Workunit name BETA_ARP1_0000360_000
Estimated app speed 2.53 GFLOPs/sec
Estimated task size 13,698 GFLOPs
CPU time 02:36:47
Elapsed time 02:39:31
Estimated time remaining 36:58:32 hh:mm:ss
Fraction done 6.708%
Virtual memory size 743.68 MB
Working set size 695.94 MB

No checkpoint set yet, so will not test restart atm.

Snagged two on two Linux machines. One is a slow 2.05 GHz APU with only 4 GB RAM and the other a 4.0 GHz FX-8350 with 16 GB RAM. Should be a good comparison for calendar run times.

This may be a little bit off topic, but there is no forum available for the project of this Beta test. From the information in the Beta test announcement by armstrdj, he has revealed some tantalizing details about the project. Each workunit will contain a simulation of a a 48 hour period of an un-named geographical area of currently unknown ("small") size. They will be doing a calendar years worth of simulations for each area with subsequent workunits being generated by information from the previous unit. Each unit needs to be run twice, so that would mean 366 work units per geographic area. If a person knew the size of the geographic areas and how many of them they were testing and the average time it takes to complete a workunit, a reasonable approximation could be made for the design scope and length of the project.
The approximate surface area of the earth is 510,000,000 km^2. If they are using an area of 100 km^2 this would require 5,100,000areas times 366 workunits per area which equals 1,866,000,000 work units. At their preliminary estimate of average time per work unit of 20+hours would give 37,332,000,000 cpu hours to process. This gives a total run time of 4,262,000 years of cpu time. Our current global run time is 1,826,997 years.
Thus, I am thinking they are either using larger geographic areas, or they are not trying to simulate the entire surface of the Earth and are using many fewer areas of the earth, maybe only land areas or populated areas or heavily trafficed ares of the oceans.
This is, of course, only speculation by a highly uninformed cruncher. Feel free to correct or speculate as you wish.
Cheers

The initial 100 workunits (200 results) had an initial deadline of 4 days. The remaining workunits have a deadline of 7 days. If a resend happens due to an error, then the resent result will have a shorter deadline. If my memory serves me correctly, this is set to 40% of the original deadline. So this would be about 3 days.

Note: resends should end in a _2, _3, etc....

Thanks,
-Uplinger

Do different tasks have different durations. I see that some are pretty long or 12 hours. I got 4 tasks that will last about 75 minutes. I have an aging i5 - 4 threads running Win 10.

EDIT 1 :But interesting. 8 minutes to go but progress bar shows only 6% complete. So, "remaining time" is not in sync with "progress" percentage.

EDIT 2: At about 4:45 remaining, the "time remaining" is starting to count up instead of down. So I guess these are going to be long tasks after all.