As you know, I've been doing a lot of fiddling and tinkering lately. Usually I keep my fiddling and tinkering away from public eyes until whatever grander scheme I'm plotting is really ready, but I thought you guys should see this. It's inspirational to me and I hope you'll get some satisfaction from it as well.
http://www.seventeenorbust.com/danor...twork-rate.png
The graph shows the real computing power of the project since its inception in April '02. The divisions along the x axis are months, with the thicker lines being year boundaries. The graph ends on June 1. It's using the new running-time estimate (n^2 log n) and is far more accurate than n^3 (cEMs). The THz measurements are only estimates, but they're probably correct to within, say, 20%; I'm afraid the "current power" measurement on the public site overestimates things by about a factor of three.
But what's really important is the trend. As you can see, in the last three months our power has more than doubled! And that's a real doubling, not an artifact caused by inaccuracies in the way test times are estimated.
(Note: sieving and factoring are not included on this graph. If you throw those into the mix, I'm sure the total power is a lot higher. Also, this graph doesn't automatically update... it's just something I generated and uploaded.)
So congratulations, and keep up the good work. Good times are coming for this project... it's in the air.
Last edited by kugano; 06-11-2004 at 04:22 AM.
Proth tests completed 516 tests (0.2%) 270252 tests
Work done 49.41 T cEMs (0.4%) 13.16 P cEMs
As you can see every 15 minutes we done 0.2% of previous job.
But as I can remeber a few weeks ago crunchin rate was 0.1% every half a hour.
Well done. I make some PR and now 13 Russia 161.92 T 466.7 days.
Hope for TOP10!Up! Up!
wbr, Me. Dead J. Dona \
Half of that seems to be courtesy of TeamPrimeRib.Originally posted by kugano
our power has more than doubled!
They want to get to the top of the Total Production stat. Let's hope that Anandtech makes it a little harder by increasing crunching power as well(well, seems like they have done a lot recently)...
Your new graph really looks great, by the way!
Oh, and I have one question concerning this graph:
Can you explain the spikes in the graph?
The first column of that table is for the "Last Day". So unfortunately, we're not doing 0.2% every 15 minutes.Originally posted by Death
As you can see every 15 minutes we done 0.2% of previous job.
somebody doing mega-flush using SBQueue.Originally posted by Mystwalker
Oh, and I have one question concerning this graph:
Can you explain the spikes in the graph?
The first column of that table is for the "Last Day". So unfortunately, we're not doing 0.2% every 15 minutes.
so it was 0.1% per day and now 0.2% per day. thats what called double.
wbr, Me. Dead J. Dona \
i believe that is a count of total number of PRP test completed and for some reason we've been testing alot of really small numbers with the super secre account. I'm not sure if this is just taking care of non-matching residues or what but the super secret account has been doing several hundred tests aday recently. In the last day we've completed 219 supersecret/secret tests. the total number of PRPs done is at 514 so about 295 first time tests fiished in last day. Not a bad rate though considering we were barely reaching 200 two months ago. And remember every susessive test takes a little longer to complete. We should see these large number of PRPs being finished daily diminish quickly as the backlogged error-fix tests are finished off.
Hopefully, those teams also will discover factoring and sieving soon, as they're really weak at factoring statsOriginally posted by Mystwalker
Half of that seems to be courtesy of TeamPrimeRib.
They want to get to the top of the Total Production stat. Let's hope that Anandtech makes it a little harder by increasing crunching power as well.
I think one factor is the end of ECC2. A Short while after it's end I came onboard with SOB (about 53 days ago). I'm sure others did too.Originally posted by kugano
..But what's really important is the trend. As you can see, in the last three months our power has more than doubled! And that's a real doubling, not an artifact caused by inaccuracies in the way test times are estimated...[/B]
Right. ATM, Rechenkraft is still leading the sieving and factoring stats, although most work of this is only done by priwo and me...Originally posted by Troodon
Hopefully, those teams also will discover factoring and sieving soon, as they're really weak at factoring stats
You alone account for ~10% of the increase!Originally posted by DVNT1
I think one factor is the end of ECC2. A Short while after it's end I came onboard with SOB (about 53 days ago). I'm sure others did too.
Posting Permissions
Reply With Quote
