2020 14e post processing reservations and results

[Xyzzy]

Originally Posted by VBCurtis

Welcome to the club, sir! I hope your new machine proves a rocket.

We are actually running this on our gaming PC using the Linux subsystem in Windows 10. We recently upgraded the CPU from 4 real cores to 8 real cores so we can run msieve in the background on a few cores and not notice it.

A few days ago we downloaded the relations for an already-taken job and it is running as we speak. We are using this job as a verification run to make sure everything works.

[pinhodecarlos]

Originally Posted by VBCurtis

Welcome to the club, sir! I hope your new machine proves a rocket.

xyzzy joining us reminds me- we don't have a benchmark matrix-solve to compare hardware, do we? Like, it might be nice to have a dropbox/etc with a sample job available, say something that produces a matrix 5M or so in size. Folks who wanted to compare hardware speeds could D/L this job and note elapsed time to run it.

Is this something worth pursuing?

There’s a thread from Tom for a msieve benchmark dataset he hosts.

[jyb]

p60 factor: 360681772014351384828649598371819822852126920885983637754829
p131 factor: 69386654609830376695244274417299417943051279633249588176410985514713922588875163770217003771152532420762472189927111249138446507713

https://pastebin.com/YcXLzWbf

[jyb]

Taking 7p3_358.

[RichD]

p105 factor: 228502788837988506687414482721359064217828618323880630011163289613812625239652445306605252128734249386667
p115 factor: 3089441369378679664834225033969735785659311345733495551660492830556613010448512149522698862085241377623332061418499

348M unique relations built an 18.8M matrix using TD=132.
Solve time about 272 hours. (-t 4)

Log at: https://pastebin.com/jyU4f3LQ

[Xyzzy]

Originally Posted by Xyzzy

A few days ago we downloaded the relations for an already-taken job and it is running as we speak. We are using this job as a verification run to make sure everything works.

This run ran flawlessly under the Linux subsystem for Windows. For some reason we thought it would not work.

Eight real cores at 4.5GHz and 3200MHz DDR4 memory made for a fairly speedy run time. We paused the process several times to play games. We also started it with just two cores before we realized that we could pause the process.

[RichD]

Taking 16381_59m1.

[jyb]

p59 factor: 69731434881474538408242438802491391624313089725160313729831
p116 factor: 15100647599480709741576256291494763911613408879774450606514577914475184036337821087073669999594477741863198035542561

https://pastebin.com/jFsM7iTa

[Dylan14]

Taking 48774625069_19m1 for post-processing.

[Xyzzy]

Originally Posted by VBCurtis

xyzzy joining us reminds me- we don't have a benchmark matrix-solve to compare hardware, do we? Like, it might be nice to have a dropbox/etc with a sample job available, say something that produces a matrix 5M or so in size. Folks who wanted to compare hardware speeds could D/L this job and note elapsed time to run it.

Is this something worth pursuing?

We will upload a 32 bit dataset to our Dropbox and share it. It will take a few weeks, though. (Our upload speed is as slow as our smilie!)

[Xyzzy]

Or we can put a smaller one up. Just link us to the files.