'nough said.

Now the short queue died!!

I see "Tasks ready to send 1,993".

Can someone please send 'em??

"Test application for CPU MD"

"Test application for CPU MD"

Exactly my point. There are no GPU work units ready to sent, only CPU ones.

My beasts are starving.
I had to feed them with short runs, but it's not to their taste.

New WU should be available now!

Good. Two shorts will finish soon........

New WU should be available now!

And, again... :-( (Short queue, too.)
____________

This seems to happen on weekends.

Typically resolved on Monday's.

Not a big deal for folks with slower GPU's -- say 750ti and slower. But folks with much faster GPU's are likely in work unit starvation <smile>

Reading this kind of "complaints" reminds me of what a wonderful community we built here.

I'm sending some new WU (WUID: PTCL_PLA2*, PTCL_LFE*). This time they are simulations of two proteins, lactoferrin and Phospholipase A2, and two well-known market drugs: ibuprofen and aspirin. Hopefully we'll be able to validate our method by correctly predicting binding affinities.

PTCL_LFE* is quite a big system and "harder" to crunch. I hope this satisfy your hunger ;)

Reading this kind of "complaints" reminds me of what a wonderful community we built here.

I'm sending some new WU (WUID: PTCL_PLA2*, PTCL_LFE*). This time they are simulations of two proteins, lactoferrin and Phospholipase A2, and two well-known market drugs: ibuprofen and aspirin. Hopefully we'll be able to validate our method by correctly predicting binding affinities.

PTCL_LFE* is quite a big system and "harder" to crunch. I hope this satisfy your hunger ;)

Let us know if you have any other problem or if with OC reduction you can fix stability. As always, new WU batches need a testing period.

Please add some short runs into the queue too.

I appears that the queue is empty again. This is rather unusual for midweek, especially on the heels of a weekend event of no work available.

Perhaps some explanation and forecast would be of value.

The approach I use is pretty much this.


When there are no new units available, and I have been able to acquire a second unit behind my active work unit, I setup a secondary GPU project and them, with a short 'suspend' of GPUGrid, pull down about 12 to 24 hours of my secondary project.

U then pull the suspend off of GPUGrid so that it can return to processing and place the secondary GPU project into no new work.

Ideally GPUGrid is able to generate work within a day or so and take over as the primary project.

The other condition that drives me away (temporarily) from GPUGrid as a primary project is the problem with VERY SLOW uploads that seems to affect a couple of my workstations (all on the same network) intermittently. In those situations, I find it can take 10 to 18 hours simply to upload a result. That too drives those workstations into secondary GPU processing land.

For this second problem, I suspect there are some infrastructure issues at might site, (noisty switch for example), but the problem is amplified by the HUGE upload files that GPUGrid generates.

The approach I use is pretty much this.


When there are no new units available, and I have been able to acquire a second unit behind my active work unit, I setup a secondary GPU project and them, with a short 'suspend' of GPUGrid, pull down about 12 to 24 hours of my secondary project.

U then pull the suspend off of GPUGrid so that it can return to processing and place the secondary GPU project into no new work.

Ideally GPUGrid is able to generate work within a day or so and take over as the primary project.

The other condition that drives me away (temporarily) from GPUGrid as a primary project is the problem with VERY SLOW uploads that seems to affect a couple of my workstations (all on the same network) intermittently. In those situations, I find it can take 10 to 18 hours simply to upload a result. That too drives those workstations into secondary GPU processing land.

For this second problem, I suspect there are some infrastructure issues at might site, (noisty switch for example), but the problem is amplified by the HUGE upload files that GPUGrid generates.

When there are no new units available, and I have been able to acquire a second unit behind my active work unit, I setup a secondary GPU project and them, with a short 'suspend' of GPUGrid, pull down about 12 to 24 hours of my secondary project.

U then pull the suspend off of GPUGrid so that it can return to processing and place the secondary GPU project into no new work.

Yup -- similar approaches -- mine's a bit more hands on. My secondary project it MooWrap mosty. Could be Collatz -- that project has periodic issues of its own.

I resolved on of my upload lag issues by replacing a switch -- apparently a single port of that switch was 'noisy' and creating problems.

I'm preparing a new batch of short WU, equilibrations should be ready to send by today. On the other hand, yesterday I sent new long WU but seems not to be enough to satiate the beast. Perfect! I'll prepare some new ones!

PD: remember that from each completed task a new task is regenerated and sent.

Looks like the queues are empty again -- and it's only Thursday night....

They are periodically getting refilled. Please be patient.

Long runs (8-12 hours on fastest card) 35 2,273 4.50 (1.06 - 17.87) 603

No problem -- I am patient <smile>

Hoping the queue gets refilled again.....

Short queue refilled!