12-10-2011, 08:42 PM
The performance reported by the ockHascat-lite (I do not use the rest of the programs) as speed, is the same for all the GPU units. So it seems that the bus speed variation is insignificant. And you can do the math, assuming: A connection between any two PCIe devices is known as a link, and is built up from a collection of 1 or more lanes. BUT:
Capacity Per lane (each direction):
v1.x: 250 MB/s (2.5 GT/s)
v2.x: 500 MB/s (5 GT/s)
v3.0: 1 GB/s (8 GT/s)
v4.0: 2 GB/s (16 GT/s)
Furthermore:
16 lane slot (each direction):
v1.x: 4 GB/s (40 GT/s)
v2.x: 8 GB/s (80 GT/s)
v3.0: 16 GB/s (128 GT/s)
As you can see the fastest 2xGPU on one card at the curent time can not produce the MINIMAL speed of the 1 lane in a normal v2 connector. Maybe in some years we will have such a speedy cards, but not yet. And even in some years CPUs, GPUs, will not be so powerful as the capacity of a 16x connection. Maybe in the era of quantum computing. Just joking.
Ask any more question if you want.
Capacity Per lane (each direction):
v1.x: 250 MB/s (2.5 GT/s)
v2.x: 500 MB/s (5 GT/s)
v3.0: 1 GB/s (8 GT/s)
v4.0: 2 GB/s (16 GT/s)
Furthermore:
16 lane slot (each direction):
v1.x: 4 GB/s (40 GT/s)
v2.x: 8 GB/s (80 GT/s)
v3.0: 16 GB/s (128 GT/s)
As you can see the fastest 2xGPU on one card at the curent time can not produce the MINIMAL speed of the 1 lane in a normal v2 connector. Maybe in some years we will have such a speedy cards, but not yet. And even in some years CPUs, GPUs, will not be so powerful as the capacity of a 16x connection. Maybe in the era of quantum computing. Just joking.
Ask any more question if you want.