Open GoogleCodeExporter opened 8 years ago
A Core2 Duo 2.53GHz Laptop will generate (during a single continuous session)
1000 flavours of anaemia in 7 minutes (1 every 0.46s), another 1000 in 8
minutes (1 every 0.48s) and a further 1500 in 16 minutes (1 every 0.64s).
Compute performance in a single session tails off rapidly beyond around 3000
items generated, presumably because the Java heap is approaching full: the heap
finally explodes somewhere after 4000 items generated in one session.
Performance metrics appear to be the same for each new SnoFyre session, and is
not obviously degraded if modest quantities of previously generated data
already exists in the backend DB - currently tested up to 54k unique
expressions, created over multiple different sessions generating 3500 items
each. Metrics suggest performance, if heap problem was fixed, may be a
sustained generation rate of around 8,000 patients per hour compute time.
Original comment by jeremy.r...@googlemail.com
on 20 Jan 2011 at 10:48
Original issue reported on code.google.com by
jay.kola
on 16 Jan 2011 at 1:49