After looking into the problem you raised, we got the following:
First, our benchmark was done on a personal computer equipped with the CPU, Intel(R) Xeon(R) E5520 @ 2.27GHz, which can get about 540 credits every core-day, far higher than 200 credits.
Second, the "Recent average credit" in the sudoku statistics page is not correct as it shows.
When we look at the "Recent average credit" on top 20 of our project, the sum of the recent average credits is about 500000 per day.
However, according to the website BOINCstats, http://boincstats.com/stats/project_graph.php?pr=sudvt
our project has about 32,000,000 credits in total on 4/28, and has about 40,000,000 credits in total on 5/21.
Therefore, our project gets about 8000000/23 = 347826 credits per day, which is obviously below 500000.
We do not know why "Recent average credit" does not show as it actually is. (If you know this problem, please let us know.)
For example, the top one "PC4" (now using "syhan" instead) is actually not running anymore.
From the above two, the averaged cores per day is about 347826/540 = 644.
In other words, about 0.073% of the search space should be done per day, and about 0.51% per week.
In reality, the current rate we have now (from the "percentage" page) is about 0.05% per day.
For the difference between 0.073 and 0.05%, we observed that
(1) the first 0.3 billion sudoku grids take more time than those among the whole 5.4 billion, and
(2) the credit reports are not so consistently among different machines
(the reason could be that the machine benchmark test by BOINC are not consistent for our application).
Although some organizations in Taiwan promised to add more computing power to help,
we would appreciate your letting us know, if you know better way to advertise our BOINC project to speed up the project world-wide.
Thank you very much.