S2HAT: map2alm performance results:
Constant per-processor work load runs
In the figures below different colors refer to different values
of the theoretical per-processor work load as defined by nside^3/nproc parameter.
The dashed lines show the perfect scaling. The width of the lines is determined by the
difference between the slowest and the fastest processor and thus reflect the deviations
from the perfect load-balancing (and/or inherent differences of the processors).
Seaborg results
Jacquard results
Seaborg
Jacquard
not available yet
|