Sieve (for advanced users)

[geoff]

Originally Posted by BlisteringSheep

All I need is the source which includes the -y flag functionality.

I have just posted the source in the same directory as the binary.

The expected number of factors doesn't take into account the modifications, as Citirx noted above.

When large values of Y are used with the `sr2sieve -y Y' switch the time it takes for the program to respond to an interrupt signal may become excessive.

[Citrix]

I also noted that if y is large and speed > 2^31p/sec, then the program does not output anything.

But for y values less than 10, this is not important.

[ValerieVonck]

Any need for further sieving?
If so, with which y?

Thank you.

[Citrix]

Just looking for time per factor at different y values.

Note that if you sieved a range at y1 then there is no point on sieving the same range at y2, where y2>y1.

So using above need average time per factor for y=1 to y=12 and what machine was used for each.

edit: No need to reserve the whole trillion range for low y's. You can find the time over the first few factors.

[ValerieVonck]

Reserving 801T => 802T, y= 12

P4 - 2800 - 1Gb

[ValerieVonck]

Range done in under 15 minutes

05/25/07 15:09:42 WARNING: Only testing for factors p of the form p=1 (mod 4096).
05/25/07 15:09:42 sr2sieve started: 2 <= n <= 49999947, 801000000000000 <= p <= 802000000000000
05/25/07 15:22:43 sr2sieve stopped: at p=802000000000000 because range is complete.
05/25/07 15:22:43 Found factors for 0 terms in 748.295 cpu sec. (expected about 49.68)


Is it advisable to use the "Y" settings for "production" use?

[Citrix]

y=12 so you expect to find 1 factor, for every 2048 actual factors. So 2048/50=41T.

So you should do a range of 41T with y=12 to get a factor.
Now to get average time over 1-2 factors, do a range of 100-150T.

edit: Don't know for sure, but I think y=5 might be the best for a .dat of range n=2M.

[ValerieVonck]

Reserving 802T => 844T
P4 - 2400 - 512 mb

[hhh]

If you want to find factors needed soon on the SoB side, you could create a .dat from 14M to 15M. Below is beeing P-1ed currently.

Yours, H.

[ValerieVonck]

finished my range

05/25/07 23:05:04 WARNING: Only testing for factors p of the form p=1 (mod 4096).
05/25/07 23:05:05 sr2sieve started: 2 <= n <= 49999947, 802000000000000 <= p <= 844000000000000
05/25/07 23:05:29 sr2sieve stopped: at p=802005932351489 because SIGINT was received.
05/25/07 23:05:29 Found factors for 0 terms in 25.968 cpu sec. (expected about 0.29)
05/25/07 23:06:19 WARNING: Only testing for factors p of the form p=1 (mod 4096).
05/25/07 23:06:19 sr2sieve started: 2 <= n <= 49999947, 802005932351489 <= p <= 844000000000000
05/25/07 23:33:36 sr2sieve stopped: at p=804090763714561 because SIGINT was received.
05/25/07 23:33:36 Found factors for 0 terms in 1659.984 cpu sec. (expected about 103.67)
05/26/07 10:53:41 WARNING: Only testing for factors p of the form p=1 (mod 4096).
05/26/07 10:53:41 sr2sieve started: 2 <= n <= 49999947, 804090763714561 <= p <= 844000000000000
05/26/07 19:21:57 sr2sieve stopped: at p=842603790606337 because SIGINT was received.
05/26/07 19:21:57 Found factors for 0 terms in 31699.796 cpu sec. (expected about 1963.85)
05/26/07 20:00:01 WARNING: Only testing for factors p of the form p=1 (mod 4096).
05/26/07 20:00:02 sr2sieve started: 2 <= n <= 49999947, 842603790606337 <= p <= 844000000000000
05/26/07 20:18:39 sr2sieve stopped: at p=844000000000000 because range is complete.
05/26/07 20:18:39 Found factors for 0 terms in 32811.092 cpu sec. (expected about 2029.59)

[Citrix]

Attached is PSP dat from 4M to 6M to test.
I tried making a dat file for PSP+SOB with n=4M to 6M for PSP and n=14M to 16M for SOB. Geoff's Srsieve became very slow when I did this, so I am just posting the PSP 2M dat.

Reserve ranges as you wish.Thanks!