Difference between revisions of "CudaVsNested"
From Ghoulwiki
Ghoulsblade (talk | contribs) |
Ghoulsblade (talk | contribs) |
||
Line 1: | Line 1: | ||
− | <pre> | + | <pre>N=32768 |
− | N= | + | SX=2048 |
− | SX= | + | SY=128 |
− | SY= | + | SZ=8 |
− | SZ= | + | I0=16 |
− | I0= | + | DATASIZE_IN_RAW=1152kb |
− | DATASIZE_IN_RAW= | + | DATASIZE_IN_STATE=128kb |
− | DATASIZE_IN_STATE= | + | DATASIZE_IN_INDEX=18kb |
− | DATASIZE_IN_INDEX= | + | DATASIZE_IN_TOTAL=1298kb |
− | DATASIZE_IN_TOTAL= | ||
DATASIZE_OUT_TOTAL=65536kb | DATASIZE_OUT_TOTAL=65536kb | ||
assert passed : (N / kThreadBlockSize <= 63*1024 && "grid_size larger than suppo | assert passed : (N / kThreadBlockSize <= 63*1024 && "grid_size larger than suppo | ||
rted (cudaGetDeviceProperties:maxGridSize: 63k currently)") | rted (cudaGetDeviceProperties:maxGridSize: 63k currently)") | ||
− | ReadTextData data/Corel_ColorMoments_9d.ascii : | + | ReadTextData data/Corel_ColorMoments_9d.ascii : 32768 lines of real data, added |
0 lines of random data | 0 lines of random data | ||
0.00 sec : reading data from file | 0.00 sec : reading data from file | ||
Line 19: | Line 18: | ||
assert passed : (INDEXPOS_2(I0-1,I0-1,I0) == INDEX_END-1) | assert passed : (INDEXPOS_2(I0-1,I0-1,I0) == INDEX_END-1) | ||
assert passed : (sz < 255) | assert passed : (sz < 255) | ||
− | -2. | + | -2.453896,-2.369326,-2.369326,...,3.864799 |
− | 0. | + | 0.04 sec : generating index data |
0.05 sec : allocate and init device mem | 0.05 sec : allocate and init device mem | ||
− | + | 1.21 sec : exec kernel on device | |
0.00 sec : receive results from device | 0.00 sec : receive results from device | ||
− | atom[0]= | + | atom[0]=21875 atom[1]=0 iNumResults=21875 kMaxResults=8388608 |
check : sequential... | check : sequential... | ||
− | check:iNumResults= | + | check:iNumResults=21959 maxc=26 for i=17803 |
− | + | 22.53 sec : check : sequential | |
− | gpu/cpu= 0.1 ok err<=0. | + | gpu/cpu= 0.1 ok err<=0.4% N=32768 size=1MB IO=16 i3:0,0 tgpu=1.2s tcpu=22. |
− | + | 5 | |
Press ENTER to exit... | Press ENTER to exit... | ||
− | |||
</pre> | </pre> |
Latest revision as of 18:24, 17 October 2007
N=32768 SX=2048 SY=128 SZ=8 I0=16 DATASIZE_IN_RAW=1152kb DATASIZE_IN_STATE=128kb DATASIZE_IN_INDEX=18kb DATASIZE_IN_TOTAL=1298kb DATASIZE_OUT_TOTAL=65536kb assert passed : (N / kThreadBlockSize <= 63*1024 && "grid_size larger than suppo rted (cudaGetDeviceProperties:maxGridSize: 63k currently)") ReadTextData data/Corel_ColorMoments_9d.ascii : 32768 lines of real data, added 0 lines of random data 0.00 sec : reading data from file assert passed : (INDEXPOS_0(I0) == INDEXSTART_1-1) assert passed : (INDEXPOS_1(I0-1,I0) == INDEXSTART_2-1) assert passed : (INDEXPOS_2(I0-1,I0-1,I0) == INDEX_END-1) assert passed : (sz < 255) -2.453896,-2.369326,-2.369326,...,3.864799 0.04 sec : generating index data 0.05 sec : allocate and init device mem 1.21 sec : exec kernel on device 0.00 sec : receive results from device atom[0]=21875 atom[1]=0 iNumResults=21875 kMaxResults=8388608 check : sequential... check:iNumResults=21959 maxc=26 for i=17803 22.53 sec : check : sequential gpu/cpu= 0.1 ok err<=0.4% N=32768 size=1MB IO=16 i3:0,0 tgpu=1.2s tcpu=22. 5 Press ENTER to exit...