dddh@node9 ~/g/cusp/performance/spmv $ nvcc -O2 -o spmv spmv.cu -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=\"sm_35,compute_35\" -I/home/dddh/g -I../../

dddh@node9 ~/g/cusp/performance/spmv $ time ./spmv ~/g/m/pwtk.mtx --value_type=double --device=0

Computing SpMV with 'double' values.

There are 2 devices supporting CUDA

Device 0: "GeForce GTX TITAN"
  Major revision number:                         3
  Minor revision number:                         5
  Total amount of global memory:                 6442123264 bytes

Device 1: "GeForce GTX 680"
  Major revision number:                         3
  Minor revision number:                         0
  Total amount of global memory:                 2147287040 bytes

Running on Device 0

Read matrix (/home/dddh/g/m/pwtk.mtx) with shape (217918,217918) and 11634424 entries

        coo_flat            :   2.2136 ms ( 10.51 GFLOP/s 127.7 GB/s) [L2 error 0.000000]
        coo_flat_tex        :   1.8492 ms ( 12.58 GFLOP/s 152.9 GB/s) [L2 error 0.000000]
        csr_scalar          :   6.5901 ms (  3.53 GFLOP/s  36.1 GB/s) [L2 error 0.000000]
        csr_scalar_tex      :   5.3945 ms (  4.31 GFLOP/s  44.1 GB/s) [L2 error 0.000000]
        csr_vector          :   2.0344 ms ( 11.44 GFLOP/s 116.9 GB/s) [L2 error 0.000000]
        csr_vector_tex      :   2.0404 ms ( 11.40 GFLOP/s 116.6 GB/s) [L2 error 0.000000]
        Refusing to convert to DIA format
        Refusing to convert to ELL format
        hyb                 :   1.0365 ms ( 22.45 GFLOP/s 229.7 GB/s) [L2 error 0.000000]
        hyb_tex             :   1.1736 ms ( 19.83 GFLOP/s 202.9 GB/s) [L2 error 0.000000]

real    0m16.576s
user    0m8.032s
sys     0m8.476s
dddh@node9 ~/g/cusp/performance/spmv $ time ./spmv ~/g/m/pwtk.mtx --value_type=double --device=1

Computing SpMV with 'double' values.

There are 2 devices supporting CUDA

Device 0: "GeForce GTX TITAN"
  Major revision number:                         3
  Minor revision number:                         5
  Total amount of global memory:                 6442123264 bytes

Device 1: "GeForce GTX 680"
  Major revision number:                         3
  Minor revision number:                         0
  Total amount of global memory:                 2147287040 bytes

Running on Device 1

Read matrix (/home/dddh/g/m/pwtk.mtx) with shape (217918,217918) and 11634424 entries

        coo_flat            :   2.8742 ms (  8.10 GFLOP/s  98.4 GB/s) [L2 error 0.000000]
        coo_flat_tex        :   2.4613 ms (  9.45 GFLOP/s 114.9 GB/s) [L2 error 0.000000]
        csr_scalar          :  18.2128 ms (  1.28 GFLOP/s  13.1 GB/s) [L2 error 0.000000]
        csr_scalar_tex      :  18.3211 ms (  1.27 GFLOP/s  13.0 GB/s) [L2 error 0.000000]
        csr_vector          :   3.1678 ms (  7.35 GFLOP/s  75.1 GB/s) [L2 error 0.000000]
        csr_vector_tex      :   3.1189 ms (  7.46 GFLOP/s  76.3 GB/s) [L2 error 0.000000]
        Refusing to convert to DIA format
        Refusing to convert to ELL format
        hyb                 :   2.0665 ms ( 11.26 GFLOP/s 115.2 GB/s) [L2 error 0.000000]
        hyb_tex             :   1.7925 ms ( 12.98 GFLOP/s 132.8 GB/s) [L2 error 0.000000]

real    0m19.464s
user    0m7.552s
sys     0m11.900s
dddh@node2:~/g/m> time ~/spmv ~/g/m/pwtk.mtx --value_type=double

Computing SpMV with 'double' values.

There is 1 device supporting CUDA

Device 0: "GeForce GTX 580"
  Major revision number:                         2
  Minor revision number:                         0
  Total amount of global memory:                 1610153984 bytes

Running on Device 0

Read matrix (/home/dddh/g/m/pwtk.mtx) with shape (217918,217918) and 11634424 entries

        coo_flat            :   3.1945 ms (  7.28 GFLOP/s  88.5 GB/s) [L2 error 0.000000]
        coo_flat_tex        :   2.7491 ms (  8.46 GFLOP/s 102.8 GB/s) [L2 error 0.000000]
        csr_scalar          :  14.6018 ms (  1.59 GFLOP/s  16.3 GB/s) [L2 error 0.000000]
        csr_scalar_tex      :   7.2921 ms (  3.19 GFLOP/s  32.6 GB/s) [L2 error 0.000000]
        csr_vector          :   2.0886 ms ( 11.14 GFLOP/s 113.9 GB/s) [L2 error 0.000000]
        csr_vector_tex      :   1.8733 ms ( 12.42 GFLOP/s 127.0 GB/s) [L2 error 0.000000]
        Refusing to convert to DIA format
        Refusing to convert to ELL format
        hyb                 :   1.6643 ms ( 13.98 GFLOP/s 143.1 GB/s) [L2 error 0.000000]
        hyb_tex             :   1.9237 ms ( 12.10 GFLOP/s 123.8 GB/s) [L2 error 0.000000]

real    0m30.359s
user    0m18.456s
sys     0m11.724s