Performance Studies using PSTSWM

Intel Paragon Protocol Performance

(transpose LT experiment I-A1 / O(P) sendrecv transpose algorithm)

1x16 Processors / Problem T10L16

Runtime Statistics min (mean-min)/min (median-min)/min (max-min)/min   3.7882e-01    0.12    0.11    0.29

Three Fastest Protocols 1st 2nd 3rd   e3    e2    c2         Number of Proctocols With Runtimes Within X% of Min 1% 5% 25%   1    3    27

1x8 Processors / Problem T21L8

Runtime Statistics min (mean-min)/min (median-min)/min (max-min)/min   1.0448e+00    0.02    0.02    0.06

Three Fastest Protocols 1st 2nd 3rd   a6    c2    a2         Number of Proctocols With Runtimes Within X% of Min 1% 5% 25%   8    28    29

1x16 Processors / Problem T21L16

Runtime Statistics min (mean-min)/min (median-min)/min (max-min)/min   1.1381e+00    0.03    0.03    0.08

Three Fastest Protocols 1st 2nd 3rd   a6    a2    a1         Number of Proctocols With Runtimes Within X% of Min 1% 5% 25%   5    22    29

1x32 Processors / Problem T21L32

Runtime Statistics min (mean-min)/min (median-min)/min (max-min)/min   1.3735e+00    0.05    0.04    0.12

Three Fastest Protocols 1st 2nd 3rd   a3    a2    a1         Number of Proctocols With Runtimes Within X% of Min 1% 5% 25%   4    17    29

1x8 Processors / Problem T42L16

Runtime Statistics min (mean-min)/min (median-min)/min (max-min)/min   9.7564e+00    0.02    0.02    0.05

Three Fastest Protocols 1st 2nd 3rd   b6    b5    d6         Number of Proctocols With Runtimes Within X% of Min 1% 5% 25%   8    28    29

DISCUSSION

Partitions of the Paragon processor grid were used that match the processor subsets that the parallel algorithms would run on in a two dimensional data decomposition: 1x8, 1x16, and 1x32.

---

Patrick H. Worley / ( worleyph@ornl.gov)
Last Modified Monday, 15-Jul-2002 10:30:05 EDT.