Bases 33-100 reservations/statuses/primes

[rogue]

I am sieving the next 52000 k, but only to 1e10. They should be sieved to 1e11, but I won't be able to do that before you complete the current range. Part of me would like to sieve the remaining k to 1e10 as well and have BOINC do the testing even though the ranges won't be sieved deep enough. That way the entire range could be completed to n=25000 in July.

If anyone wants to take ranges of 1000 k and sieve to 1e11 I would more than willing to give them away. It should take about 10 days to sieve a range that size to 1e11. Sieving is better for non-AVX machines since the sieving software doesn't have AVX code. In other words an AVX machine will sieve at about the same speed as a non-AVX machine, but an AVX machine will be about twice as fast as a non-AVX machine for PRP testing.

[henryzz]

Originally Posted by rogue

I am sieving the next 52000 k, but only to 1e10. They should be sieved to 1e11, but I won't be able to do that before you complete the current range. Part of me would like to sieve the remaining k to 1e10 as well and have BOINC do the testing even though the ranges won't be sieved deep enough. That way the entire range could be completed to n=25000 in July.

If anyone wants to take ranges of 1000 k and sieve to 1e11 I would more than willing to give them away. It should take about 10 days to sieve a range that size to 1e11. Sieving is better for non-AVX machines since the sieving software doesn't have AVX code. In other words an AVX machine will sieve at about the same speed as a non-AVX machine, but an AVX machine will be about twice as fast as a non-AVX machine for PRP testing.

We just need to find someone who can upgrade srxsieve to AVX.

[rogue]

Originally Posted by henryzz

We just need to find someone who can upgrade srxsieve to AVX.

Agreed, but it it not likely to be me. I am not good at x86 assembly.

[grueny]

r72 tested to n=900e3 ; no prime

continuing to n=1e6.

[henryzz]

Originally Posted by rogue

Agreed, but it it not likely to be me. I am not good at x86 assembly.

Is there anyone at primegrid who could do it? They use sr2sieve pretty heavily. When I had a look it was mostly just parallel mulmods which shouldn't be that hard for someone who knows what they are doing.

[rebirther]

S39 tested to n=100k (50-100k)

88 primes found - 259 remain

Results emailed - Base released

[rebirther]

Reserving R78 to n=100k (63-100k) for BOINC

[Batalov]

You may want to rebuild your LLR with gwnum 28.6 and push it to clients with the new tasks. This is important to do as early as possible because your BOINC throughput is substantial and every test run with old binary is another tiny chance to miss a prime.

(You just replace gwnum folder in LLR with gwnum folder from p95src version 28.6, then edit two instances of the string 28.5 in the lmenu.c and Llr.h file and recompile.)

[rebirther]

Originally Posted by Batalov

You may want to rebuild your LLR with gwnum 28.6 and push it to clients with the new tasks. This is important to do as early as possible because your BOINC throughput is substantial and every test run with old binary is another tiny chance to miss a prime.

(You just replace gwnum folder in LLR with gwnum folder from p95src version 28.6, then edit two instances of the string 28.5 in the lmenu.c and Llr.h file and recompile.)

ok, I have rebuilt the 3.8.14 with gwnum 28.6 for windows. I need it also for linux32/64 and mac64 if someone can compile it. I will add the apps after all small current running bases are finished.

[rebirther]

R78 tested to n=100k (63-100k)

9 primes found - 63 remain

23462*78^69162-1
72638*78^70230-1
11412*78^72798-1
45661*78^73022-1
58412*78^83824-1
78053*78^84433-1
51476*78^88677-1
68571*78^91386-1
3633*78^94500-1

Results emailed - Base released

[rebirther]

R79 tested to n=50k (25-50k)

1491 primes found - 6120 remain

Results emailed - Base released