If you run a dedicated PhysX GPU, buy big...

If you run a dedicated PhysX GPU, buy big...

I think allot of people are under the impression that a 9 series card is all you would need to process PhysX, and that a 200 series card would just be overkill...

This is not the case in Cryostasis...

The hotest config to run this game, is (2) 200 Series GPU's in SLI processing graphics...
and another 200 Series GPU operating in dedicated PhysX mode.

That produces a higher frame rate than (3) card SLI, or (3) card SLI with a 9 series dedicated PhysX processor. The GPU you select for PhysX, has a large impact on frame rates you generate running the game.


In this thread, I was collecting the Cryostasis Tech Demo FPS info, for 1920x1200 RES.
http://www.evga.com/forums/tm.asp?m=568129&mpage=2

These are the posted reports I found on the net.

Updated...

Cryostasis Tech Demo:
Demo was run using the High preset...


So in summary at 1920x1200 res, this is what we have:

walterman -------------- Xeon 3350 @ 3.6GHz, High settings, X16, 1920x1200.
(1) 8800GTX. (Processing Graphics)
No Dedicated PhysX GPU
Average FPS= 22.0, Minimum FPS= 13

Whisperfang ----------- Q9450 @ 2.66GHz, High settings, X16, 1920x1200.
(1) 280 @ 602/1296/1107. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 28.3, Minimum FPS= 15.8, and Maximum FPS=84.6

Whisperfang ----------- Q9450 @ 3.01GHz, High settings, X16, 1920x1200.
(1) 280 @ 602/1296/1107. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 29.6, Minimum FPS= 16.3, and Maximum FPS=98.1
Whisperfang: "This leads me to conclude that at 1920x1200 with all options on high and hardware PhysX with a single GTX280 and a "modern" CPU that the 280 is in fact the limiting factor.
I added ~15% more CPU horsepower and ~13% more memory speed and basically got nothing in return".

Roliath ------------------ Q6600 @ 3.2GHz, High settings, X16, 1920x1200.
(2) 260's SLI Mode. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 31.6, Minimum FPS= 18.3, and Maximum FPS=75.3

Talonman -------------- Q6600 @ 3.81GHz, High settings, X16, 1920x1200.
(1) 280 @ 756/1512/1269. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 35.9, Minimum FPS= 21.1, and Maximum FPS=109.3

Bench Extreme ------ QX 9650 @ 4.20GHz, High settings, X16, 1920x1200.
(1) 285. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 36.9, Minimum FPS= 23.2

StretchMaK ------------ E4800 @ 3.6GHz, High settings, X16, 1920x1200.
(2) 260's SLI Mode. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 38.3, Minimum FPS= 22.1, and Maximum FPS=96.7

Bench Extreme ------ QX 9650 @ 4.20GHz, High settings, X16, 1920x1200.
(1) 285. (Processing Graphics)
(1) 285. (Dedicated PhysX)
Average FPS= 44.7, Minimum FPS= 26.3

Shamino --------------- i7 @ 2.8GHz, High settings, X16, 1920x1200.
(3) 280's SLI Mode. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 46.0, Minimum FPS= 17.9, and Maximum FPS=109.2

slaWter ----------------- i7 965 Extreme @ 3.84GHz, High settings, X16, 1920x1200.
(2) 280's SLI Mode. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 50.0, Minimum FPS= 29.8, and Maximum FPS=150.4

Bench Extreme ------ QX 9650 @ 4.20GHz, High settings, X16, 1920x1200.
(2) 285's SLI Mode. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 51.5, Minimum FPS= 29.9

Talonman -------------- Q6600 @ 3.81GHz, High settings, X16, 1920x1200.
(1) 295 @ 648/1512/1188. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 53.7, Minimum FPS= 30.7, and Maximum FPS=127.2

Bench Extreme ------ QX 9650 @ 4.20GHz, High settings, X16, 1920x1200.
(3) 285's SLI Mode. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 58.0, Minimum FPS= 31.0

Shamino --------------- i7 @ 2.8GHz, High settings, X16, 1920x1200.
(3) 280's SLI Mode. (Processing Graphics)
(1) 9800GT. (Dedicated PhysX)
Average FPS= 58.6, Minimum FPS= 24.9, and Maximum FPS=132.1

Dentlord ---------------- QX9650 @ 3.6GHz, High settings, X16, 1920x1200.
(3) 280's @ 685/1475/1220 SLI Mode. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 59.6.

Dentlord ---------------- QX9650 @ 4.2GHz, High settings, X16, 1920x1200.
(3) 280's @ 685/1475/1220 SLI Mode. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 62.5.

Krog --------------------- i7 920 @ 3.9GHz, High settings, X16, 1920x1200.
(3) 280's @ 720/1440/1300 SLI Mode. (Processing Graphics)
No Dedicated PhysX GPU.
Average FPS= 64.4.

freakysqeeky: System was at: Q9550 @ 4.0GHz, High settings, X16, 1920x1200.
(1) 295 set to C=660, SH=1423, and M=1210. (Processing Graphics)
(1) 9800GTX+ set to C=790, SH=1977, and M=1160. (Dedicated PhysX)
Average FPS= 65.5, Minimum FPS= 32.2, and Maximum FPS=135.4

Bench Extreme ------ QX 9650 @ 4.20GHz, High settings, X16, 1920x1200.
(2) 285's in SLI Mode. (Processing Graphics)
(1) 285. (Dedicated PhysX)
Average FPS= 75.1, Minimum FPS= 32.4

Talonman: System was at: Q6600@ 3.81GHz, High settings, X16, 1920x1200. (Updated entry to match freakysqeeky's 295 clock settings.)
(1) 295 set to C=660, SH=1423, and M=1210. (Processing Graphics)
(1) 280 set to C=702, SH=1512, and M=1188. (Dedicated PhysX)
Average FPS= 76.7, Minimum FPS= 35.9, and Maximum FPS=139.0

Talonman: System was at: Q6600@ 3.81GHz, High settings, X16, 1920x1200.
(1) 295 set to C=702, SH=1548, and M=1188. (Processing Graphics)
(1) 280 set to C=756, SH=1512, and M=1296. (Dedicated PhysX)
Average FPS= 78.5, Minimum FPS= 36.2, and Maximum FPS=140.4
____________________________________________________________________________________________

So in summary at 1680x1050 res, this is what we have:

RMXO: ------------ Q6600 @ 3.6GHz High settings, X16, 1680X1050.
(1) 285 set to C=730, SH=1615, and M=1375. (Processing Graphics)
No dedicated PhysX GPU.
Average FPS= 44.9, Minimum FPS= 22.8, and Maximum FPS=99.3

RMXO: ------------ i7 920 @ 4GHz High settings, X16, 1680X1050.
(1) 285 set to C=740, SH=1650, and M=1425. (Processing Graphics)
No dedicated PhysX GPU.
Average FPS= 53.2, Minimum FPS= 33.1, and Maximum FPS=158.5



Blazemore: -------- Q6600 @ 3.6GHz High settings, X16, 1680X1050.
(1) 295 set to C=648, SH=1368, and M=1242. (Processing Graphics)
(1) 9600 GSO (sp 96) set to C=550, SH=1350, and M=1600. (Dedicated PhysX)
Average FPS= 70.5, Minimum FPS= 30.3, and Maximum FPS=152.3

Blazemore: -------- Q6600 @ 3.6GHz High settings, X16, 1680X1050.
(1) 295 set to C=648, SH=1512, and M=1242. (Processing Graphics)
(1) 9600 GSO (sp 96) set to C=700, SH=1750, and M=1800. (Dedicated PhysX)
Average FPS= 72.5, Minimum FPS= 31.3, and Maximum FPS=157.0


Talonman: -------- Q6600 @ 3.81GHz High settings, X16, 1680X1050. (Ambient Occlusion turned on.)
(1) 295 set to C=648, SH=1512, and M=1188. (Processing Graphics)
(1) 280 set to C=756, SH=1512, and M=1269. (Dedicated PhysX)
Average FPS= 75.9, Minimum FPS= 34.3, and Maximum FPS=125.6

Talonman: -------- Q6600 @ 3.81GHz High settings, X16, 1680X1050. (Ambient Occlusion turned 0ff.)
(1) 295 set to C=648, SH=1512, and M=1188. (Processing Graphics)
(1) 280 set to C=756, SH=1512, and M=1269. (Dedicated PhysX)
Average FPS= 82.7, Minimum FPS= 35.8, and Maximum FPS= 168.8
____________________________________________________________________________________________


Note that @ 1920x1200, freakysqeeky was running a 295, and a 9800GTX+ for PhysX. This config generated:
Average FPS= 65.5, Minimum FPS= 32.2, and Maximum FPS=135.4


Note that I am running a 295 at the exact same speed, and a 280 for PhysX. My config generated:
Average FPS= 76.7, Minimum FPS= 35.9, and Maximum FPS=139.0


If a 9800GTX+ for PhysX is truly all we would ever need, our numbers should be closer.

I think the dedicated PhysX processor we selected, is the only reason I am getting a better FPS out of this game.

If you are thinking of getting a GPU to run dedicated PhysX on, i say buy big.

I dont know if Cryostasis will be one of a kind with it's large PhysX workload, but probably won't be? I expect the processing power it takes to give a PhysX game the max performance, is greater than some initially may think.

nice lil comparison.

but does nobody compare a true phys-x card to these setups? i dont play any phsyx games yet, but ive got a 9500gt(ddr3) and a dedicated bfg ageia phys-x card too...howd that scale against such things

I havent seen the Ageia card used in the benchmark. I would love to.

Most feel the ageia card's processing power is in the same ball park as an 8 series card...

I am not 100% sure on that, but think a 280 would smoke an official ageia PPU.

P3N1X0R said:

nice lil comparison.

but does nobody compare a true phys-x card to these setups? i dont play any phsyx games yet, but ive got a 9500gt(ddr3) and a dedicated bfg ageia phys-x card too...howd that scale against such things

Click to expand...

TalonMan said:

I havent seen the Ageia card used in the benchmark. I would love to.

Most feel the ageia card's processing power is in the same ball park as an 8 series card...

I am not 100% sure on that, but think a 280 would smoke an official ageia PPU.

Click to expand...

It gets obliterated by a gpu, not even on par with an 8800gt

I'll look for the thread with the info

Thanks...

That sounds correct to me.

Since when is informing some one of the correct place to post something crapping on a post, this is about physics processing so why that section?

I am full well aware of the PhysX section, but this is more about how fast of a GPU to select for running PhysX.

I don't wan't to debate if you like PhysX, or if PhysX has a future.. It does.

So many guys post in the GPU section, 'What card should I buy to run as a dedicated PhysX processor?"
I think the most common answer that has been given is: Around an 8 series card should do. Anything else would be overkill.
I don't believe that to be the best answer anymore. This info, being in the GPU section, will help get the word out.

I opted to go with the GPU area with more traffic.

Lets stay on topic.

So, I got a 2nd GTX280.

So what you are saying is that fr Physx games i am better off not SLIing and using the 2nd GTX 280 for Physx?

Is this true for all Physx games or just Cryostasis?

What about Mirror's Edge? Any info there?

Either way, thanks for all the info.

StretchMaK ------------ 2x 260's SLI Mode --- E4800 @ 3.6GHz
Average FPS= 38.3, Minimum FPS= 22.1, and Maximum FPS=96.7

Extreme Bench ------------ GTX 285 + GTX 285 for PhysX, QX 9650 @ 4.20GHz
Average FPS= 44.7, Minimum FPS= 26.3

It might be close, but my money would be on the dedicated PhysX rig.

I only have quality Cryostasis info so far, but think a PhysX card helps more than running 1 more card in SLI...

If this will continue when some of the big boy game makers, offer their first PhysX games up, is hard to say. But I bet it will.
We already know a game patch is going to be released, that will increase the amount of PhysX in this game.
We can also assume so will the PhysX workload...

Again, another reason to buy big!

The fact that my system can generate more FPS than an i7 rig, runiing (3) 280's in SLI, along with a lesser dedicated physX processor, was rather shocking to me....

Shamino ------------ Tri-280's, 9800GT dedicated PhysX GPU, i7 @ 2.8GHz
Average FPS= 58.6, Minimum FPS= 24.9, and Maximum FPS=132.1

I did note in my other thread, the Tech Demo uses all 4 cores of my Q6600.
It appears to me, the actual game only uses 1 core...

Demo...


Game...


It looks as though games that run lot's of PhysX, at 1920x1200, we should give them all the PhysX processor we can get our hands on.
Also consider changing our standard answer on what card people should get for a dedicated PhysX processor.




Also note that if you tried using a 295 for dedicated PhysX, it would only use 1/2 of the 295.
PhysX does not support SLI, so both sides of your 295 could not be used to calculate PhysX.

The 280's and 285's are the fastest PhysX processors we have. (For now...)

If the trend you discovered continues in other games and tomorrows games, then I'm down.

Yeah Man!

If any of you SLI boys decide to test this out, and run one card in dedicated PhysX mode, post back...

I would like to know how things turned out.

If I didn't fold 24/7, my main rig has SLI'd 280s and a 285 which could be used in dedicated PhysX.
Maybe I'll check out the game but it's graphics don't excite me from YouTube videos I've seen.

Kowan said:

If I didn't fold 24/7, my main rig has SLI'd 280s and a 285 which could be used in dedicated PhysX.
Maybe I'll check out the game but it's graphics don't excite me from YouTube videos I've seen.

Click to expand...

The video's look crap compared to the actual game....especially the dynamic water...

I heard that the original physx card is supposed to be the equivalent to an 8500GT 256MB card.

Wow, things have changed quite a bit since then, lol.

Kowan said:

If I didn't fold 24/7, my main rig has SLI'd 280s and a 285 which could be used in dedicated PhysX.
Maybe I'll check out the game but it's graphics don't excite me from YouTube videos I've seen.

Click to expand...

I bet OC'ed SLI'ed 280's, running with a OC'ed 285 dedicated PhysX GPU could put some big numbers up...

thanks for doing the legwork for us on this one talon. you got plans to compare any other games, at least between 2 sli and 1 sli 1 physx? it would be cool to see if this is true across a few different titles. also do you think the terribly low clock on the i7 had anything to do with shaminos low numbers? i mean he does have 3x280s and a 9800gt. thats a lot of gpu power. and 1GHz cpu speed is a big difference.

Your welcome.

I would like to know what shamino clocked his GPU’s at… That might be the biggest factor?

The CPU might be in on the deal too. Hard to say on why shaminos score is a bit lower.

Shamino ------------ Tri-280's, 9800GT dedicated PhysX GPU, i7 @ 2.8GHz
Average FPS= 58.6, Minimum FPS= 24.9, and Maximum FPS=132.1

Dentlord with 3x GTX 280 @ 685/1475/1220...air...none...qx9650 @ 3.6GHz...water produced:
Average FPS= 59.6.

Dentlord with 3x GTX 280 @ 685/1475/1220...air...none...qx9650 @ 4.2GHz...water produced:
Average FPS= 62.5.

Krog with 3x GTX 280 @ 720/1440/1300 ...water...none... i7 920 @ 3.9GHz...water produced:
Average FPS= 64.4.

Still pritty close.

I do have a few shots of Mirrors Edge, and the FPS i'm getting with a dedicated PhysX processor in my other thread:

http://www.evga.com/forums/tm.asp?m=568129&mpage=2

but we still need more testing.

you looking for tests done on a 3.5GHz Pentium D and a single 9800GT? ok i didnt think so nvm.

yeah if krog gets 6 frames more with the same 3x280s without any physx card, then there must be something between the cpu and gpu clocks that accounts for the discrepancy. krog needs to get a 260 for physx and then post results. 70fps maybe??

OK, so here is a question..
I have an 8600GTS just sitting around doing nothing..is it even worth it to put it in as a physics card?
or would it make no difference at all and just waste space and make heat...

Red Falcon said:

I heard that the original physx card is supposed to be the equivalent to an 8500GT 256MB card.

Wow, things have changed quite a bit since then, lol.

Click to expand...

Thats Hourse Hocky......I runt he orig PhysX card with my 8800GTX's (SLI'ed) and I still notice in UT3 that the orig Physx card runs circles around any Video card dedicated to Physx. IMHO.....

Crosshairs said:

OK, so here is a question..
I have an 8600GTS just sitting around doing nothing..is it even worth it to put it in as a physics card?
or would it make no difference at all and just waste space and make heat...

Click to expand...

try it out! takes what, 5 minutes to install? give it another half hour or so to get it set up as physx. run some tests, play some games, see if you get a performance boost. if not, take it back out. i would say yes it should make a difference, depending on how old your video card(s) is(are) it may even be substantial. youll have to decide for yourself if its worth the extra heat and power.

Crosshairs said:

OK, so here is a question..
I have an 8600GTS just sitting around doing nothing..is it even worth it to put it in as a physics card?
or would it make no difference at all and just waste space and make heat...

Click to expand...

ORIGINAL: Shamino said:

Ok so let's see if dedicated physx card is of any help!!!

Test setup:

Core i7 @ 2.8GHz
3x2GB Rams
EVGA X58 Classified E759
3 x EVGA GTX280s - Triple SLI
EVGA 9800GT
Cryotasis Demo Benchmark - All details maxed out, 16xAF, 1920x1200


PhysX Enabled on GTX 280s - Non dedicated scenario:
Non Dedicated:
Average FPS - 46.0
Minimum FPS - 17.9
Maximum FPS - 109.2


PhysX Enabled on 9800GT - Dedicated PhysX scenario:
Dedicated:
Average FPS - 58.6
Minimum FPS - 24.9
Maximum FPS - 132.1


Dedicated PhysX gave ~ 27% boost in FPS.

Click to expand...

A 9800GT gave Shamino a 27% boost in speed on his i7 rig, I bet an 8600GTS would help yours too.

But for max performance, the bigger the better on dedicated PhysX processor.

TalonMan said:

A 9800GT gave Shamino a 27% boost in speed on his i7 rig, I bet an 8600GTS would help yours too.

But for max performance, the bigger the better on dedicated PhysX processor.

Click to expand...

OK, Thanks for the reply.Im running a single 260 GTX right now.....so Ill stick the 8600 in and see what happens...

I broke down my SLI , but maybe Ill stick the other 260 back in for physics if Im not happy with the 8600

Crosshairs said:

OK, so here is a question..
I have an 8600GTS just sitting around doing nothing..is it even worth it to put it in as a physics card?
or would it make no difference at all and just waste space and make heat...

Click to expand...

Worth it for what? You can't even play the tech demo. I don't know what is the point of this thread. To see what difference a GPU can make in a tech demo? It is as worthless as getting a higher score in 3DMark.

Back for another go?

If you check the link I gave, it also has my screen shots, and the FPS I generate playing the actual game...

They are in line with the Tech Demo FPS, so the info provided by the benchmark is also valid for the game.

What FPS were you getting on the Demo? Give us your stats to help the discussion.

Again, the point of this thread is to provide the new information that to get max performance in PhysX games, you do need more than an 8 series dedicated PhysX GPU.

This has not been the opinion of most people.

Sending a message to a mod to watch this thread... (Dr. Evil notified.)

alg7_munif said:

Worth it for what? You can't even play the tech demo. I don't know what is the point of this thread. To see what difference a GPU can make in a tech demo? It is as worthless as getting a higher score in 3DMark.

Click to expand...

Sigh. The playable demo has been out for three weeks now.....and you know the actual game too?

trinibwoy said:

Sigh. The playable demo has been out for three weeks now.....and you know the actual game too?

Click to expand...

The game is out, the playable demo is out but the data provided in this thread are from the tech demo, which is as useless as the scores from 3DMark.

TalonMan said:

These are the posted reports I found on the net.

Cryostasis Tech Demo:

So in summary at 1920x1200 res, this is what we have:

Click to expand...

The title of this thread: "If you run a dedicated PhysX GPU, buy big..."
You should buy big based on the data from a tech demo? It is like saying you should buy a GPU based on its 3DMark score or based on canned benchmarks using cutscenes.

alg7_munif said:

The game is out, the playable demo is out but the data provided in this thread are from the tech demo, which is as useless as the scores from 3DMark.

Click to expand...

So what? Do you have a smidgen of evidence that the actual game shows different scaling characteristics from the techdemo? Have you done the comparison yourself? What exactly is your problem?

I don't agree that you should run PhysX on your GTX 285 just for one game but your criticism of the data provided here is just noise and doesn't contribute anything useful. Like Talonman suggested, go run your own comparison of demo vs techdemo and let us know if they scale differently.

+1

Thank you...

I sense much envy here...*presses report button*

Keep the results comming TalonMan and don't let a troll derail you

Thank you sir.. I intend to.

alg7_munif said:

Worth it for what? You can't even play the tech demo. I don't know what is the point of this thread. To see what difference a GPU can make in a tech demo? It is as worthless as getting a higher score in 3DMark.

Click to expand...

What are you talking about?
You are the only one who does not see the point of this thread...please, if you have nothing constructive to add, just stay out of it.

trinibwoy said:

So what? Do you have a smidgen of evidence that the actual game shows different scaling characteristics from the techdemo? Have you done the comparison yourself? What exactly is your problem?

I don't agree that you should run PhysX on your GTX 285 just for one game but your criticism of the data provided here is just noise and doesn't contribute anything useful. Like Talonman suggested, go run your own comparison of demo vs techdemo and let us know if they scale differently.

Click to expand...

Using 4 cores in the tech demo but only one core in the game shows that the game is not the same as the tech demo.

TalonMan said:

I did note in my other thread, the Tech Demo uses all 4 cores of my Q6600.
It appears to me, the actual game only uses 1 core...

Click to expand...

And yet, the minimum FPS I get in the benchmark is 36 FPS, and the lowest FPS I have seen in the game is 36 FPS.

Crosshairs said:

What are you talking about?
You are the only one who does not see the point of this thread...please, if you have nothing constructive to add, just stay out of it.

Click to expand...

If you want me to be constructive, I can be constructive. TalonMan said that this thread is intended to inform those people who ask "What card should I buy to run as a dedicated PhysX processor?".

TalonMan said:

So many guys post in the GPU section, 'What card should I buy to run as a dedicated PhysX processor?"
I think the most common answer that has been given is: Around an 8 series card should do. Anything else would be overkill.
I don't believe that to be the best answer anymore. This info, being in the GPU section, will help get the word out.

Click to expand...

The problem is, he came up with the conclusion "If you run a dedicated PhysX GPU, buy big..." but he only gave the data from a single game and he is not even using the real world gameplay data. He just gave the data from a tech demo.

It would be much better if he can give the data from the actual gameplay to prove his point. Giving the data from other games too would also be useful. I'm not asking for a full gameplay evaluation like what [H] do but I've been around here long enough, I just can't accept it when someone tells me to buy a specific graphic card because the data show that the card is faster during a cutscene of a game.

i think the main point here isnt that you need to have a 200 series card for physx or go home. his point is simply that there exist some games where you could actually make use of something more. everyone thought "well as long as you have an 8 series, upgrading your physx wont make a difference because it is just overkill." talon is trying to show that this may not actually be the case, as his research seems to show that a 200 series as physx could possibly give a better fps than an 8 series SOMETIMES in SOME GAMES POOSSIBLY. all hes saying is the common knowledge on this one seems to be wrong, and we should be aware of that. further testing to solidify his hypothesis is surely on the way. right talon?

ekuest said:

i think the main point here isnt that you need to have a 200 series card for physx or go home. his point is simply that there exist some games where you could actually make use of something more. everyone thought "well as long as you have an 8 series, upgrading your physx wont make a difference because it is just overkill." talon is trying to show that this may not actually be the case, as his research seems to show that a 200 series as physx could possibly give a better fps than an 8 series SOMETIMES in SOME GAMES POOSSIBLY. all hes saying is the common knowledge on this one seems to be wrong, and we should be aware of that. further testing to solidify his hypothesis is surely on the way. right talon?

Click to expand...

In some games or in a tech demo of a game?

The problem is, his data doesn't show whether a faster card would make a difference for the gameplay or not. If you look at the Cryostasis settings, you will see that you can only turn the hardware physics setting on or off, you can't pick low, medium or high. In this case, it is still unclear whether you need a 200 series card to use PhysX or you should just go home (don't use PhysX) because the data provided is from a tech demo.
http://img297.imageshack.us/img297/9641/cryo1.jpg

The Cobra said:

Thats Hourse Hocky......I runt he orig PhysX card with my 8800GTX's (SLI'ed) and I still notice in UT3 that the orig Physx card runs circles around any Video card dedicated to Physx. IMHO.....

Click to expand...

i know it runs circles around my 9500gt ddr3 model too.....yet when i run the nvidia phsy-x screensaver, and run phys-x on the 9500 it actually performs better than my dedicated phys-x card....

well if it CAN happen in a tech demo, i assume it CAN happen in a real game. perhaps the real cryostasis will not prove to perform better with a 200 series on physx, but the data still suggests that better physx performance CAN be had with a better than 8 series card. and if it CAN happen in a tech demo, i would say it WILL happen in a full game soon enough, whether or not that game is cryostasis.