs can finally make a decent argument as to why hate PhysX so much. I hope it blocks these ungrateful crybabies completely from playing physx games.

      

    here we go again

    20 Apr 10 at 6:21 am

  14. “Regeneration” :

    “Update: After deeper investigation, this issue has identified as a rare overflow bug. Looks like Nvidia PhysX System Software 9.10.0222 doesn’t cripple performance, but I’ll run more tests just to be sure.”

    Eran Badit, that name sounds familiar…
    http://www.techpowerup.com/index.php?65144

      

    applejack

    20 Apr 10 at 6:40 am

  15. GenL
    but you are not trying to prevent people from being misleaded
    You’re saying that to me ?
    how about people that were misleaded by original straightforward statement, all that “another bull move by nVidia” and stuff ? They will now spread that crap on forums, and mislead another users.

    What about another websites, who have picked up this news ? What about their readers ?
    I understand that NGOHQ is closer to you, but without double standarts, please.

    and
    Looks like Nvidia PhysX System Software 9.10.0222 doesn’t cripple performance
    It seems I was right – again. /sarcasm/ How is that damn possible ?!

      

    Zogrim

    20 Apr 10 at 9:27 am

  16. did you contact ngohq for their reply before you posted benchmarks with different hw specification and conspiracy theories?

      

    Ron

    20 Apr 10 at 11:40 am

  17. Ron
    did you contact ngohq
    Yes, I’ve asked confirmation in comments to that article.

    posted benchmarks with different hw specification
    I assumed benchmarks are legetimate, because ATI CPU PhysX = NV CPU PhysX. It always was. And claim like “it was 20 fps now it is 5″ is not a proof.

    Where is a confirmation to that 20 fps statement, btw ? If 5 fps was a rare bug, than it can be easily avoided.

    Moreover, just to prevent question like your, I’ve added two explanations. I was not saying that ngohq was wrong, I was saying that I can’t reproduce it on my system.

    Thus, since it was “overflow bug”, than explanation one occur only when non-Nvidia GPU is spotted in the system is incorrect, explanation two – has provided not very accurate information.. without malicious intent is correct.

    conspiracy theories?
    The only conspiracy theory here was that CPU PhysX with ATI GPU is not equal to CPU PhysX with NV GPU – and it was not mine.


    Upd: oh, I see your point. I was supposed to spend several hours on testing, and them send results silently to ngohq ?
    Maybe.. but this is against my views a little. If you have posted something – be ready to face the music. My apologies.

      

    Zogrim

    20 Apr 10 at 12:20 pm

  18. Zogrim
    Even if the original statement was misleading, why counter it with another possible-misleading article? Isn’t that a bull movie too? If you are trying to keep the fresh news coming, you shouldn’t go that far and just fight one fud with another fud.

    None of these news sites is closer to me, thanks for your opinion on that though. Things like this are preventing them from being added to my favs.

    >ATI CPU PhysX = NV CPU PhysX
    Performance may be the same for now, but main PhysX files already has all means to decide if you are running nvidia card or not. Let’s hope they will never use it for CPU mode, though.

      

    GenL

    20 Apr 10 at 3:02 pm

  19. GenL
    Even if the original statement was misleading, why counter it with another possible-misleading article? Isn’t that a bull movie too?
    Ok, what is your recommendation than ? What will you do in such case ?
    And, thanks, btw – different adequate opinions are always for good.. keep us learn and evolve :)

    If you are trying to keep the fresh news coming
    Actually, I was really trying to prevent incorrect statement to propagate futher.
    Maybe realisation was not so brilliant, but intentions were good.

    another fud
    C’mon)

    Let’s hope they will never use it for CPU mode, though.
    Doing this is just pointless, as it would reduce PhysX SDK usage by devs, and that is not what NV wanted.

      

    Zogrim

    20 Apr 10 at 3:34 pm


Leave a Reply

*
Copyright © 2009-2014. PhysXInfo.com | About PhysXInfo.com project | Privacy Policy
PhysX is trademark of NVIDIA Corporation