Today marks a milestone of sorts. It is the first day that the average time per work unit is less than one day. This could be the result of just a mix of faster machines which completed and reported tasks. However, the results from Nov. 27 show an average of .973 days per work unit, this is about 23 1/3 hours. The average credit per work unit for the day is about 743 BOINC points, a little less than the overall average of 807 at this point.
Cheers

Sgt Joe said:

My mind went to Office Space, and I laughed out loud.

Hey Mike, could you provide more info on the machines that ran the units? Are they the same specs? Did one host stop and re-start several times? I noticed that even though one machine claims less time spent crunching, it was returned two and a half hours later (or was that the result of the machine's cache?)... Or at least tell us what each machine's CPU benchmarks are in terms of how many cobblestones they can compute per unit time. I mean, the OS helps, but there are so many other variables that could interfere. Windows 7 machines could include an atom processor or an i7-6950X. Those machines could use the fasted ssds, or an ancient 64 gb hdd. That apparently can affect the benchmarks too. (In theory this shouldn't affect the benchmark, but the cobblestone tests aren't perfect, and network transfer and disk storage are officially cited as variables potentially affecting cobblestone benchmarks...) Without knowing this sort of information, it's impossible to tell whether it was a perfect storm of variables outside WCG's control, or something they need to fix

Hey, sorry man, I misread your post; I thought you had two machines that were wingmen for each other haha. I was like "Whoa, how unusual!" But given reality, yeah, there's not a whole lot of information to go on, other than the OS. WCG will probably have to look into it then, since I think they have access to a lot more host information on their side.

I also have a 3770k machine; it's not a bad little cruncher even considering its age! How much CPU does Einstein use for its GPU crunching? I remember back in the day I crunched for GPUGrid as well, but each unit ate about .7 threads. Made it hard to know how to allocate resources efficiently, but produced a lot more work overall. Considering getting back into the GPU game...

One machine claiming zero (0.0) points, getting all it can eat for workunitId = 1382983903 ...

Result Name          OS           AVN Status Sent Time         Due / Return Time CPUh  Claimed/Granted
ARP1_0030704_000_1-- Linux        727 Valid   12/7/19 01:05:55  12/7/19 16:14:05 15.14     0.0/738.9
ARP1_0030704_000_0-- Linux Fedora 727 Valid   12/7/19 00:45:43  12/7/19 16:17:08 14.62   738.9/738.9

[Generated by wcgformat]