Combined Sieve

[henryzz]

Would it be possible for min-merit in gap_test.py to default to whatever was used in the stats call?

[SethTro]

Originally Posted by henryzz

Would it be possible for min-merit in gap_test.py to default to whatever was used in the stats call?

A great suggestion; Implemented and available (after you `git pull`)

Originally Posted by robert44444uk

Hi Seth, a couple of questions

• Do the --prp-top-percent and --min-merit handles still work?
• Will the program pick up automatically after either ctrl c is pressed, or the power outage, or other automated system halts come into play? If so, is the command above the correct one to use?

I added these options to `misc/run.sh` (as -p and -m)

I should write some documentation for `misc/run.sh` but I'm tired tonight; I think the best way to think of run.sh is a convient way of making sure that `gap_stats` gets run directly after `combined_sieve` and then `gap_test.py` after that. It doesn't do anything special or additional.

[SethTro]

I made a theoretical inspection to modulo_search and found two ~10% improvements so combined_sieve for deep sieves should be ~20-30% faster.

If anyone had an Intel CPU I'd be curious to see the output before and after `git pull` from

$ make benchmark
 $ ./benchmark 1000000 modulo_search_euclid_all_small

I've also added

$ make benchmark_google
$ ./benchmark_google

Which gives similar output

 
----------------------------------------------------------------------------------------
Benchmark                                              Time             CPU   Iterations
----------------------------------------------------------------------------------------
BM_module_search_euclid/25/15000                    50.2 ns         50.2 ns     13925674
BM_module_search_euclid/25/100000                   34.3 ns         34.3 ns     20490547
BM_module_search_euclid/35/15000                     109 ns          109 ns      6432297
BM_module_search_euclid/45/15000                     174 ns          174 ns      4020660
BM_module_search_euclid/60/15000                     285 ns          285 ns      2462220
BM_module_search_euclid/60/100000                    264 ns          264 ns      2633543
BM_module_search_euclid_stack/25/15000              49.8 ns         49.8 ns     14064605
BM_module_search_euclid_stack/25/100000             34.5 ns         34.5 ns     20435383
BM_module_search_euclid_stack/35/15000               112 ns          112 ns      6307416
BM_module_search_euclid_stack/45/15000               176 ns          176 ns      3989369
BM_module_search_euclid_stack/60/15000               286 ns          285 ns      2450658
BM_module_search_euclid_stack/60/100000              265 ns          265 ns      2607589

Which produces similar results. It's slightly more accurate but requires installing google/benchmark (apt install libbenchmark-dev)