Rastalovich
New member
So being as PhysX is everywhere badged on nVidia cards, and a revolution in physics enabled software is meant to be happening, I, like so many other people, have thought "well what's the best setup for me?".
I've looked at a good selection of review sites, and been linked to a load, that in all honesty are dedicated to what the new hardware and drivers can do, some coupled with using an older existing 8 series card for a comparison. None looked from the pov of people who owned the Aegia PhysX card already. Sure they looked at mixtures of setups and made some comparisons of newer hardware, but there was always a simple stat missing that left me thinking "ok.. but if I already have a Aegia PhysX card - what then ?".
I do get the impression that the Aegia cards were dismissed immediately cos they are deemed 'older tech' and can't possibly be held as an option - but this is crazy, I mean OK it's a fair assumption, but u can't make these assumptions unless u run the test. In my mind, it's similar to people dismissing an AMD built pc without seeing what it actually does just by looking at benchmarks and not how it plays ur game.
So, having already got my gaming pc in pieces, and some hardware on the way, I figured I'd run the test myself to finally put this thing to bed.
This test rig consists of:
ABit IX48 GT3
QX9650 @ 3g stock with OCZ Vendetta2
2x2g OCZ PC31280 7-7-7 @ 1360
At this point u have to appreciate that the mobo is not currenty extreme friendly, in fact the bios isn't memory friendly. The stats above are what I can guarantee the cpu/mem are. Any changes made cannot be said to be true. Although the memory is rated 1.9v, it seems to want to operate between 1.5 and 1.65v, remember we aren't X58 here. 7-7-7-24 is the norm, 8-8-8-25 allowed me to get 9x370 at some point. Saved the profile and it wouldn't load back - so eff it all, it's a good setup @ stock. ABit don't come out with a new bios (there's rumors of people being emailed upd8s), I'll have to change oc mobos for anything +ve.
also in hand:
XFX GTX280 1G @ stock 602
XFX 8800GT 512m @ stock 670
BFG Aegia PhysX 128m @ stock (whatever that is - but years ago I'm sure people oc'd these somehow)
Running the text using nVidia's 178.24 driverset and FluidMark as the bencher. 2 main scenarios, GTX280 as the main display card, and 8800GT as the main card. Tests were repeated with the insertion of cards to be sure nothing spooky was going on.
The OS of choice here is Windows XP Professional SP3. Now I did ponder installing Vista 64 bit too, but having seen the FluidMark (which I chose to run without any settings changes - just click GO), it doesn't do anything graphically dramatic, there are no Dx10 effects going on that I could see. If there is a case for 10>9 I may rethink, but I extremely doubt it. It's not so much an assumption, that I don't really like making, but the output is that basic.
Tests simply selecting the PhysX control panel, choosing ur weapon, and running FluidMark. Rerunning the tests to be sure nothing weasily is going on.
GTX280 Main card
No Physic Acceleration ------ 1962
PCI Ageia PhysX ------------- 2394
8800GT doing PhysX ------- 10127
GTX280 Acting on it's own - 11112
8800GT Main card
No Physic Acceleration ----- 1532
PCI Ageia PhysX ------------ 1929
GTX280 doing PhysX - well I know it'd be stoopid, but I did try, and the CD-IN header on the mobo wouldn't allow the card to go in a lower PCIe slot, and the SB blocked the other one. It would've been interesting, but blame XFX for bulking so much plastic around the cooler.
8800GT Acting on it's own - 7720
I did just try sticking a MSI 8600 GTS in to give a 8800GT comparison. MSI had chosen to stick a big m-f grill on the reverse of the card that spans more than another slot, with made it impossible. It would have been nice, but I thing the 8800GT answers enough.
In summary: It would have been interesting to have a few more pieces of hardware at hand. Perhaps a 256m Aegia card. The 8600 to see how it works compared to the 8800GT.
What I can see from this is the only time u could entertain a PCI Ageia card presently is if u r on budget and dont plan on replacing a 7 series card. In this case u would have to download a seperate driverset and use the Ageia card as it was originally intended. It's true that u would be limiting ur experience - but hey a persons budget is their budget. If u want to build a pc for some1 and have a 7950 card at hand, it'd help the dx9 experience. What games will support this moving forward is questionable, but there are a good collection of existing ones.
In terms of the GTX280 owner. I'm not sure u'd bother with any option other than using the card to do everything.
There is an important line to be drawn tho. To this point we've used a benchmark, and we all know what they're all about. Unrealistic in comparison to gaming. It could well be the case that although the lowest score above is 1532 with the 8800GT and the QX9650 doing the business, it would be game-dependent on whether this is enough for the game ur playing. If no game out there claims 1000 equivalent marks to do what it needs to do - why not stick to just cpu physics ? Being slightly concerned, stick an Ageia card in for around 2000 equivalent marks. This would all be situation dependent ofc. If u have arguably a 8800GT, and more importantly a GTX280 - there may not be any point in thinking about it. The gpu processing is obviously doing the job 7x & 11x what the other options can do.
How does this impact fps in ur game ? There is obviously a % gpu cost to doing the work. If ur doing 100s of fps on the game, u won't worry about it. If ur doing 50 u may think about the options to smooth things out.
I'm selling my BFG Aegia PhysX card in work next week
I've looked at a good selection of review sites, and been linked to a load, that in all honesty are dedicated to what the new hardware and drivers can do, some coupled with using an older existing 8 series card for a comparison. None looked from the pov of people who owned the Aegia PhysX card already. Sure they looked at mixtures of setups and made some comparisons of newer hardware, but there was always a simple stat missing that left me thinking "ok.. but if I already have a Aegia PhysX card - what then ?".
I do get the impression that the Aegia cards were dismissed immediately cos they are deemed 'older tech' and can't possibly be held as an option - but this is crazy, I mean OK it's a fair assumption, but u can't make these assumptions unless u run the test. In my mind, it's similar to people dismissing an AMD built pc without seeing what it actually does just by looking at benchmarks and not how it plays ur game.
So, having already got my gaming pc in pieces, and some hardware on the way, I figured I'd run the test myself to finally put this thing to bed.
This test rig consists of:
ABit IX48 GT3
QX9650 @ 3g stock with OCZ Vendetta2
2x2g OCZ PC31280 7-7-7 @ 1360
At this point u have to appreciate that the mobo is not currenty extreme friendly, in fact the bios isn't memory friendly. The stats above are what I can guarantee the cpu/mem are. Any changes made cannot be said to be true. Although the memory is rated 1.9v, it seems to want to operate between 1.5 and 1.65v, remember we aren't X58 here. 7-7-7-24 is the norm, 8-8-8-25 allowed me to get 9x370 at some point. Saved the profile and it wouldn't load back - so eff it all, it's a good setup @ stock. ABit don't come out with a new bios (there's rumors of people being emailed upd8s), I'll have to change oc mobos for anything +ve.
also in hand:
XFX GTX280 1G @ stock 602
XFX 8800GT 512m @ stock 670
BFG Aegia PhysX 128m @ stock (whatever that is - but years ago I'm sure people oc'd these somehow)
Running the text using nVidia's 178.24 driverset and FluidMark as the bencher. 2 main scenarios, GTX280 as the main display card, and 8800GT as the main card. Tests were repeated with the insertion of cards to be sure nothing spooky was going on.
The OS of choice here is Windows XP Professional SP3. Now I did ponder installing Vista 64 bit too, but having seen the FluidMark (which I chose to run without any settings changes - just click GO), it doesn't do anything graphically dramatic, there are no Dx10 effects going on that I could see. If there is a case for 10>9 I may rethink, but I extremely doubt it. It's not so much an assumption, that I don't really like making, but the output is that basic.
Tests simply selecting the PhysX control panel, choosing ur weapon, and running FluidMark. Rerunning the tests to be sure nothing weasily is going on.
GTX280 Main card
No Physic Acceleration ------ 1962
PCI Ageia PhysX ------------- 2394
8800GT doing PhysX ------- 10127
GTX280 Acting on it's own - 11112
8800GT Main card
No Physic Acceleration ----- 1532
PCI Ageia PhysX ------------ 1929
GTX280 doing PhysX - well I know it'd be stoopid, but I did try, and the CD-IN header on the mobo wouldn't allow the card to go in a lower PCIe slot, and the SB blocked the other one. It would've been interesting, but blame XFX for bulking so much plastic around the cooler.
8800GT Acting on it's own - 7720
I did just try sticking a MSI 8600 GTS in to give a 8800GT comparison. MSI had chosen to stick a big m-f grill on the reverse of the card that spans more than another slot, with made it impossible. It would have been nice, but I thing the 8800GT answers enough.
In summary: It would have been interesting to have a few more pieces of hardware at hand. Perhaps a 256m Aegia card. The 8600 to see how it works compared to the 8800GT.
What I can see from this is the only time u could entertain a PCI Ageia card presently is if u r on budget and dont plan on replacing a 7 series card. In this case u would have to download a seperate driverset and use the Ageia card as it was originally intended. It's true that u would be limiting ur experience - but hey a persons budget is their budget. If u want to build a pc for some1 and have a 7950 card at hand, it'd help the dx9 experience. What games will support this moving forward is questionable, but there are a good collection of existing ones.
In terms of the GTX280 owner. I'm not sure u'd bother with any option other than using the card to do everything.
There is an important line to be drawn tho. To this point we've used a benchmark, and we all know what they're all about. Unrealistic in comparison to gaming. It could well be the case that although the lowest score above is 1532 with the 8800GT and the QX9650 doing the business, it would be game-dependent on whether this is enough for the game ur playing. If no game out there claims 1000 equivalent marks to do what it needs to do - why not stick to just cpu physics ? Being slightly concerned, stick an Ageia card in for around 2000 equivalent marks. This would all be situation dependent ofc. If u have arguably a 8800GT, and more importantly a GTX280 - there may not be any point in thinking about it. The gpu processing is obviously doing the job 7x & 11x what the other options can do.
How does this impact fps in ur game ? There is obviously a % gpu cost to doing the work. If ur doing 100s of fps on the game, u won't worry about it. If ur doing 50 u may think about the options to smooth things out.
I'm selling my BFG Aegia PhysX card in work next week
