GPU

GPU cluster We just wanted to let everyone in this list know that the NSF CRI GPU servers (qa-xp-001 - qa-xp-008) are up and running.  They are set up as compute nodes.  You would normally submit jobs to these servers using the following queue in your job script:

#$ -q gpu@@csecri-titanxp

For code development and/or testing, you can also run "interactively" on one of these nodes using qrsh command.  Below is an example on how to run caffe interactively on one of the qa-xp- nodes.  Please contact us if any problems are encountered or if you have any questions.

thanks, --Dodi

[dheryadi@crcfe02 ~]$ qrsh -q gpu@@csecri-titanxp
[dheryadi@qa-xp-001 ~]$ 
[dheryadi@qa-xp-001 ~]$ module load caffe/2016-gpu
[dheryadi@qa-xp-001 ~]$ caffe device_query -gpu all
I1014 12:47:20.131463 18640 caffe.cpp:112] Querying GPUs all
I1014 12:47:20.348227 18640 common.cpp:177] Device id:                     0
I1014 12:47:20.348317 18640 common.cpp:178] Major revision number:         6
I1014 12:47:20.348331 18640 common.cpp:179] Minor revision number:         1
I1014 12:47:20.348345 18640 common.cpp:180] Name:                          TITAN X (Pascal)
I1014 12:47:20.348379 18640 common.cpp:181] Total global memory:           12781551616
I1014 12:47:20.348398 18640 common.cpp:182] Total shared memory per block: 49152
I1014 12:47:20.348409 18640 common.cpp:183] Total registers per block:     65536
I1014 12:47:20.348420 18640 common.cpp:184] Warp size:                     32
I1014 12:47:20.348440 18640 common.cpp:185] Maximum memory pitch:          2147483647
I1014 12:47:20.348461 18640 common.cpp:186] Maximum threads per block:     1024
I1014 12:47:20.348481 18640 common.cpp:187] Maximum dimension of block:    1024, 1024, 64
I1014 12:47:20.348496 18640 common.cpp:190] Maximum dimension of grid:     2147483647, 65535, 65535
I1014 12:47:20.348510 18640 common.cpp:193] Clock rate:                    1531000
I1014 12:47:20.348526 18640 common.cpp:194] Total constant memory:         65536
I1014 12:47:20.348539 18640 common.cpp:195] Texture alignment:             512
I1014 12:47:20.348554 18640 common.cpp:196] Concurrent copy and execution: Yes
I1014 12:47:20.348570 18640 common.cpp:198] Number of multiprocessors:     28
I1014 12:47:20.348585 18640 common.cpp:199] Kernel execution timeout:      No
I1014 12:47:21.089072 18640 common.cpp:177] Device id:                     1
I1014 12:47:21.089128 18640 common.cpp:178] Major revision number:         6
I1014 12:47:21.089141 18640 common.cpp:179] Minor revision number:         1
I1014 12:47:21.089153 18640 common.cpp:180] Name:                          TITAN X (Pascal)
I1014 12:47:21.089205 18640 common.cpp:181] Total global memory:           12781551616
I1014 12:47:21.089224 18640 common.cpp:182] Total shared memory per block: 49152
I1014 12:47:21.089236 18640 common.cpp:183] Total registers per block:     65536
I1014 12:47:21.089246 18640 common.cpp:184] Warp size:                     32
I1014 12:47:21.089256 18640 common.cpp:185] Maximum memory pitch:          2147483647
I1014 12:47:21.089267 18640 common.cpp:186] Maximum threads per block:     1024
I1014 12:47:21.089295 18640 common.cpp:187] Maximum dimension of block:    1024, 1024, 64
I1014 12:47:21.089309 18640 common.cpp:190] Maximum dimension of grid:     2147483647, 65535, 65535
I1014 12:47:21.089321 18640 common.cpp:193] Clock rate:                    1531000
I1014 12:47:21.089344 18640 common.cpp:194] Total constant memory:         65536
I1014 12:47:21.089359 18640 common.cpp:195] Texture alignment:             512
I1014 12:47:21.089388 18640 common.cpp:196] Concurrent copy and execution: Yes
I1014 12:47:21.089411 18640 common.cpp:198] Number of multiprocessors:     28
I1014 12:47:21.089432 18640 common.cpp:199] Kernel execution timeout:      No
I1014 12:47:21.721165 18640 common.cpp:177] Device id:                     2
I1014 12:47:21.721210 18640 common.cpp:178] Major revision number:         6
I1014 12:47:21.721218 18640 common.cpp:179] Minor revision number:         1
I1014 12:47:21.721225 18640 common.cpp:180] Name:                          TITAN X (Pascal)
I1014 12:47:21.721232 18640 common.cpp:181] Total global memory:           12781551616
I1014 12:47:21.721268 18640 common.cpp:182] Total shared memory per block: 49152
I1014 12:47:21.721274 18640 common.cpp:183] Total registers per block:     65536
I1014 12:47:21.721282 18640 common.cpp:184] Warp size:                     32
I1014 12:47:21.721288 18640 common.cpp:185] Maximum memory pitch:          2147483647
I1014 12:47:21.721295 18640 common.cpp:186] Maximum threads per block:     1024
I1014 12:47:21.721302 18640 common.cpp:187] Maximum dimension of block:    1024, 1024, 64
I1014 12:47:21.721308 18640 common.cpp:190] Maximum dimension of grid:     2147483647, 65535, 65535
I1014 12:47:21.721314 18640 common.cpp:193] Clock rate:                    1531000
I1014 12:47:21.721321 18640 common.cpp:194] Total constant memory:         65536
I1014 12:47:21.721365 18640 common.cpp:195] Texture alignment:             512
I1014 12:47:21.721376 18640 common.cpp:196] Concurrent copy and execution: Yes
I1014 12:47:21.721390 18640 common.cpp:198] Number of multiprocessors:     28
I1014 12:47:21.721403 18640 common.cpp:199] Kernel execution timeout:      No
I1014 12:47:22.377786 18640 common.cpp:177] Device id:                     3
I1014 12:47:22.377833 18640 common.cpp:178] Major revision number:         6
I1014 12:47:22.377841 18640 common.cpp:179] Minor revision number:         1
I1014 12:47:22.377847 18640 common.cpp:180] Name:                          TITAN X (Pascal)
I1014 12:47:22.377856 18640 common.cpp:181] Total global memory:           12781551616
I1014 12:47:22.377863 18640 common.cpp:182] Total shared memory per block: 49152
I1014 12:47:22.377895 18640 common.cpp:183] Total registers per block:     65536
I1014 12:47:22.377902 18640 common.cpp:184] Warp size:                     32
I1014 12:47:22.377912 18640 common.cpp:185] Maximum memory pitch:          2147483647
I1014 12:47:22.377918 18640 common.cpp:186] Maximum threads per block:     1024
I1014 12:47:22.377925 18640 common.cpp:187] Maximum dimension of block:    1024, 1024, 64
I1014 12:47:22.377933 18640 common.cpp:190] Maximum dimension of grid:     2147483647, 65535, 65535
I1014 12:47:22.377938 18640 common.cpp:193] Clock rate:                    1531000
I1014 12:47:22.377949 18640 common.cpp:194] Total constant memory:         65536
I1014 12:47:22.377974 18640 common.cpp:195] Texture alignment:             512
I1014 12:47:22.377996 18640 common.cpp:196] Concurrent copy and execution: Yes
I1014 12:47:22.378016 18640 common.cpp:198] Number of multiprocessors:     28
I1014 12:47:22.378023 18640 common.cpp:199] Kernel execution timeout:      No
[dheryadi@qa-xp-001 ~]$ exit
logout
[dheryadi@crcfe02 ~]$