Dedicated physx card

[voxgm]

So I bought 2 Gtx 560ti for $50. My board only supports crossfire so sli isent an option. Would using one as a dedicated physx card work ok. I could see a problem with running a Gtx 650 as a dedicated physx card if the main card was something like a Gtx 980 ti. The 650 dedicated I'm sure is slower then the 980ti at processing physx so there would be no benefit. Maybe even decreased preformance with a config like that. But being as the cards are the same then it should improve performance overall right??? Because the cards have the same speed. One would take the physx load off the main card??

 

[bgunner]

The GPU's do not need to be the same speed for one to useful as a phys-X card. The reason being is that the only thing a phys-X card does is process GPU bound physics. This take the pressure off the main GPU so it can render faster/Higher FPS. While it would be a bit of a waste of processing power you can use it like that if you so choose it no ill effects.

 

[pigeoncracker]

PhysX is a flawed process. You would not see any performance increase setting one of these cards up for it. Except if you were running Metro Last Light, then you might get a 5 FPS difference. Not worth it with the extra card drawing power and heating up your system.

I'd stick with the single card for now and upgrade the board for SLI later.

Using a dedicated card for PhysX sets that card aside for rendering ONLY PhysX. Meaning all power of that card is going towards rendering it. People have typically used much weaker cards alongside better cards for a slight performance increase. I tried it once, never did it again.

That said, how many games that have been coming out have used PhysX? I think I've only played 2 games with even a PhysX option. It's not common and really won't be used.

 

[bgunner]

PhysX is a flawed process. You would not see any performance increase setting one of these cards up for it. Except if you were running Metro Last Light, then you might get a 5 FPS difference. Not worth it with the extra card drawing power and heating up your system.

I'd stick with the single card for now and upgrade the board for SLI later.

Using a dedicated card for PhysX sets that card aside for rendering ONLY PhysX. Meaning all power of that card is going towards rendering it. People have typically used much weaker cards alongside better cards for a slight performance increase. I tried it once, never did it again.

That said, how many games that have been coming out have used PhysX? I think I've only played 2 games with even a PhysX option. It's not common and really won't be used.

you will see a much greater FPS jump if a weaker primary card is is used. This does not say set a more powerful card as a phys-X card, just that if the primary card is not supper powerful then you will see a huge gain. There are actually a lot of games that have phys-X capability. I own 6 or 8 of them myself.

 

[voxgm]

I do play metro 2033, metro last light, sacred 2, overlord 2, witcher 3, fallout 4 and some others. Thats why I asked...

 

[voxgm]

I'm confused with pigeoncracker saying there is no increase preformance when I saw benchmarks for several setups showing a very decent increase. Most of the benches however are using a slower card then the main. Instead of 2 of the same or 2 with close preformance

 

[pigeoncracker]

you will see a much greater FPS jump if a weaker primary card is is used. This does not say set a more powerful card as a phys-X card, just that if the primary card is not supper powerful then you will see a huge gain. There are actually a lot of games that have phys-X capability. I own 6 or 8 of them myself.

Fair enough. I'm not a fan of it because I got a lot of crashing, but that could just be me. You will definitely see a gain with a weaker primary card, but I don't like it. My opinion.

It won't harm anything if you go with it. If it crashes, remove it. If it doesn't. You're good and roll with it for a bit.

 

[pigeoncracker]

I'm confused with pigeoncracker saying there is no increase preformance when I saw benchmarks for several setups showing a very decent increase. Most of the benches however are using a slower card then the main. Instead of 2 of the same or 2 with close preformance

I'm sorry. I'll clear that up. You will see a performance gain. I don't believe that it will be a significant enough of a gain for you to fully notice it. Any game that doesn't use PhysX will see no gain.

Most benches use a slower card than the main because I'm pretty sure that's what Nvidia had in mind when they developed the system. Reuse older, redundant cards to gain FPS with your upgrade from Nvidia. It was not designed with using two of the same card in mind. Regardless, it should perform the same as the benches you've seen.

 

[voxgm]

Ah ok I gotcha. Best why to test is to go in hands on and see what happens. Just have to test myself and see if it's worth it or not. Makes since as all system will probably show different results. I'll test metro with and without a dedicated card and see what happens. Tho metro last light maybe preform better or worse then other games I would imagine. Depending.

 

[bgunner]

I'm confused with pigeoncracker saying there is no increase preformance when I saw benchmarks for several setups showing a very decent increase. Most of the benches however are using a slower card then the main. Instead of 2 of the same or 2 with close preformance

This is not surprising in the least, That statement had a lot of falsity's in it. you will see a gain, a measurable gain, But if you can play the game on max settings @60 FPS, on a 60Hz monitor, on just 1 card using a phys-X card will show very little gain.

 

[bgunner]

Ah ok I gotcha. Best why to test is to go in hands on and see what happens. Just have to test myself and see if it's worth it or not. Makes since as all system will probably show different results. I'll test metro with and without a dedicated card and see what happens. Tho metro last light maybe preform better or worse then other games I would imagine. Depending.

metro and metro last light are very GPU intensive and are a very difficult games on systems to play, the redux versions are even more so.

 

[pigeoncracker]

So, I decided to do some testing (because I was bored. Humor me.) and I found that I don't gain any FPS at all, really. Sharing what's happening down with me.

I really don't know what I'm doing anymore.

http://imgur.com/a/qUhpH

This was all done off of Metro Last Light Redux's bench.

 

[bgunner]

So, I decided to do some testing (because I was bored. Humor me.) and I found that I don't gain any FPS at all, really. Sharing what's happening down with me.

I really don't know what I'm doing anymore.

http://imgur.com/a/qUhpH

This was all done off of Metro Last Light Redux's bench.

Humm your CPU doesn't have graphics in it, plus PhysX only works on Nvidia cards. Your results are interesting though as they seem to show a drop in FPS BUT did you run the bench 3 times after a ten run warm up? this preps the system for heat, loads the memory, allows the hard drive/ssd's to bring the info forward and ready to read the info. then you average the three benches together and get your average frame rate.

 

[pigeoncracker]

So, I decided to do some testing (because I was bored. Humor me.) and I found that I don't gain any FPS at all, really. Sharing what's happening down with me.

I really don't know what I'm doing anymore.

http://imgur.com/a/qUhpH

This was all done off of Metro Last Light Redux's bench.

Humm your CPU doesn't have graphics in it, plus PhysX only works on Nvidia cards. Your results are interesting though as they seem to show a drop in FPS BUT did you run the bench 3 times after a ten run warm up? this preps the system for heat, loads the memory, allows the hard drive/ssd's to bring the info forward and ready to read the info. then you average the three benches together and get your average frame rate.

I did average three benches together for the results. I was playing Overwatch for about three hours before and that heats it up fairly well regardless. I did nearly immediately go to the bench after. I know my CPU doesn't have graphics, but there was a PhysX CPU option in the menu. I was bored enough to see if it made a difference at all.

There was a slight improvement when adding a PhysX card, about two more FPS when adding the 550 ti as a PhysX card for the 960 and a 1 FPS difference when using the 550 ti as the main card and the 960 as the PhysX card. I have a GTS 250 lying around somewhere. I'll see if I get any different results when using that if I have the time, just to see how old the hardware can be.

My usual system specs, if you're curious:
GPU: MSI GTX 960 2G OC (1463MHz core, 7806MHz effective memory speed)
CPU: AMD FX-8320 (4.5GHz)
RAM: 16GB DDR3 1866MHz
Motherboard: MSI 970 Gaming

The bench was running off a 7200rpm hard drive, not my SSD.