Doing some test runs for rough comparisons.

Gaussian

Running one small example using g03 …

• swallowtail/petaltail (N=2, imw) 65 mins
• sharptail (N=2, bss12) 121 mins
• sharptail (N=8!, bss12, so across nodes) – bad idea, does not work

Amber sander.MPI

Running two small benchmarks (jac and factor_ix)

• swallowtail/petaltail (N=2, imw) – 2:50 and 4:47 (mins:secs)
• sharptail (N=2, bss12) – 3:52 and 7:08 (mins:secs)
• sharptail (N=8!, bss12, so across nodes) – 4:30 and 5:17 (mins:secs)

Amber pmemd

Running one program (sdixit example)

• swallowtail/petaltail (N=8, imw) – 1:05 (mins:secs)
• sharptail (N=2, bss12) – 3:49 (mins:secs)
• sharptail (N=8!, bss12, so across nodes) – 6:13 (mins:secs)

Back