So where's the PhsyX drivers?

i could put a 9500GT in my Sugo03 system, hopefully the PCIe 4x slot wouldn't bottleneck it any.
 
Atech said:
I migh chime in, just for the fun of it...but need to sort a driver bug on FreeBSD first.
(Because I own a PPU, and with out one of those in the bag, this chart says nothing)
Click to expand...

I would love to see your data! :D
 
R3MF said:
i could put a 9500GT in my Sugo03 system, hopefully the PCIe 4x slot wouldn't bottleneck it any.
Click to expand...

Bus-bandwith has never been important to physics, as the havy calutaions are done inside the PPU, and only the vetordata is transfered over the BUS.
Lantency of the bus is much more important.
 
Keep us honest! :cool:

FYI - The approximate speed of an 8600 GT.
This is the one that the ageia PPU would have to beat, to be able to run with the 8 series acting as PPU's.

http://www.gpureview.com/GeForce-8600-GT-card-513.html

Manufacturer: nVidia
Series: GeForce 8
GPU: G84
Release Date: 2007-04-17
Interface: PCI-E x16
Core Clock: 540 MHz
Shader Clock: 1180 MHz
Memory Clock: 700 MHz (1400 DDR)
Memory Bandwidth: 22.4 GB/sec
FLOPS: 75.52 GFLOPS
Pixel Fill Rate: 4320 MPixels/sec
Texture Fill Rate: 8640 MTexels/sec



The official ageia card:

http://www.stageselect.com/News/NewsViewer.aspx?newsid=2036&fromint=1

"AGEIA PhysX chip that is capable of approximately 533,000 max collisions per second; 58 GFLOPS sphere-sphere collisions per second; and 22 billion instructions per second".
 
TalonMan said:
Trying to keep our data recorded... :)

Member ____ CPU / Speed _______ GPU _____ PPU _________ Res ____ Sentinel Test
TalonMan ------- Q6600@3.78GHz ----- GTX 280 ---- None --------- 1920x1200 ------ 72-73FPS

Acoustique ----- XXXXXXXXXXXXX ------- 8800GTS --- None ---------- 1680x1050 ------- 55FPS
Acoustique ----- XXXXXXXXXXXXX ------- 8800GTS --- 8600 GT ----- 1680x1050 ------- 72FPS

nst6563 ---------- Q6600@3.6GHz ------ GTX 260 ---- None ---------- 1440x900 --------- 79FPS
nst6563 ---------- Q6600@3.6GHz ------ GTX 260 ---- 8800 GTS --- 1440x900 -------- 93-96FPS

Is this correct? ;)
Click to expand...

prolly should add iam running a e6300@2.8GHz

seems to be adding an extra 20fps, can anyone score an extra 30fps?
 
I didn't think there would be THAT big of a difference between running the physx on the 260 and on a seperate card, because my resolution isn't that high and the 260 is already a very powerful card...but seeing the results I'm leaving the gts installed ;)

With the new drivers coming out, do you want to what drivers to the chart?

fyi - for reference. With NO physx acceleration, I get a solid 21fps :p

 
I wonder how efficient it would be with a GTX 280 plus an 8800GTX as the PPU @ 1920x1200...
 
I think there might be diminishing returns on using faster and faster cards for the PPU function, but it would be very interesting to try it out. Anyone have an 8600gt, 9600gt (which is probably just a tad slower than my oc'd gts640), 8800GTX and a GTX260 or 280? ... or just some very deep pockets? :p
 
TalonMan said:
Keep us honest! :cool:

FYI - The approximate speed of an 8600 GT.
This is the one that the ageia PPU would have to beat, to be able to run with the 8 series acting as PPU's.

http://www.gpureview.com/GeForce-8600-GT-card-513.html

Manufacturer: nVidia
Series: GeForce 8
GPU: G84
Release Date: 2007-04-17
Interface: PCI-E x16
Core Clock: 540 MHz
Shader Clock: 1180 MHz
Memory Clock: 700 MHz (1400 DDR)
Memory Bandwidth: 22.4 GB/sec
FLOPS: 75.52 GFLOPS
Pixel Fill Rate: 4320 MPixels/sec
Texture Fill Rate: 8640 MTexels/sec



The official ageia card:

http://www.stageselect.com/News/NewsViewer.aspx?newsid=2036&fromint=1

"AGEIA PhysX chip that is capable of approximately 533,000 max collisions per second; 58 GFLOPS sphere-sphere collisions per second; and 22 billion instructions per second".
Click to expand...

Comparing FLOPS like that is the best reason why FLOPS are useless when comparing different architechtures ;)
 
Acoustique said:
prolly should add iam running a e6300@2.8GHz

seems to be adding an extra 20fps, can anyone score an extra 30fps?
Click to expand...

Thanks for ths info:

Member ____ CPU / Speed _______ GPU _____ PPU _________ Res ____ Sentinel Test
TalonMan ------- Q6600@3.78GHz ----- GTX 280 ---- None --------- 1920x1200 ------ 72-73FPS

Acoustique ----- e6300@2.8GHz ------- 8800GTS --- None ---------- 1680x1050 ------- 55FPS
Acoustique ----- e6300@2.8GHz ------- 8800GTS --- 8600 GT ----- 1680x1050 ------- 72FPS

nst6563 ---------- Q6600@3.6GHz ------ GTX 260 ---- None ---------- 1440x900 --------- 79FPS
nst6563 ---------- Q6600@3.6GHz ------ GTX 260 ---- 8800 GTS --- 1440x900 -------- 93-96FPS
Note from nst6563:With NO physx acceleration, I get a solid 21fps. (Physx on CPU)
 
 
Am using the GTX280 for graphics and my old 8800GTX as dedicated PPU. I am just downloading the new drivers and full physX pack, so currently using the 177.79 Beta and 8.07.18 PhysX drivers released some time ago.

I must say it works very well on GRAW2 with 1.05 patch and Warmonger.

I am usually playing with vsync enabled and for both GRAW2 at Ageia Island level and Warmonger I run at constant 60FPS, will try tomorrow to disable vsync and post the FPS if interested.
 
Running MKZ Benchmark, that was part of the pack today..

On my single 280,

1920x1200: AVE FPS=69, MIN FPS=38, MAX FPS=84.
1280x1024: AVE FPS=71, MIN FPS=44, MAX FPS=85.

Looks like doing the extra work at high res dosen't hurt my 280's performance.
I bet it using it's 1GB of ram! ;)

Nurien Demo
AVE FPS=70, MIN FPS=48, MAX FPS=124.
 
With these latest drivers do you still need a second monitor plugged in for Vista when using a separate GeForce card for Physx?
 
k ran that liquid demo
It would appear liquids need something powerful

88gts512 @ 25fps
88gts512 + 86gt @ 17fps

Re-tried warmonger with the same results as before
 
I guess the wheels fall off the 8600 GT when acting as a PPU on the more demanding demos?
 
Q9450@3.5GHz 2GB Ram GTX 280

MKZ Benchmark
1920x1200
avg 78 fps
min 42 fps
max 91 fps

1280x1024
avg 80 fps
min 49 fps
max 92 fps
 
Acoustique said:
k ran that liquid demo
It would appear liquids need something powerful

88gts512 @ 25fps
88gts512 + 86gt @ 17fps

Re-tried warmonger with the same results as before
Click to expand...

I'm not totally sure of you mean fluid or that nvidia liquid test.

Anyway, Fluid is broken vs multimonitors, disable any secondary monitors. I got stuck at 25fps with my GTX 280 too.
It only uses raw physics with very little textures so using the second card as physX is useless in that benchmark unless it's faster than the main card(which is very unlikely).
 
Tried PhsyX on my modest rig (e6420, 9600GT, 2 gigs of ram, XP Pro SP3) and loaded up the UT3 PhysX maps. My system froze up tight as a drum on Tornado, in fact UT3 froze up when I selected a non PhysX map as well after installing these drivers. Never had an issue with previous sets. I did get to play Heat Ray and the other map and the performance was what I expected with my rig..mediocre. I did get Tornado to run the second time I tried it and I had the lowest performance on that. All in all the effects were good, but the performance is not where I want it to be on my rig and the lockups didn't help. I imaged my machine back a few days (too lazy to unistall the drivers, run driver cleaner, install new drivers..etc..I had shit to do) and all is well again.
 
Im glad others are getting some freeze ups because I did too. I got a few just on the menu pages of UT3 and couple within the game itself. also even my old Unreal 2 game was freezing up a few times. before you say that cant be related I want you to know that back when I installed UT3, Blacksite Are51 or any other new Unreal engine game I would get prompted to uninstall Unreal Gold. anyway I uninstalled the physx and reinstalled and it seems to be working fine so far.
 
cannondale06 said:
Im glad others are getting some freeze ups because I did too. I got a few just on the menu pages of UT3 and couple within the game itself. also even my old Unreal 2 game was freezing up a few times. before you say that cant be related I want you to know that back when I installed UT3, Blacksite Are51 or any other new Unreal engine game I would get prompted to uninstall Unreal Gold. anyway I uninstalled the physx and reinstalled and it seems to be working fine so far.
Click to expand...


I uninstalled the Ageia software(UT3 originally installed that ) then my old drivers and rebooted into safe mode and ran driver cleaner pro and driver sweeper and removed all remnants of the old Nvidia drivers and the Ageia software. I then booted into regular mode and installed the new Nvidia drivers/PhysX package. I've uninstalled, reinstalled drivers thousands of times so i'm not a noob.

I'll probably wait a bit and then give it another shot. I read around a bit and people are having some issues with shit locking up. Don't know if it's the driver or the PhysX package, I might just use WinRAR to extract the driver set and install just the drivers and try some non-PhysX games and then install PhysX again. I would prefer them as separate packages anyway. I noticed they can be uninstalled separately in add/remove programs.

Thank God for images..i'm a lazy ass when it comes to my own system.
 
TalonMan said:
I guess the wheels fall off the 8600 GT when acting as a PPU on the more demanding demos?
Click to expand...

Now see why GFLOPS means nothing, when comparing different architechtures?

Anyway it won't be untill next week I get my benches done, left home to be with the GF...and I only took my laptop with me, my workstation was too big to carry ;)
 
Thanks for the info: :)
Sentinel Test: Start FRAPS, then start WarMonger's Tutorial, but don't move. Only spin to the left and put your crosshairs centered directly on the Green gun labled 'Sentinel'. Recore your FPS. Mine held constant at 72 or 73 FPS.

Member ____ CPU / Speed _______ GPU _____ PPU _________ Res ____ Sentinel Test
TalonMan ------- Q6600@3.78GHz ----- GTX 280 ---- None --------- 1920x1200 ------ 72-73FPS

Backlash ------- Q6600@3.51GHz ----- GTX 260 ---- None ---------- 1920x1200 -------- 71FPS
Backlash ------- Q6600@3.51GHz ----- GTX 260 ---- 8800 GT ----- 1920x1200 -------- 87FPS

nst6563 ---------- Q6600@3.6GHz ------ GTX 260 ---- None ---------- 1440x900 --------- 79FPS
nst6563 ---------- Q6600@3.6GHz ------ GTX 260 ---- 8800 GTS --- 1440x900 ------- 93-96FPS

Acoustique ----- e6300@2.8GHz ------- 8800GTS --- None ---------- 1680x1050 ------- 55FPS
Acoustique ----- e6300@2.8GHz ------- 8800GTS --- 8600 GT ----- 1680x1050 ------- 72FPS

Note from nst6563: With NO physx acceleration, I get a solid 21fps. (Physx on CPU)

Note: When using an 8600 GT as a dedicated PPU, it will register a FPS increase with an 8800 GTS as the GPU, running our Sentinel Test in WarMonger. It did however hurt the system's performance running Nvidia's Particle Fluid Demo. When Acoustique used his same two card config running the Particle Fluid Demo, the 8600 GT actually hurt his systems performance having to brunt all of the PhysX responsibilities. He did better running only on his 8800 GTS alone, calculating both PhysX and Graphics. It appears the wheels fall off of the 8600 GT, when acting as a dedicated PPU, on the more demanding applications? The 8600 GT helped Acoustique in WarMonger, but hurt his performance in the Particle Fluid Demo.
 
I should note the 8800GT is clocked @ 700|2000 1750.

EDIT:
Added a shot for the GT to my last results.
 
Yeh iam starting to think that liquid tech demo from nvidia gave lower results becuase it might of been rendering and doing physx with the 86
thats just from a quick 5 min test i did b4 work this morning

but more testing ahoy
Ill try get something i can repeat in cellfactor
 
My first results were 1920x. Here are my results @ a 1440x res. Same settings, or max. CPU is a Q6600 G-0 @ 3.51GHz.

GTX 260 @ 760|2600 1580 = 93FPS



Adding an 8800GT @ 700|2000 1750 = 123FPS



nst6563 said:
If I do the same (load tutorial and turn to target the sent gun). I get between 93 and 96 fps with the GTX260 @ 756c/1512s/1188m and the 8800gts640 @ 700c/1600s/1000m doing the physx.



Now when I disable the 8800GTS and have the graphics AND physx done by the GTX260 I get significantly less...almost 20fps less.




:EDIT: OK...wtf is up with my pictures... going to re-upload them and try again...
:EDIT2: Pix fixed.
Click to expand...
 
Updated!! :)

Sentinel Test: Start FRAPS, then start WarMonger's Tutorial, but don't move. Only spin to the left and put your crosshairs centered directly on the Green gun labled 'Sentinel'. Recore your FPS. Mine held constant at 72 or 73 FPS.

Member ____ CPU / Speed _______ GPU _____ PPU _________ Res ____ Sentinel Test
TalonMan ------- Q6600@3.78GHz ----- GTX 280 ---- None --------- 1920x1200 ------ 72-73FPS

Backlash ------- Q6600@3.51GHz ----- GTX 260 ---- None ---------- 1920x1200 -------- 71FPS
Backlash ------- Q6600@3.51GHz ----- GTX 260 ---- 8800 GT ----- 1920x1200 -------- 87FPS

Backlash ------- Q6600@3.51GHz ----- GTX 260 ---- 8800 GT ----- 1440x900 --------- 93FPS
Backlash ------- Q6600@3.51GHz ----- GTX 260 ---- 8800 GT ----- 1440x900 -------- 123FPS

nst6563 ---------- Q6600@3.6GHz ------ GTX 260 ---- None ---------- 1440x900 --------- 79FPS
nst6563 ---------- Q6600@3.6GHz ------ GTX 260 ---- 8800 GTS --- 1440x900 ------- 93-96FPS

Acoustique ----- e6300@2.8GHz ------- 8800GTS --- None ---------- 1680x1050 ------- 55FPS
Acoustique ----- e6300@2.8GHz ------- 8800GTS --- 8600 GT ----- 1680x1050 ------- 72FPS

Note from nst6563: With NO physx acceleration, I get a solid 21fps. (Physx on CPU)

Note: When using an 8600 GT as a dedicated PPU, it will register a FPS increase with an 8800 GTS as the GPU, running our Sentinel Test in WarMonger. It did however hurt the system's performance running Nvidia's Particle Fluid Demo. When Acoustique used his same two card config running the Particle Fluid Demo, the 8600 GT actually hurt his systems performance having to brunt all of the PhysX responsibilities. He did better running only on his 8800 GTS alone, calculating both PhysX and Graphics. It appears the wheels fall off of the 8600 GT, when acting as a dedicated PPU, on the more demanding applications? The 8600 GT helped Acoustique in WarMonger, but hurt his performance in the Particle Fluid Demo.
 
I ran that metal knigh zero bench

and scored an avg of 45 using 88gts512 single
and only averaged 27fps when using the 8600gt fo phsyx

Things dont look to good ATM for the 86gt being used for phsyx
 
Thought I'd chime in after I did some tweaking. Seems that overclocking the PCI-E bus does have an impact on this stuff as well. I didn't however test to see if it had any impact on NON-Physx content though.

But oc'ing my PCI-E bus to 130Mhz increased my max fps to the 104-105 range in Warmonger.



something else I noticed while messing around. The speed of the card doing physx seems to make a difference as well depending on what's being calculated or how heavy the physx content is. I dropped my 8800gts640 from 680Mhz to the stock (I think...been so long ago) 600Mhz and ran the nvidia fluid demo. Looked ok, got around 30fps (I think it's capped), however the water seemed...sluggish...as if it were "thick" water. Then I increased the clock speed of the gts640 to 760Mhz (my gts was a beast...love it) and re-ran the demo and the water movement was much more fluid and seemed less "thick". Can someone try the same type of test and see if they get the same results or am I just on crack? :p
 
nst6563 said:
Thought I'd chime in after I did some tweaking. Seems that overclocking the PCI-E bus does have an impact on this stuff as well. I didn't however test to see if it had any impact on NON-Physx content though.

But oc'ing my PCI-E bus to 130Mhz increased my max fps to the 104-105 range in Warmonger.



something else I noticed while messing around. The speed of the card doing physx seems to make a difference as well depending on what's being calculated or how heavy the physx content is. I dropped my 8800gts640 from 680Mhz to the stock (I think...been so long ago) 600Mhz and ran the nvidia fluid demo. Looked ok, got around 30fps (I think it's capped), however the water seemed...sluggish...as if it were "thick" water. Then I increased the clock speed of the gts640 to 760Mhz (my gts was a beast...love it) and re-ran the demo and the water movement was much more fluid and seemed less "thick". Can someone try the same type of test and see if they get the same results or am I just on crack? :p
Click to expand...
stock 8800gts 640mb is only 500mhz. how in the hell did you get that to 760? I have never heard of an overclock that high on the 90nm 8800gts 640mb. :confused:
 
cannondale06 said:
stock 8800gts 640mb is only 500mhz. how in the hell did you get that to 760? I have never heard of an overclock that high on the 90nm 8800gts 640mb. :confused:
Click to expand...

Voltage and cooling mods ;) Believe me...it's not stable at that speed for long, basically enough to bench it and then let it cool. If I had a chilled water setup I could probably run it that way for a while, but I normally run it at 710Mhz @ 1.35v. I got it to 680Mhz on stock voltage.
 
nst6563 said:
Thought I'd chime in after I did some tweaking. Seems that overclocking the PCI-E bus does have an impact on this stuff as well. I didn't however test to see if it had any impact on NON-Physx content though.

But oc'ing my PCI-E bus to 130Mhz increased my max fps to the 104-105 range in Warmonger.
Click to expand...

So this means that perhaps PCI-E 2.0's time has finally come? Anyone collecting stats should probably add a column for mobo PCI-E version so it can be tracked.
 
You must log in or register to reply here.
Back
Top