Am 11.12.2012 um 22:14 schrieb Joseph Farran: > Greetings. > > How do I request the CPU type in qrsh / qsub with SGE 8.1.2? > > Googling this question shows some answers of the type "qrsh -l arch=xxx. > However, all my nodes in my qhost shows the same type of arch: > > # qhost -F | grep arch > hl:arch=lx-amd64 > hl:arch=lx-amd64 > hl:arch=lx-amd64 > hl:arch=lx-amd64 > hl:arch=lx-amd64 > ... > > This site http://dsavas.staff.shef.ac.uk/teaching/sge.pdf talks about using: > > -l arch=intel* > -l arch=amd*
I assume they overwrote simply the default setting in the exechost definition (`qconf -me node01` => complex_values arch=foobar). But I wouldn't suggest to do so, but to introduce a new complex "mtype" or whatever and fill it with an appropriate value (it could even be put in a load sensor to set it automatically and there might be more possibilities, whether the machines have sse4 or not and for fma3/4). Some packages like Gaussian provide different binaries for each CPU type and depending on your jobscript you could either request a preferred one or chose the right one automatically during execution (if you put your binaries in congruent directories where you select the type of CPU at an upper level). You can use a modified `arch` script and put its output in the path to the binaries in your jobscript for this. -- Reuti > What am I missing in my GE setup to have the architecture type fill-in > correctly? > > Here is qhost -F of one of my Intel nodes: > compute-19-2 linux-x64 4 0.01 23.7G 434.3M 7.8G 0.0 > hl:arch=linux-x64 > hl:num_proc=4.000000 > hl:mem_total=23.656G > hl:swap_total=7.812G > hl:virtual_total=31.468G > hl:load_avg=0.010000 > hl:load_short=0.050000 > hl:load_medium=0.010000 > hl:load_long=0.000000 > hl:mem_free=23.232G > hl:swap_free=7.812G > hl:virtual_free=31.044G > hl:mem_used=434.332M > hl:swap_used=0.000 > hl:virtual_used=434.332M > hl:cpu=0.100000 > hl:m_topology=SCCSCC > hl:m_topology_inuse=SCCSCC > hl:m_socket=2.000000 > hl:m_core=4.000000 > hl:np_load_avg=0.002500 > hl:np_load_short=0.012500 > hl:np_load_medium=0.002500 > hl:np_load_long=0.000000 > hl:cores_in_use=0.000000 > > Thanks, > Joseph > _______________________________________________ > users mailing list > [email protected] > https://gridengine.org/mailman/listinfo/users _______________________________________________ users mailing list [email protected] https://gridengine.org/mailman/listinfo/users
