Hey all! So for the first time I'll be running 2 GPU's and find myself wondering if my current setup will make the cut.
So I'm planning my system for the game Star Citizen. According to the Devs it's going to be pretty heavy physics wise. And it's going to be a PhysX enabled game. So my setup will be one current gen (at the time) high end card to run an Oculus Rift and a second card being a previous gen high end card to run PhysX.
My current setup is an i5-3570k so I'm not worried about CPU power, but with the limit at 8x lanes for 2 cards I'm wondering about PCIe bandwidth. I haven't been able to find anything about how much bandwidth is used by PhysX and I do know that higher FPS (which you kind of need for the Rift, it would seem 75-90fps) uses more bandwidth.
Now I have done a bit of Google searching and found an article at Tech PowerUp that shows scaling with a 680 that shows that x8 is fine and x4 shows only a 2% performance loss. And a article at Anand Tech with 2 Titans in SLI that shows up to a 7% boost going from x16/x16 PCIe 2.0 to 3.0. Which is a very different result from the TPU article.
So my options seem to be:
1) Upgrade to PCIe 3.0 x16/x16
2) Wait for PCIe 4.0 and do a new x8/x8
3) Wait for PCIe 4.0 and do x16/x16
I'm leaning to 2 or 3 since the game won't be out until likely late next year so that will be right around SkyLake. Any thoughts from my fellow enthusiasts?
So I'm planning my system for the game Star Citizen. According to the Devs it's going to be pretty heavy physics wise. And it's going to be a PhysX enabled game. So my setup will be one current gen (at the time) high end card to run an Oculus Rift and a second card being a previous gen high end card to run PhysX.
My current setup is an i5-3570k so I'm not worried about CPU power, but with the limit at 8x lanes for 2 cards I'm wondering about PCIe bandwidth. I haven't been able to find anything about how much bandwidth is used by PhysX and I do know that higher FPS (which you kind of need for the Rift, it would seem 75-90fps) uses more bandwidth.
Now I have done a bit of Google searching and found an article at Tech PowerUp that shows scaling with a 680 that shows that x8 is fine and x4 shows only a 2% performance loss. And a article at Anand Tech with 2 Titans in SLI that shows up to a 7% boost going from x16/x16 PCIe 2.0 to 3.0. Which is a very different result from the TPU article.
So my options seem to be:
1) Upgrade to PCIe 3.0 x16/x16
2) Wait for PCIe 4.0 and do a new x8/x8
3) Wait for PCIe 4.0 and do x16/x16
I'm leaning to 2 or 3 since the game won't be out until likely late next year so that will be right around SkyLake. Any thoughts from my fellow enthusiasts?