[QUOTE=Dubslow;470994]

I'll start with C179_128_87[/QUOTE]

[code]linear algebra completed 13278827 of 13279057 dimensions (100.0%, ETA 0h 0m)
lanczos halted after 209998 iterations (dim = 13278832)
recovered 40 nontrivial dependencies
BLanczosTime: 560188

commencing square root phase
handling dependencies 1 to 64
reading relations for dependency 1
read 6637108 cycles
cycles contain 20318558 unique relations
read 20318558 relations
multiplying 20318558 relations
multiply complete, coefficients have about 591.14 million bits
initial square root is modulo 201151
sqrtTime: 2041
p69 factor: 117785364136821512925601021797774153264921697097231933631065994380963
p111 factor: 124356721327251738236724520252489516481197189145346038072712997655982168684651560177493966260727065140467266891
elapsed time 157:59:13[/code]

132 failed, I believe 120 also failed IIRC hence the chosen td=100. Elapsed time is slightly artifically inflated by a few pauses on the order of an ~hour each.

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


I'll replace this job with C206 from phi_13(phi_7(5231)/79).

Edit: Could someone please explain what the phi means? I figured out via FDB that this number is the cofactor of ((5231^7-1)/371330)^13-1, but I'm not quite sure how that corresponds to the title I reserved. It's not the Euler phi, right? Looks to me more like [$]\sigma((\frac{\sigma(5231^6)}{371330})^{12})[/$].

C192_1920647391913_19 and C219_129_92 are done.

[QUOTE=Dubslow;471027]I'll take C167_6228362269_23.[/QUOTE]

[code]
matrix is 11093476 x 11093701 (4975.6 MB) with weight 1287974535 (116.10/col)
sparse part has weight 1193394360 (107.57/col)
using block size 8192 and superblock size 589824 for processor cache size 6144 kB
commencing Lanczos iteration (4 threads)
memory use: 4235.4 MB
linear algebra at 0.0%, ETA 128h 7m093701 dimensions (0.0%, ETA 128h 7m)
checkpointing every 90000 dimensions93701 dimensions (0.0%, ETA 129h38m)
linear algebra completed 11093458 of 11093701 dimensions (100.0%, ETA 0h 0m)
lanczos halted after 175434 iterations (dim = 11093473)
recovered 35 nontrivial dependencies
BLanczosTime: 475366

commencing square root phase
handling dependencies 1 to 64
reading relations for dependency 1
read 5547356 cycles
cycles contain 18343222 unique relations
read 18343222 relations
multiplying 18343222 relations
multiply complete, coefficients have about 507.86 million bits
initial square root is modulo 1295958571
sqrtTime: 1930
p58 factor: 8470359578370412058969804375004353570376895978369804952261
p109 factor: 2753987064573008793490389067269262131434294898646249917899256051708431399479973137914218412394079297311398911
elapsed time 134:06:00[/code]

td=132 IIRC. Is this an ECM miss?

[url]https://pastebin.com/4aFx006g[/url]

To replace it, taking "C176 from phi_11(phi_13(467)/(157*14610077))".

Edit: A random question for anyone who can answer. Is disk speed (SSD vs slow HDD) a bottleneck in filtering (or matrix solving or even the sqrt)? IOW, is there any noticeable penalty from using a slow HDD relative to an SSD?

Reserving C176 from phi_11(phi_13(467)/(157*14610077)) please.

[QUOTE=richs;471545]Reserving C176 from phi_11(phi_13(467)/(157*14610077)) please.[/QUOTE]

The post immediately prior to yours was me reserving it. Matrix already in progress.

Taking phi_5(phi_13(43609)) from 15e.

[QUOTE=Dubslow;471546]The post immediately prior to yours was me reserving it. Matrix already in progress.[/QUOTE]

Sorry, I didn't see it. No harm done since I had not started.

[QUOTE=Dubslow;471539]

td=132 IIRC. Is this an ECM miss?

Edit: A random question for anyone who can answer. Is disk speed (SSD vs slow HDD) a bottleneck in filtering (or matrix solving or even the sqrt)? IOW, is there any noticeable penalty from using a slow HDD relative to an SSD?[/QUOTE]

Naw, a GNFS 167 should be ECM'ed to around 1.5*t50. A p58 is a ways above that ECM depth.

An SSD cuts by a factor of ~10 the remdups time and the reading-relations time within filtering. Matrix solving and sqrt aren't meaningfully affected. On 14e tasks, many of the relation sets have ~25% duplicates, so running remdups (on SSD or spinnyrust) cuts filtering time quite a bit- useful for jobs where you might run multiple target_density choices, or fear the relation set is oversieved.

[QUOTE=richs;471549]Sorry, I didn't see it. No harm done since I had not started.[/QUOTE]
I was never particularly worried, hardly the end of the world, but I figured quick and simple post means fewest losses. :smile:

[QUOTE=VBCurtis;471554]Naw, a GNFS 167 should be ECM'ed to around 1.5*t50. A p58 is a ways above that ECM depth.

An SSD cuts by a factor of ~10 the remdups time and the reading-relations time within filtering. Matrix solving and sqrt aren't meaningfully affected. On 14e tasks, many of the relation sets have ~25% duplicates, so running remdups (on SSD or spinnyrust) cuts filtering time quite a bit- useful for jobs where you might run multiple target_density choices, or fear the relation set is oversieved.[/QUOTE]

Thanks. I thought the msieve relations reading time was significantly slower than e.g. with remdups, which would indicate that the disk read speed is not the issue?

I ask because the SSD in my server is only 40 GB, which isn't really a lot. I think if I ever want to do larger jobs (can't really imagine why though :smile:) I'd probably just have to take the performance hit of running off the HDD.

l5715a factors
 

Log at [url]https://pastebin.com/w2mhs2wp[/url]
[code]prp64 factor: 1326547679693216195415207979047945539038010136629272997152371571
prp116 factor: 30763928607881574259775815051085662011220279524652271470895115875379106200949720706713542697293478846342970519422591[/code]
Matrix 18.9M at density somewhere in the 90s; the log does not contain all the filtering runs, and I chose the matrix with the shortest ETA among the 3 or 4 filtering runs I tried. The dataset was oversieved, and I played with reducing max-rels and got a matrix of 17.0M with density ~134; however, this had a marginally longer ETA than the 18.9M matrix that used the full dataset.

[QUOTE=Dubslow;471539]taking "C176 from phi_11(phi_13(467)/(157*14610077))"[/QUOTE]

[code]matrix is 4775799 x 4776024 (1986.9 MB) with weight 507168699 (106.19/col)
sparse part has weight 473083513 (99.05/col)
using block size 8192 and superblock size 589824 for processor cache size 6144 kB
commencing Lanczos iteration (4 threads)
memory use: 1670.8 MB
linear algebra at 0.0%, ETA 14h48m776024 dimensions (0.0%, ETA 14h48m)
checkpointing every 310000 dimensions024 dimensions (0.0%, ETA 15h42m)
linear algebra completed 4775628 of 4776024 dimensions (100.0%, ETA 0h 0m)
lanczos halted after 75528 iterations (dim = 4775795)
recovered 34 nontrivial dependencies
BLanczosTime: 73922

commencing square root phase
handling dependencies 1 to 64
reading relations for dependency 1
read 2387732 cycles
cycles contain 7531148 unique relations
read 7531148 relations
multiplying 7531148 relations
multiply complete, coefficients have about 185.64 million bits
initial square root is modulo 4592947
sqrtTime: 525
p76 factor: 2275133046546010463392960754692250408893595743989442021480630407479677451561
p101 factor: 14849250759317137524992662778356622406936708163352353630108803001889391917906208971722727991
869773621
elapsed time 21:24:06[/code]

[url]https://pastebin.com/y3n3A7WD[/url]

Taking C212_14009_59.