Concurrent Data Transfer on GTX480

Accelerated Computing CUDA CUDA Programming and Performance
1

One of my awaiting feature from the Fermi is the Concurrent Data Transfers that is stated in NVIDIA CUDA Programming guide
“Devices of compute capability 2.0 can perform a copy from page-locked host memory to device memory concurrently with a copy from device memory to page-locked host memory.”

But right now when I try this feature on my GTX 480, i could not make this happen. Is there any other condition for Concurrent Data Transfer or simply GTX 480 is incapable of doing that.

Is that a bug, if it is then when the bug fixes are released.
If some one could make it happen, could you please post a snap of the code.

Thank you for the information

2

According to Sumit here:

[url=“http://forums.nvidia.com/index.php?showtopic=165055&mode=threaded&pid=1073830”]The Official NVIDIA Forums | NVIDIA

The GeForce series only has 1 DMA engine, and so cannot do the bidirectional transfer. Only the Fermi-based Tesla cards can do this.