Hello,

I just noticed a weird behavior of s@h 4.02 when run with BOINC 4.02: every newly started WU directly report that approx. 50% (prog=0.5xyz) of the work has already been done for the WU. It didn't happen with s@h 3.20.

I'm running BOINC on an SMP machine, with two s@h processes regularly started by BOINC.

The completion time for a given WU isn't different from what it was with 3.2, except that now i immediately process the first 50% then it takes twice the time to preccess the second half, resulting in the same time for one WU. It seems to me like an error, but I can't figure out the problem.

What is happening ? It is ruinning my stats ;-)

Hoping I've been clear enough...

---

--
Olivier Mehani