Discussion in 'Games, Gaming & Game-demos' started by krisha, Sep 4, 2008.
Patch has been released for the game. Offering many single player fixes, performance optimizations and PhysX improvements. There are two versions of the patch, depending on which publisher you bought the game from.
If your copy was published by 505 Games download this patch. If your copy was published by Aspyr, download this patch.
I found it at my local Target store for $7. Is it worth it? Should I buy?
I love it. It runs like a complete dog on anything less than a GTX 260 and a E/Q6600, but it is very claustrophobic, scary in places and very "Eastern Bloc".
Yeah for $7 def
I'm going to dig up this thread for hopefully, some confirmation on performance after the 1.1 patch. I recall trying the demo on my GTX285, and ended up getting terrible performance whenever any PhysX events started, which in the demo mostly consisted of melting ice when turning on various heat sources. The rest of the time the game ran just fine, and I enjoyed what I could play.
So with the patch mentioning performance improvements specifically, is there anyone that can confirm the patch fixing performance during these PhysX/shader moments? I'd love to buy the game, but am not based simply on performance, would be a shame to pass on it if the patch indeed fixes the issue.
When i played it i found a physx glitch,.. you need to disable it on your gpu and enable it in game.. kinda strange but if i didn't to that, i got 20fps max.
Its the opposite like in Mirrors Edge where you need to enable it on your gpu first, so it works in the game later..
The "PhysX" or 1.1 (Not to be confused with the first 1.01) patch improved the PhysX implementation quite a bit but it's still a very demanding game as they also increased the effect quality, still a GTX285 should have no problems with the game, bit CPU limited though but it should run fine.
(The PhysX effects aren't some side effects or extras here and really enhance the game, still heavily used even if disabled as it runs via CPU then which takes quite a hit even without extras like water and ice particle effects.)
(It's used everywhere after all, water and fluids like the ice, cloth and frost effects and things like character ragdolls.)
Probably one of the better PhysX implementations of what it can do outside of pure PhysX titles like Cell Factor or that other game which is all PhysX enabled.
(But too old for CUDA functionality and recent improvements via APEX are far better looking and realistic like say Arkham Asylum or the particle effects in Dark Void.)
So your saying the PhysX run better in software mode than hardware? Maybe I'll re download the demo and try that.
Anyone else running the full game with PhysX/sm4.0 just fine with, or without this method?
:Edit: Ah Jonas got a post in there.
Well, it's sounding like the full game has improved then, so maybe I shouldn't worry about it. I certainly have no doubts that it is a demanding game, and don't expect to be getting a constant 60 frames per second, but I was experiencing horrible micro stuttering combined with frame rates in the teens whenever the PhysX kicked in, so it made it nearly unplayable.
Thanks for the feedback.
Aside from a few levels that ran poorly I worked trough the game with a AGEIA PPU and ATI card which wasn't really that optimized driver wise at the time but performance was fine (Around 30 or so, not fantastic or anything but playable.) so with something like a GTX260 or even a dedicated card for PhysX it should run very fluidly.
I couldn't get the game to run properly with 2x 5770s and an overclocked 9600GT.
Ah, you had a dedicated PhysX card, I guess I can't be so sure about my 285 then.
I've always found PhysX to run pretty badly on anything but a dedicated card. I've been through an 8800GTX, GTX260, and then stepped up to the GTX285 I have now and experienced horrible frame skipping and mouse lag in PhysX games on all 3 cards. The platform is just flawed without a dedicated card it seems.
It should run better on a proper NVIDIA card even if it's used for both the game and physics than it should on a ATI/PPU combination.
Kinda hard to say without a demo or something though but there wasn't one for the updated version as far as I know.
(PPU is comparable to a 9600 GT but slower as it has to rely on the CPU and also the GPU, it's kinda dated and sorta dead by now though, haha.)
The demo seems to be somewhat updated, as it has the water cannon that was added in 1.1, but I can't be sure if it includes the performance related aspects of the patch. I hope it doesn't anyway. :3eyes: Guess I'm still undecided unfortunately.
If it has the water cannon it should be built upon the 1.1 patch, wasn't aware that there was a updated demo available, perhaps it got released alongside the games US release. (Or perhaps just before with the promo by EVGA when buying the GTX200 series of cards gave you a code for the game.)
Base game was a bit unoptimized (Also partly due to non-optimized Catalyst drivers for it, came much later.) but it ran pretty well after the patch. (Still not really perfect but better than it used too and again that was with a ATI card and a AGEIA PPU.)
What PhysX drivers are you using and are you certain the demo uses GPU PhysX acceleration, should be able to open a options.ini file in the game directory and it should say something like hardware_physics=1 if it's enabled, updated Forceware drivers have the option of showing a on screen message wether PhysX uses the CPU or the GPU as well so that might be useful to check to verify that it indeed uses CUDA acceleration via the GPU and not the far slower CPU.
(Trying to run the game in "software" with physics enabled won't really run well at all, CPU can't handle all those effects and calculations in addition to the main game CPU routines ro how to describe it in a good way.)
I suppose I never checked it with the ForceWare overlay, but I remember checking out the ini file, and everything looked to be in order. As far as driver versions, I remember trying a couple with the same results.
For the PhysX driver I was mostly thinking that the new interface (Latest version.) that changes it from C:\Program Folder\Ageia to \NVIDIA could have interefered a bit with older titles but it shouldn't.
(The game would not work at all if it couldn't load the PhysX binaries.)
Might be a good idea to rename any local PhysX files in the game directory however and let the latest version take over. (NX*.dll should be left alone but PhysX*.dll files can be renamed, PhysXLoader.dll for example might interfere.)
INI is probably fine as well, however as I said if it can't initiate PhysX via CUDA or rather the GPU all the effects are handled by the CPU and as shown in many tests and benchmarks it's nowhere near fast (Even for i7 975's overclocked.) enough to handle all those effects along with the main game so you'll be left with single digit framerates most of the time, hah. (For the more demanding effects at least, cloth works fine via CPU as does ragdoll but fluids and particles, well fluids are also particle sorta but whatever, will really hit performance due to the massive amount of them used for the effects and that they can interact with each other and everything else.)
EDIT: To clarify a bit with PhysX loader or PhysX device .dll files older versions would only hold information about the PPU and not CUDA capable GPU's so it would not initialize properly, sorry for being a bit vague with the previous description.
(I'm no PhysX expert but there's a lot to read about it and seeing how the hacks and mods for it work you learn a bit.)
I've simply decided i'm gonna wait till i got a GTX 480/470 to buy this game, it's cheap on steam now for 12.99 too, but like others i'm afraid of the performance... lol
I have indeed done some of the .DLL rerouting in a couple older titles, but never tried it in the Cryostasis demo. With it being such a recent title, and one touted specifically as an NV PhysX title I would be surprised if it was looking for a PPU, but I'll give everything a shot when I download the demo again.
As far as CPU running the PhysX, I do know just how bad they are at it compared to GPU's in theoretical circumstances, but NV's implementation seems to be flawed to a rather extreme degree. Even blowing up a vehicle in GRAW2 will literally pause the game for half a second while it renders all of maybe 30 pieces of debris (verified on multiple peoples PC's).