Gold - 2.27.2020

Steve

Judging from your other badges, you have enough capacity to have crunched a lot more arp, but must not be employing the right strategy.

If you want more, please say how many cores/threads you have and your project priorities and I can suggest a better strategy.

Mike

Consistent with WCG doubling the supply.

Mike

Steve

Your problem is probably that you are limiting WCG to arp and hst. Hst is another limited availability project so your machines are not requesting more work often enough because that only happens when you return units.

With either of them, you need a mix with projects that are readily available and that means mcm and mip at present.

I am well on the way to Emerald for arp with only a 4 core/8 thread i7.

I have set my device profile (tab on left margin) to custom. Then available projects to arp, mip and mcm. You can add in hst if you wish.

mip and mcm are readily available and relatively short units so create updates frequently. That way you are more likely to be doing an update when arp (and hst) units are being released. There are only 4000 arp units per day, so appear infrequently. Hst is also infrequent.

Further down the device profile you will find the cache settings. I have set mine to 1.5 and 1.5. That means that my threads can work on one unit and hold one in reserve for when that has finished.

The next section is project limits. I have set arp to 8, mcm to 6 and mip to 4 - reasons later.

Do you use app_config? If not, I can let you have it. It restricts the number of units your machine can work on at any one time. I have set arp to 4, mcm to 5 and mip to 3.

Now for the reasons: arp are long units that use a lot of capacity but are intermittently available so I restrict it to work on 4 of my 8 threads with another 4 as cache. They take 24 hours on my machine so restricting to 1 set working and 1 set as cache keeps my machine logged as 'reliable' by WCG. There is a lot about 'reliable' in other threads.

In case arp runs dry, mcm and mip are set to 5 and 3 respectively in app_config with 1 extra each in project limits. They are readily available so I only hold one extra each to cover the few minutes between uploading one unit and downloading the next.

If I have at least 4 arp units at a time, they rapidly top the priority and start crunching and the other 4 threads are split between mcm and mip, normally 3/1 or 2/2, respectively.

Mip is another project with capacity problems so my settings limit it to 3 units at ant one time.

This way you don't get capacity problems and your machine requests more work more frequently than if you were just crunching arp (and hst).

For your other 2 machines, multiply the app_config limits by 4 or 8, then double the arp project limit and add just the one to the other project limits.

Try it. It works for me.

If you need app_config, please ask. It is easy to set up.

Mike

Steve

For each of the 3 character codes add the number 1 to make a 4 character code.

Mike