Search of 17clues tests band 6 analysis

http://pagesperso-orange.fr/gpenet/SK17TEST/test.htm

The band 6 has been the first expanded for several reasons

 

40 17s with the distribution 665 are known, a big count for low indexes

 

Band index 6 has a huge number of bands 2 and bands 3 attached

           98 249 555 bands 3, 1.8% of the total

           5 417 020 bands 2 for the first entry 0.89% of the total

           4 790 632 bands 2 for the second entry.

 

The full expansion for this band 1 has been disturbed by power failures and windiws updates, so we have the batches summary on part of the process only.

 

With band 1, the 2X3Y has not been called in the summary collected.

 

With more than 18 bands3 per band 2 in average, the run time per solution grid is good in this area.

 

As far as I can see, the XY count is in the range 35-40 millions for this band 1.

The count is very similar for both entry. A higher count in 656 mode for entry 2 covers the 566 count in entry 1.

 

And the run time is very similar as well for both entries, may be slightly higher for the entry file 2, but as batches has been done on different computers, the comparison is not highly reliable.

 

Run times seen vary significantly from low band2 index to high band2 index. To make it simple, we can take the range 1.5—4.5 seconds per band 2 (giving 3s per band 2 in average) using my mix of cores.

 

This is for one entry file and 18 bands 3.

 

At the end, the throughput in this area seems to be around 3 bands 3 per second, may be

 

Using this as average for the whole process, we come to around 60 core*year of processing far below the check that no 16 exists done by Gary Mc Guire, but far above the power that I can deliver.

 

A better view of the future run time estimate will come in one or 2 months.