Jump to content
Dante Unbound: Share Bug Reports and Feedback Here! ×

Performance and the Future of PhysX Effects


Jengerer

Recommended Posts

1 hour ago, cyborgmaster said:

Nvidia post new driver 381 and update for phyX !!!!!

Interesting mister the developper. No ?

There is really a new Physx driver 9.17.0329 .Devs does it fix the problem?

Link to comment
Share on other sites

I've had driver version 381.65 tested on 3 separate systems and there doesn't appear to be any leaks on this version, so that may be a bit of good news for possibly re-enabling PhysX Effects. If anyone would like to try and update their drivers and test on their own system, you may follow the instructions for this sample application (use at your own risk and all that!) and report your findings here.

Link to comment
Share on other sites

On 2017-04-07 at 12:12 AM, [DE]Jengerer said:

I've had driver version 381.65 tested on 3 separate systems and there doesn't appear to be any leaks on this version, so that may be a bit of good news for possibly re-enabling PhysX Effects. If anyone would like to try and update their drivers and test on their own system, you may follow the instructions for this sample application (use at your own risk and all that!) and report your findings here.

using it and well I see no issues with it it seems fixed ^^

Link to comment
Share on other sites

47 minutes ago, [DE]Jengerer said:

I've had driver version 381.65 tested on 3 separate systems and there doesn't appear to be any leaks on this version, so that may be a bit of good news for possibly re-enabling PhysX Effects. If anyone would like to try and update their drivers and test on their own system, you may follow the instructions for this sample application (use at your own risk and all that!) and report your findings here.

Just ran the Leaky test you mentioned here on my Nvidia GT 740M (yes I will get a new computer soon). On the previous GE Force driver, windows task manager told me that the commited task memory increased by 10-20 MB every second for a few minutes. I then installed 381.65 and found that the commit memory size did not increase at all after 5 minutes. Thanks for the nice find Jengerer!

Link to comment
Share on other sites

1 hour ago, [DE]Jengerer said:

I've had driver version 381.65 tested on 3 separate systems and there doesn't appear to be any leaks on this version, so that may be a bit of good news for possibly re-enabling PhysX Effects. If anyone would like to try and update their drivers and test on their own system, you may follow the instructions for this sample application (use at your own risk and all that!) and report your findings here.

I've been running 381.65 since release on a medium grade system and a high grade system, without noticing any true issues for Warframe or more highly resource intensive games.

Medium Grade system specs:

  • i7-4790k
  • GTX 970
  • 16 gigs DDR3
  • Win10 (all concurrent updates)

High Grade specs:

  • i7-6900K
  • GTX 1080 SLI (x2)
  • 64 gigs DDR4
  • Win10 (all concurrent updates)
Link to comment
Share on other sites

Same here, @[DE]Jengerer. Tennos replies before this post of Nvidia Driver 381.56: PhysX       From: NVIDIA

***********************************************************

Software Module Versions

• NView - 148.47

• HD Audio Driver - 1.3.34.26

• NVIDIA PhysX System Software - 9.17.0329

• GeForce Experience - 3.4.0.70 • CUDA - 8.0

************************************************************

I swear, there was some thing about memory leak, but it was indirect. Well, I do hope this get fix for the PhysX players who enjoys them.

Link to comment
Share on other sites

11 hours ago, [DE]Jengerer said:

I've had driver version 381.65 tested on 3 separate systems and there doesn't appear to be any leaks on this version, so that may be a bit of good news for possibly re-enabling PhysX Effects. If anyone would like to try and update their drivers and test on their own system, you may follow the instructions for this sample application (use at your own risk and all that!) and report your findings here.

Like to help with it. But i get this result:

7lp3sqih.jpg

Any Suggestions?

Link to comment
Share on other sites

1st run:

20000 frames

Commit: stayed at 197MB for the entire duration

Working set:started at 113MB until around frame 1000, then increased to114MB and stayed there for the rest of the duration.

2nd run:

40000 frames

Commit: stayed at 197MB for the entire duration

Working set:started at 113MB until around frame 1000, then increased to114MB and stayed there for the rest of the duration.

 

Gtx970 2x in SLI

Driver 381.65

Link to comment
Share on other sites

21 hours ago, Extroah said:

Please no Havok. Havok is freaking out a lot, especially in Skyrim in regards to clothes physics with collision detection (like havok cloaks stretching across the whole map or havok hair freaking out when colliding), so i can't even start and imagine how that would work out with particle effects.

uhh... Havok is one of the most widely used Physics libraries, used in countless thousands of games.
Havok isn't the issue, it's the game if physics related things are functioning poorly (Bethesda games are a complete disaster and should be banned from the industry).

any Physics library can produce terrible results.

Link to comment
Share on other sites

6 minutes ago, [DE]Jengerer said:

Okay, yeah, the sample application wasn't built with support for 1000-series cards. I've submitted a new release here that should work (I've verified on a 1080 that the leaks don't occur on 381.65, but the more data points the better).

Thanks that works.  Tested and the memory size stays fixed at Committed: 203MB and Working Set: 114MB.

Link to comment
Share on other sites

10 minutes ago, [DE]Jengerer said:

Okay, yeah, the sample application wasn't built with support for 1000-series cards. I've submitted a new release here that should work (I've verified on a 1080 that the leaks don't occur on 381.65, but the more data points the better).

Okay. It works. But till wich Frame schould i let it run?

Commited: 256 - 258 MB | Working Set: 114 MB

Link to comment
Share on other sites

1 minute ago, kAzz-GER said:

Okay. It works. But till wich Frame schould i let it run?

Commited: 256 - 258 MB | Working Set: 114 MB

Shouldn't need more than a few seconds; once the "Active Particles" count reaches 500, it should stop using more memory. Thanks for the tests, everyone. I think this bodes well for this driver version.

Link to comment
Share on other sites

19 minutes ago, [DE]Jengerer said:

Shouldn't need more than a few seconds; once the "Active Particles" count reaches 500, it should stop using more memory. Thanks for the tests, everyone. I think this bodes well for this driver version.

Active Particles starts with 500 ... oO

vxnla2my.jpg

Link to comment
Share on other sites

3 hours ago, Anubis.ft.Chacal said:

Its activated again. So are the devs  thinking about use Nvidia FleX? I just downloaded the Beta Gamengine and its awesome *-* Warframe is awesome and it would be the perfect Fusion Ha!

Flex sucks.In Killing Floor 2 its unusable.You get massive fps drops from this.

When you dont know Killing Floor 2 here is a video that demonstrates Flex very well but trust me it runs like crap.

 

Link to comment
Share on other sites

3 hours ago, K0bra said:

Flex sucks.In Killing Floor 2 its unusable.You get massive fps drops from this.

When you dont know Killing Floor 2 here is a video that demonstrates Flex very well but trust me it runs like crap.

 

Like I said, Ive tested it on gameengine. This game sucks, its not warframe. Who write the game is free to improve everything with codes. FleX is more complex than PhysX (which just create some glowing beatles with a lit of physics) and a lot of knowledge is needed...so thats why Killing Floor sucks. Just look at other videos and ull see.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

Guest
This topic is now closed to further replies.
×
×
  • Create New...