[QUOTE=RichD;497428]OK, let's keep it simple. 20M it is. No use adding in more variables to an already labor intensive job. The next thing we know is you will be asking for a 20% pay increase. :smile:[/QUOTE]

It’s no big deal changing the starting point of Q, especially if it significantly improves yield. Just a detail that sometimes gets lost. I’ll keep an eye on it from now on.

I recently used 10M and 15M as starting Q values for several small jobs.

In other news, I've just unreserved C240_4603759_37_minus1.

The long form of that statement is that I wanted to post-process that 31-bit LPs task on a new computer, but despite significant oversieving (nearly 290M raw relations, a bit less than 220M unique), the matrix is too large to fit in the 8 GB of RAM that computer is equipped with...
I filtered at TD=110, the highest one I ever used, yielding a 18M matrix, by far the largest one I ever attempted tackling. During the filtering phase, msieve didn't trigger the OOM killer, but at the beginning of the LA phase, msieve's RAM consumption jumps beyond ~8.5 GB (I temporarily added 1 GB of swap space on the SSD, and killed nearly all daemons) and it gets destroyed by the OOM killer when the computer's all out of virtual memory. Higher target densities should work, given how much oversieving there was, but I'm looking at reducing the RAM consumption by 2 or 3 GBs, which is probably more than a higher TD could achieve. Not to mention that the computer has a single RAM stick, so memory bandwidth would be lackluster anyway.

[QUOTE=debrouxl;497440]... but despite significant oversieving (nearly 290M raw relations, a bit less than 220M unique), the matrix is too large to fit in the 8 GB of RAM that computer is equipped with...[/QUOTE]

The turning point seems to be around 190M or so unique for 31 LPB, depending on the number of quality relations (don't ask me to define that). After that, overseiving kicks in and so does the size of the matrix. The filtering switch parameter FILTER_MAXRELS needs to be used.

I'll take C240_4603759_37_minus1 but it will be at least a week before I can start it.

Taking C206_30027973_31_minus1

Reserving 861001^41-1.

C221_73659281_29m1 factored
 

[CODE]p104 factor: 20185303748522730793037775562743983472304720720702577782645332787255974318317015294363318791410058897189
p117 factor: 949148194283355082424252367652475645632030557162437516701965982615396477940613349297295599511836763004453311420329081[/CODE]
203M unique relations built a 11.5M matrix using TD=124 (128 failed).
Solve time just under 90 hours. (-t 4)

Log at: [url]https://pastebin.com/MNj1sEkS[/url]

15307^59-1 factored
 

[code]
Tue Oct 9 11:01:54 2018 p84 factor: 111595227867218703045973161044064562306597068045773455167580724966392860087580908199
Tue Oct 9 11:01:54 2018 p160 factor: 4742501942795667268049847878960680363594879337517323036093054402034911248783755878442887140510735825066451345441733222441856437927557685189178232658962431015043
[/code]


[url]https://pastebin.com/H2prihej[/url]

Releasing C240_4603759_37_minus1.

Taking C216_49852639_29A.

[QUOTE=RichD;497738]Releasing C240_4603759_37_minus1.

Taking C216_49852639_29A.[/QUOTE]

Rich gave up C240_4603759_37_minus1 because I accidentally started working on it (whoops, I downloaded the wrong data set), so please reserve that to me now. And C206_30027973_31_minus1 is taking longer than expected for that reason.

Reserving 2047572230657338751575051_11m1.

Somehow I requested way too large a Q range! By the time I realized it, all the workunits were in received state.

14401^59-1 factored
 

[CODE]p95 factor: 23770271074573778355726380289661586799780118137589509474429690170976216858675079385815135539721
p134 factor: 25667916339307611393960842831160818084793544148845392198436986205574721700131160722510789582931394830999267089204621967772310209360737[/CODE]
193M unique relations built a 13.1M matrix using TD=116 (120 failed).
Solve time about 104 hours. (-t 4)

Log at: [url]https://pastebin.com/Ksdj4H0F[/url]