PrimeGrid
Please visit donation page to help the project cover running costs for this month

Toggle Menu

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise
1) Message boards : Cullen/Woodall prime search : Future CW-sieve. (Message 150941)
Posted 7 days ago by Profile Michael GoetzProject donor
For base 2, the NVIDIA sieve is faster, but that sieve does not support other bases.


To be clear, there's no need for additional *G*CW sieving at this point (or anywhere in the near future.) The interest is only for base 2 sieving.
2) Message boards : Number crunching : PrimeGrid's 16th Birthday Challenge (Message 150641)
Posted 46 days ago by Profile Michael GoetzProject donor
And is it just me, or does the birthday cake banner only have 15 candles?


You observe correctly but, as discussed on Discord, the candles count from 0, & the 0th one isnt shown :).


We decided to leave a candle missing to honor all that was lost during the pandemic.
3) Message boards : Number crunching : remaning time (Message 150622)
Posted 48 days ago by Profile Michael GoetzProject donor
I keep getting cpu tasks which show remaining time as over 9 days with a deadline of 6 days so I keep having to abort them as I will not get any points for them and every time they upload I get another one so it must be time to stop the project which is sad as I have over 160 million points and been doing it for over 12 years. can this be corrected so there is enough time remaining

Paul


Don't abort them.

1) Time estimates can be wrong for a variety of reasons.

2) As long as your computer is running the task, the server will extend the deadline as needed. See this thread for details.

3) If you return the correct result, you'll still get credit for it even if it's past the deadline, provide the task hasn't been purged from the database. (Which would be really, really late.)
4) Message boards : Number crunching : Flops decreasing (Message 150603)
Posted 49 days ago by Profile Michael GoetzProject donor
One of the first definitions of the Cobblestone (the unit of BOINC credits) was 1 GigaFLOPS == 200 Cobblestones. This is what the standard BOINC server code still uses for the "Current GigaFLOPS" on the server status page, it is just the cumulated RAC of all participants divided by 200. PrimeGrid obviously changed the code to show TeraFLOPS instead of GigaFLOPS in order to avoid large numbers; I don't know if they also changed the formula.


I'm the person who added those statistics to PrimeGrid's server status page, so I can talk about this authoritatively.

As you say, the standard in BOINC servers is to use the GFLOP/Cobblestone formula to reverse engineer the total GFLOPs. I was not at all satisfied with this, as it's got many, many flaws, but to paraphrase, "It's the worst imaginable system; except for all the rest."

Any other way of calculating GFLOPs has worse problems, especially since we have some non-BOINC ways of getting credit such as PRPNet and manual sieving. But even without that, it would be very, very hard to "count" GFLOPs task by task accurately.

RAC is simply the best estimate we have of total GFLOPs.

Furthermore, it's the way *every* BOINC project does it, so it gives us an apples-to-apples comparison to everyone else. As flawed as it is, at least it's measuring the same thing at each project.

Getting back to the original question, I don't think anyone mentioned one of the major factors in the decrease: It's summer in the northern hemisphere. It's too hot to crunch.
5) Message boards : Problems and Help : How to abort tasks from PrimeGrid website? (Message 150515)
Posted 63 days ago by Profile Michael GoetzProject donor
It can't be done. The tasks will just time out and be resent by the server.
6) Message boards : Problems and Help : SGS is not listed in the Applications link (Message 150514)
Posted 63 days ago by Profile Michael GoetzProject donor
Why is it called TPS


Its original name was Twin Prime Search.
7) Message boards : Problems and Help : SGS is not listed in the Applications link (Message 150503)
Posted 64 days ago by Profile Michael GoetzProject donor
It's not really a big deal as all the LLR tasks are almost all identical but somehow along the line I'm guessing it got deleted and never replaced

Well, I just took a look at Applications.
It's the first one listed as Sophie Germain (LLR) -- "llrTPS"
Was it there already or did someone include it after you posted ?


It's always been there. That page is generated automatically straight from the database. It's not set up manually.
8) Message boards : Project Staging Area : Shouldn't Linux and Win10 be about the same for FPS and PRS (Message 150497)
Posted 64 days ago by Profile Michael GoetzProject donor
AffinityWatcher. Almost.

That CPU is internally divided in half, and while Linux does a good job of keeping Threads confined to the same half of the CPU, Windows does not. AffinityWatcher is a tool that does that for LLR, but it won’t work on PFGW.
9) Message boards : Problems and Help : Weird host (Message 150467)
Posted 70 days ago by Profile Michael GoetzProject donor
This host: https://www.primegrid.com/show_host_detail.php?hostid=1057098 is getting a lot of tasks without returning any...

They are still going !
PrimeGrid doesn't have the ability to stop giving tasks to a host that never returns any results ?


In egregious cases, the server will temporarily stop sending tasks. The threshold is very high, however. Affected tasks will automatically get resent, and it’s not really a big problem, except, perhaps, to the person running that computer.

If, on the other hand, the server’s limits were more restrictive, while potentially “solving” a non-problem, it would create loads of real problems for other people whose computers managed to trigger the protections because they were testing something, or were running a video game, or there was a power spike, etc. Do not underestimate the awesome ability of a modern computer to mangle an astronomical number of tasks in a very short period of time!

The system works well the way it is. We do not want to replace a small problem with a much larger problem.
10) Message boards : The Riesel Problem : K = 192971 prime found by Propper (Message 150373)
Posted 82 days ago by Profile Michael GoetzProject donor
But the point is that it is possible to generate the additional files, tarball them, and ship them off to anyone that wants to quickly verify the work. Yeah, it is a couple extra steps, but it seems insanely worthwhile to do.


But it has to have *already* been done for the Gap Numbers. If it hasn't (and there's no reason to believe it has been, and lots of reason to believe it hasn't), then we're back to square one: if anyone is going to test those Gap Numbers, then the tests Propper has already done on them are wasted.


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2021 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 0.53, 0.81, 1.24
Generated 1 Aug 2021 | 4:23:13 UTC