File sortreduce.cu compiled on Mar 23 2015 at 05:06:38
Device name: Quadro K2100M
CUDA Capability version: 3.0
CUDA Driver / Runtime Version: 7,000/7,000, CUDA_VERSION: 7,000
Thrust v1.8
TIME thru Starting                           : CPU:  0.007 (d= 0.007)
Some redundant ops are done for timing purposes.
N= 100,000,000

Comparing sort on 4 types of memory:
TIME thru Sync                               : CPU:  1.151 (d= 1.144)

Host:
TIME thru Create int vec on host             : CPU:  1.198 (d= 0.047)
TIME thru Transform rand int vec on host     : CPU:  1.359 (d= 0.161)
TIME thru Sort on host                       : CPU:  3.150 (d= 1.791)
TIME thru Sync                               : CPU:  3.150 (d= 0.000)

Device:
Before creating device vector: cuda_mem_tot= 2,147,287,040, cuda_mem_free= 1,250,750,464
TIME thru Before create int vec on device    : CPU:  3.150 (d= 0.000)
TIME thru Create int vec on device           : CPU:  3.635 (d= 0.485)
After creating device vector: cuda_mem_tot= 2,147,287,040, cuda_mem_free= 1,250,750,464
TIME thru Copy host to device                : CPU:  3.698 (d= 0.062)
TIME thru Transform rand int vec on device   : CPU:  4.439 (d= 0.741)
TIME thru Sort on device                     : CPU: 11.076 (d= 6.638)
TIME thru Copy device to host                : CPU: 11.139 (d= 0.062)

Mapped:
Starting mapped test: cuda_mem_tot= 2,147,287,040, cuda_mem_free= 1,250,750,464
TIME thru cudaHostAllocMapped                : CPU: 11.242 (d= 0.103)
TIME thru Transform mapped array             : CPU: 11.980 (d= 0.737)
TIME thru sort mapped array                  : CPU: 18.708 (d= 6.729)
TIME thru free and sync                      : CPU: 18.791 (d= 0.083)

Managed:
Starting managed test: cuda_mem_tot= 2,147,287,040, cuda_mem_free= 1,249,939,456
TIME thru cudaMallocManaged                  : CPU: 18.797 (d= 0.006)
after cudaMallocManaged: cuda_mem_tot= 2,147,287,040, cuda_mem_free= 849,608,704
TIME thru Transform managed array            : CPU: 20.889 (d= 2.092)
TIME thru Transform managed device array     : CPU: 21.607 (d= 0.718)
TIME thru sort managed array                 : CPU: 28.227 (d= 6.619)
TIME thru sort managed device array          : CPU: 34.076 (d= 5.850)
TIME thru free and sync                      : CPU: 34.118 (d= 0.042)

Comparing reduce on 4 types of memory:
TIME thru Starting testreduce                : CPU: 34.119 (d= 0.001)

Host:
TIME thru Create float vec on host           : CPU: 34.166 (d= 0.047)
TIME thru Transform rand float vec on host   : CPU: 34.435 (d= 0.269)
TIME thru Generate rand float vec on host    : CPU: 36.455 (d= 2.020)
sumh= 36,028,797,018,963,968.000
TIME thru Reduce on host                     : CPU: 36.568 (d= 0.113)

Device:
TIME thru Create float vec on device         : CPU: 36.979 (d= 0.411)
TIME thru Transform rand float vec on device : CPU: 37.496 (d= 0.516)
sumd= 4,941,592,339,152,896.000
TIME thru Reduce on device                   : CPU: 37.664 (d= 0.168)

Mapped:
TIME thru cudaHostAllocMapped                : CPU: 37.788 (d= 0.124)
TIME thru Transform rand float vec on mapped : CPU: 38.298 (d= 0.511)
sumd= 4,941,592,339,152,896.000
TIME thru Reduce on mapped                   : CPU: 38.465 (d= 0.167)

Managed:
Starting managed test: cuda_mem_tot= 2,147,287,040, cuda_mem_free= 1,283,850,240
TIME thru cudaMallocManaged                  : CPU: 38.471 (d= 0.006)
after cudaMallocManaged: cuda_mem_tot= 2,147,287,040, cuda_mem_free= 883,818,496
TIME thru Transform managed array            : CPU: 41.104 (d= 2.633)
TIME thru Transform managed device array     : CPU: 41.568 (d= 0.463)
sumd= 4,941,592,339,152,896.000
TIME thru Reduce on managed                  : CPU: 41.742 (d= 0.174)
TIME thru free and sync                      : CPU: 41.782 (d= 0.041)