Hi,
I have cuda 3.2 running and tested different Hardware (C1060,C2070,GTX480,GTX580) with the bandwidthTest (internal throughput).
Why the C2070 is underperforming ?? I switched ECC off.
card Theoretical Bandwidth measured value percent
C1060 512/8 byte20.8GHz=102GB/s 73 72
C2070 384/8 byte21.5GHz=144GB/s 90 63
GTX480 384/8 byte21.8GHz=173GB/s 118 68
GTX580 384/8 byte22.0GHz=192GB/s 140 73
Why the C2070 is so bad. Is the memory clock really 1.5 Ghz ???
Thanks.