I signed up as soon as it started, run my machine essentially 24/7, and have completed 8 work units. I get MIP1 work units and climateprediction work units. If some cores go idle, I allow a bunch of seti@home or rosetta work units to come through.

I believe ARP1 is still active, but very slow in providing work units.

I have 36 Linux only cores set to run HST1 and ARP1 only. I also do GPU only work outside of WCG. Most of the time, nearly all CPU cores are starved so several months ago, I set up crontabs on each machine to ask for work from WCG once per hour. That has helped but still a lot of cores remain unused at any given time. I do not like pestering WCG but it appears the only way at this time to make any progress for these projects since WU availability is so limited.

As an observation, not criticism, I find it ironic that projects are presented to a grid environment to perform tasks that could take life times using mainframe systems in an effort to significantly reduce overall project computation time and yet not be able to keep all resources busy 100% of the time.

STARBASEn

hst1 and arp1 are in limited supply at random times, so programming updates is not very helpful.

Rather than wasting the cores, why not run mcm1 and/or mip1? They are in regular supply and have short computation times. If you are in the Projects tab, click on Device Profiles and then on your machine in the next screen. You will find some radio buttons at the top of the Device Profiles page. Click on the Custom Profile button.

Under Cache Settings set both to 1.5 days.

Under Available Projects click on arp, hst, mcm and mip. Then go down to Project Limits and set the total of mcm and mip to the number of cores on the machine (plus 1). Set each of arp and hst to the number of cores on the machine.

Then Save.

This should result on an 8 core machine in requesting work which would include arp and hast about every 20 mins on average. Proportionately more frequently if more cores on the machine and less if less.

With extra machines all set like that, your average interval of request should come down to 5 minutes with a total of 32 cores. That is much more frequent than your crontabs.

The interval would increase when you get some of the longer computation arp and hst units.

Give it a try. I have 140 days of arp crunched with only 8 threads.

arp is very demanding on storage. WCG are working to solve that before stepping up the supply. That has no time estimation as yet.

Mike

To get an idea of home many work units are being processed each day, you can click on the project name on the My Contribution page.

At the top of the page is the Statistic section, and then the Yesterday section shows the total number of work unit results returned the previous day.

I'm getting a few on my main PC today...haven't checked the others!