My Test of Cell Factor running with and without physX

lowrider007

Weaksauce
Joined
May 16, 2006
Messages
72
oh dear again, at first when I fired up Cell Factor I honeslty thought wow, this looks really nice, maybie Agiea have got it right after all, then I started hearing rumers on the net that that people were playing this game without the PPU :eek: , ummmmmmm, it clearly states on Ageia's website,
"Min. Requirements:
AGEIA PhysX Accelerator (Yes, it is required!)"

I have finally got around to testing Cell Factor with and without PPU,

Without PPU 1280x1024 8xAF HDR Using High Graphics Shortcut : 13 FPS


with PPU 1280x1024 8xAF HDR Using High Graphics Shortcut : 15 FPS


-------------------------------------------------------------------------------------------

Without PPU 1024x768 8xAF HDR Using High Graphics Shortcut : 13 FPS


With PPU 1024x768 8xAF HDR Using High Graphics Shortcut : 12 FPS


-------------------------------------------------------------------------------------------

Without PPU 800x600 8xAF HDR Using High Graphics Shortcut : 13 FPS


With PPU 800x600 8xAF HDR Using High Graphics Shortcut : 12 FPS


-------------------------------------------------------------------------------------------

Without PPU 640x480 8xAF HDR Using High Graphics Shortcut : 13 FPS


With PPU 640x480 8xAF HDR Using High Graphics Shortcut : 19 FPS


Pretty much the same.
 
sing with me!


ta da dum dum dum, another card bites the dust.... ta da dum dum dum... another one bites the dust



seriously, this is highly dissapointing :(
 
thats pretty funny that it is exactly the same. pretty disappointing if you ask me.
 
LOL

I was thinking that the game played fairly decently without a PPU this weekend. It looks like Cell Factor is a horrendous dog with or without a PPU.

The one thing I wanted to test was CPU utilization to see if the software Ageia physics engine was working as hard as it should, i.e. CPU limited.

Too bad the PPU doesn't come with a fun unit inside. Cell Factor bored me to death in less than an hour. Even my nephews didn't want to play it.
 
I think the water mark is there because it is an Ageia sponsored game. They probably thought, "wow, wouldn't it be great if our logo was on every screenshot?"

Try taking the card out of the system and running the tests again. I wonder if there is a bug that enables PhysX if the card is present, regardless of the settings.
 
What would be funny is if the PPU isn't being used at all and the CPU is doing everything. Looking at these screens it's not outside the realm of possibility.

3 cheers for the $300 paperweight!
 
Chris_Morley said:
I'd like to know the same.
It's always there.

cf7ch.png


I just tested CF without a PPU with all graphics options at minimum (why it looks ugly above), 1024x768. Like I suspected, the Ageia software physics engine isn't trying very hard:

badcf2ro.png


The red section is initial loading + a very short botmatch game. There is a constant amount of explosions and debris going on. The blue section is a single player reload and 3 distinct physics actions 10 seconds apart: 1 grenade with lots of debris, 1 psychic push of many boxes and 1 gravity grenade on many objects.

The framerate is still terrible even though the game isn't CPU or video card limited. IMO, there's another tick limiting speed. And more wildly i'd guess that the tick favors the PPU when the game is run without a PPU on fast CPUs. All IMO, of course.
 
First. How did you get cellfactor to work, without the ppu?

Second. Ive learned not to believe everything that people say. Some people really get a kick out of making other people miserable. So I have learned to take what people say with a grain of salt.

Thru the years of building systems, and always looking forward to advancements in hardware. Everytime a new devise is about to come out. There is always a group of people who are on the negative side, no matter what the devise is. I can remember so many things thru the years that people have said. Every devise that has come out, there has always been a group to knock it down. Maybe the PPU is not the way of the future, and just maybe it is.

I for one am glad the advancements that have been made. And look forward to the future. I have bought the ppu, and really have not noticed any degrading value in my framerates. And look forward to future games using it. If it doesnt work out, then there will be another devise to look forward to installing.
 
Dusty said:
First. How did you get cellfactor to work, without the ppu?
It's trivial: just add EnablePhysX=false in the shortcut.

I would also be skeptical, but all you have to do is test it yourself. I'm pretty much convinced that Ageia is sandbagging software physics performance after looking at the CPU utilization. And I have little to disbelieve about lowrider007's performance claims after playing CF in software physics mode myself.

(edit) I gave the instructions above, but here's a link: http://www.hardforum.com/showpost.php?p=1029430929&postcount=2
 
I tried running the demo (without a PPU) on the main system in my sig. It performed like an absolute DOG. Framerate was terrible with all graphics at their medium settings.

It was odd though, because the FPS would go up and down dramatically. If I shot a grenade in a pile of stuff, the FPS would dip, then resume normal speeds while the stuff flew through the air.

I was disappointed to see CPU utilization never hit 100% on both of my "CPUs". The PhysX engine was supposed to be highly threaded. I saw no evidence of SMP support.
 
What! a gpu bottleneck can mask the effects of a cpu bottleneck?!

Now compare a celeron to a conroe in cellfactor ;o)
 
Wow that's just terrible....

I had a feeling something like this would happen... what was Ageia thinking? Why did they release so early without anything that takes advantage of the card?

Or do you think the card will always be useless?
 
Try it with this:

DisableAllFramerateCaps = False

And get back to us with min/max/averages - not instantaneous screengrabs, preferably at 800x600 with everything set low.

I'll get round to it sometime tomorrow.
 
I never tested FPS, but having task manager up on my 2nd monitor, in hardware mode it never used more than one CPU at any given time, in software mode it certainly did use more than one CPU.

jebo_4jc said:
I was disappointed to see CPU utilization never hit 100% on both of my "CPUs". The PhysX engine was supposed to be highly threaded. I saw no evidence of SMP support.

Because both processors weren’t hitting 100%, or were you only seeing activity on one CPU?
 
GotNoRice said:
I never tested FPS, but having task manager up on my 2nd monitor, in hardware mode it never used more than one CPU at any given time, in software mode it certainly did use more than one CPU.



Because both processors weren’t hitting 100%, or were you only seeing activity on one CPU?
Because both processors never hit 100%. I'm not sure how it is with a dual core CPU while playing single threaded games, but with HT, both "CPUs" fluctuate between about 10% and 90% usage, with the average usage at any one time being about 50%. In other words, sometimes one CPU will be about 20% loaded, while the other one is 80%, sometimes it will be 50/50, sometimes it will be 75/25. Cellfactor showed the same behavior in my brief playing with it last night. I'll try to run some benchmarks. We should try to compile benchmarks to see exactly what is going on here.
 
hmmm... I think there up to something... its a good thing i have not intrest in cell factor.

lets bench it on a game not sponsored by them! ohh wait, non exist yet...
 
I figured I’d do some Cellfactor testing using Fraps to do the measurments and see what I got.

test1physx.JPG


This is the first scene. I took that screenshot right before I began testing, and when I ran the follow up tests I pulled up the screenshot on my laptop and tried to position myself in exactly the same position. The testing consisted of me hitting the F11 key (to start the benchmark) and then right clicking my mouse to launch the grenade, and I would not touch the keyboard or mouse again until the benchmark was complete. I set it to stop benchmarking automatically after 20 seconds, to give debris and whatever a chance to fully settle.

I benchmarked it in 4 modes, with and without the PhysX card, and with and without my 2nd CPU enabled (using /onecpu switch in boot.ini).

PhysX Enabled, Multiple CPU:
Min: 28
Max: 43
Avg: 38.1

PhysX Disabled, Multiple CPU:
Min: 19
Max: 42
Avg: 38.65

PhysX Enabled, Single CPU:
Min: 26
Max: 37
Avg: 34.4

PhysX Disabled, Single CPU:
Min: 15
Max: 32
Avg: 27.45


test2physx.JPG


This is the 2nd scene, I tried to go for one with more debris. I tried shooting a gravity barrel thingy as a test, but every single time I did it, it seemed to fly off in a different direction. Shooting grenades into big stacks of debris seemed to be reproducible, so that’s what I did. In this case I launched the grenade from the edge of a ledge down onto the debris.

PhysX Enabled, Multiple CPU:
Min: 17
Max: 42
Avg: 36.5

PhysX Disabled, Multiple CPU:
Min: 10
Max: 40
Avg: 32.4

PhysX Enabled, Single CPU:
Min: 21
Max: 39
Avg: 32.4

PhysX Disabled, Single CPU:
Min: 11
Max: 31
Avg: 22.1

Edit: One more thing, when running Cellfactor in Software mode, none of the cloth was there. None of these scenes had any cloth in it that I know of, but I guess at this point PhysX in software can't do cloth while in hardware it can.
 
GotNoRice said:
I figured I’d do some Cellfactor testing using Fraps to do the measurments and see what I got.
Nice work. The minimums are helped the most on your tests.

Your settings were 800x600 and what settings in case others want to post comparisons?
 
pxc said:
Nice work. The minimums are helped the most on your tests.

Your settings were 800x600 and what settings in case others want to post comparisons?

I just used the "CellFactor Low Graphics" shortcut, adding and removing "EnablePhysX=false" (after also enabling or disabling the card itself via device manager). Interesting thing though, the shortcut was actually too large to fit "EnablePhysX=false" on the end of the target field, it cut off the last "e". I had to remove the part where it specifiec the resolution as 800x600 in order to fit the "EnablePhysX=false" in the target field of the shortcut, but since it was in 800x600 in before, the game still ran in 800x600 and the results should not be affected.
 
GotNoRice said:
I just used the "CellFactor Low Graphics" shortcut, adding and removing "EnablePhysX=false" (after also enabling or disabling the card itself via device manager). Interesting thing though, the shortcut was actually too large to fit "EnablePhysX=false" on the end of the target field, it cut off the last "e". I had to remove the part where it specifiec the resolution as 800x600 in order to fit the "EnablePhysX=false" in the target field of the shortcut, but since it was in 800x600 in before, the game still ran in 800x600 and the results should not be affected.
Great work. Thanks for your effort.
 
My athlon XP 3200+ and 6800GS AGP can handle cellfactor in all its HDR AF et al. glory excepting only the mightiest of "random stuff flying everywhere" explosions. Does this mean the physx card is useless? Hardly, the level displays hundreds of objects, fluids and cloth require thousands. Still, it would be nice to actually *see* such effects in use somewhere...

Edit: UT2k7 should include some more of the advanced effects, such as fluids(bloodsplatter) and other sfx. It's just disappointing that boxes was the way they chose to go initially :/
 
kuyaglen said:
What were the settings?
1280x1024 16AF HDR, Using the High Graphics shortcut.

I(illa Bee said:
why is there a PhsyX watermark on the bottom right of all the pics?
The physX watermart is there regardless if your using a PPU or not.

tazzmissionx said:
Which driver are you using? Ageia released 2.4.3b (beta) over a week ago.
Yes I am using these drivers, these are ones that Agiea recently said would fix the fps drops in GRAW, I've not seen any difference in performance tho.

GotNoRice said:
I figured I’d do some Cellfactor testing using Fraps to do the measurments and see what I got.
Nice test mate
 
I think this is really a matter of Aegia choosing the wrong type of thing to showcase their physics processor. Explosions are the easiest things to make dramatic, but if a cpu like mine can handle a couple hundred live bodies, a current-gen top-end cpu could probably handle a lot more. Neither, however, would be able to do fabric tearing, fluid flow(although I really didn't like the way Aegia did "fog" in their one demo), et al. Those are the kinds of things they should be making available in software mode specifically so that people without the PPU can say "Wow, this is a slideshow, I need one of these." Not, "Hey, this runs alright."
 
what is your system setup?

I tried CF without a physx card (before mine arrived) and it ran like a dog on my system (had to turn off HDR and lower some of the settings as my card doesn't do well @ 1280x1024 with high AA, etc. Even with those lowered to basically nothing I lagged something feirce when I gravity naded any large pile of stuff, or exploded a large pile of things.

Once I installed my physx card afterwards (and kept same settings) -I'll check later tonight for specific settings).. I noticed a good deal improovement as when things blew up and tons of things went flying everywhere it seemd alot more natural and smooth.

Perhaps the CPU type effects or hinders this more? What is your CPU/GPU stats?
 
Bicarus said:
what is your system setup?

physX Driver : 2.4.3 BETA-1 Release Date: May 5, 2006
Forceware version : 87.25

3.4 @ 3.6 Preshott (Intel)
2gig ddr2 @ 533 (Cosair)
7800gtx @ 500/1.3 (Leadtek)
physX PPU 128mb (BFG)

Nothing special I know but enougth to run most of todays games, well, apart from Oblivion thats is :(
 
Lord of Shadows said:
What! a gpu bottleneck can mask the effects of a cpu bottleneck?!

Now compare a celeron to a conroe in cellfactor ;o)

QFT
 
lowrider007 said:
With PPU : 15 FPS




Without PPU : 15 FPS


Wow man! How did you do 2 screen caps where every itsy bitsy fragment is in the exact same position?
 
Tenebrus said:
Wow man! How did you do 2 screen caps where every itsy bitsy fragment is in the exact same position?

good call... that should be impossible shouldn't it? Are there scripting capabilities available in Cell factor or something?
 
J-Mag said:
good call... that should be impossible shouldn't it? Are there scripting capabilities available in Cell factor or something?
This is the reason FRAPS benchmarks over a set amount of time are worthwhile.
 
jebo_4jc said:
This is the reason FRAPS benchmarks over a set amount of time are worthwhile.

Yeah but the individual still has to control the character for each run...
 
I've made some FRAPS videos, yes actual recordings, i have a dual Raptor setup in RAID 0 so recording demos to my primary drive is pretty smooth.

In 800 600 and capturing in half mode with sound and 4xAA to smooth out the scene, HDR is turned off, everything else is set to high, texture res is max and AF is set to 1x

I get about 20fps frapsing, and about 45 fps in normal gameplay.

Thats kinda high for something that REQUIRES a PPU to play.

System specs are 64 4000+ @ 2.7Ghz, 2Gb RAM 7800 GTX (256Mb) overclocked.

Its a pretty fast CPU for now a days, not multicore but still kicks the pants off most dual cores in games anyhow.

With AA turned off and HDR on, and with all the default settings set in the "high graphics" shortcut it's totaly playable on my system WITHOUT a PPU

One thing to mention here is that without the PPU you dont get blood physics or cloth physics in the map, I cant see cloth physics slowing down the game much, the blood might however.

I might fraps some more tomorrow and try and find some decent hosting to get it on so you guys can see some "proof" if you like, i'll make sure to capture all the menu's as well.
 
lowrider007 said:
1280x1024 16AF HDR, Using the High Graphics shortcut.


The physX watermart is there regardless if your using a PPU or not.


Yes I am using these drivers, these are ones that Agiea recently said would fix the fps drops in GRAW, I've not seen any difference in performance tho.


Nice test mate

I've just tried the same settings as you (but with 1280 960 cos i have a CRT) and it's totaly playable without a PPU. FRAPS reporting around 17fps during large explosions and about 30 fps most of the rest of the time, it feels really smooth for 30 fps as well, some games are just like that...
 
Back
Top