Total is closing in on one year very rapidly and might come in as low as 250-275 days if enough hosts join in/keep crunching.
The contribution of the optimized Windows science app is nothing short of amazing.
More relevant is probably the considerable increase in active users—although it may well be that some are being attracted back to the project by the recent improvements.
@Stef:
Thank you for this graphs, they are really useful!
The one I'm looking at almost every day is the "Percent per day". Something that puzzles me is that from the month-scale overview it looks like the value is mostly below the 400m mark and only crossed it once, while in the 2-week graph it's clearly continously above that level.
AFAIK we don't keep a record of the past days of this value, It's actually just a report of the present status of our WU generator. Would it be possible to give us access to the numbers you build the graph from?
@Odysseus:
Quote:
More relevant is probably the considerable increase in active users—although it may well be that some are being attracted back to the project by the recent improvements.
I'm rather inclined to think that many machines have been shut down during the hotter days of the year, some to prevent overheating, some because people rather spent their time outside or on vacation. The "Active Hosts" graph look dramatic, but watch the scale - the baseline is not at 0.
In any case we should have left around 200 days for S5R1, which is really a good reason to celebrate!
@Stef:
The one I'm looking at almost every day is the "Percent per day". Something that puzzles me is that from the month-scale overview it looks like the value is mostly below the 400m mark and only crossed it once, while in the 2-week graph it's clearly continously above that level.
Hmm, this looks like a display problem in the month-case - or even a bug in the software.
Yep. I tried several timescales. It looks like an averaging bug/feature.
The values are several times below 0.400%/d, it could be an effect from the low pass filter.
The trend of hosts active in last 7 days graph provided by stef on the same web page as the other graphs posted in this thread shows a very encouraging reversal in the last few weeks.
As this recovery seems to have started a week or two after the implementation of the faster distributed application, it is providing an additional increment to the increased output.
The graphical display of S5
)
The graphical display of S5 progress is superb with both long term and two week graphs.
Total is closing in on one year very rapidly and might come in as low as 250-275 days if enough hosts join in/keep crunching.
The contribution of the optimized Windows science app is nothing short of amazing.
RE: The graphical display
)
I know, now that the WU's are back to under 1 hour for machine, I've dropped Rosetta to allow more time for Einstein :)
Thanks to all who worked on the optimized apps!!!
Human Stupidity Is Infinite...
This shows a nice curve about
)
This shows a nice curve about 2-3 weeks ago.
Yep :-) if that doesn't
)
Yep :-) if that doesn't rock... can't wait to have my new PC join the pack.
RE: Total is closing in on
)
More relevant is probably the considerable increase in active users—although it may well be that some are being attracted back to the project by the recent improvements.
@Stef: Thank you for this
)
@Stef:
Thank you for this graphs, they are really useful!
The one I'm looking at almost every day is the "Percent per day". Something that puzzles me is that from the month-scale overview it looks like the value is mostly below the 400m mark and only crossed it once, while in the 2-week graph it's clearly continously above that level.
AFAIK we don't keep a record of the past days of this value, It's actually just a report of the present status of our WU generator. Would it be possible to give us access to the numbers you build the graph from?
@Odysseus:
I'm rather inclined to think that many machines have been shut down during the hotter days of the year, some to prevent overheating, some because people rather spent their time outside or on vacation. The "Active Hosts" graph look dramatic, but watch the scale - the baseline is not at 0.
In any case we should have left around 200 days for S5R1, which is really a good reason to celebrate!
BM
BM
RE: @Stef: The one I'm
)
Hmm. I see.
Here is the database dump.
http://homepage.hispeed.ch/einstein/einstein2.xml.bz2 (lots of numbers)
The first column is the absolute percentage.
I don't have a dump for "percent per day", because this value is calculated at runtime. So all i can do is a graph with a better resolution:
http://homepage.hispeed.ch/einstein/diffoobar.png
Stef
RE: Here is the database
)
Thank you very much!
Hmm, this looks like a display problem in the month-case - or even a bug in the software.
Anyway - thanks a lot again!
BM
BM
RE: Hmm, this looks like a
)
Yep. I tried several timescales. It looks like an averaging bug/feature.
The values are several times below 0.400%/d, it could be an effect from the low pass filter.
The trend of hosts active in
)
The trend of hosts active in last 7 days graph provided by stef on the same web page as the other graphs posted in this thread shows a very encouraging reversal in the last few weeks.
As this recovery seems to have started a week or two after the implementation of the faster distributed application, it is providing an additional increment to the increased output.