Hi all,

I just sent 500 WU on acemd beta, half of them with 450 credits and the other half 3000, only for the equilibration step. So this is a perfect occasion to check how the beta queue works ;). Once this is finish I will sent the complete simulations. This will allow to compute the energy of binding for about one hundred of different ligands to trypsin, and if this works, we'll extend to more than one thousand!!

Noelia.

Thanks Noelia.

These tasks are like a little vacation for my GTX 580, since the GPU load is only 15~30%. It's a warm day here in Southern California, and we are running 15 deg(C) cooler than the usual 77~78 deg with Primegrid's GPU apps.

Three completed successfully so far :)

Best wishes

---

My gpu's were bought to work, not take vacations! Such a low gpu utilization means extremely low points are awarded, which will not make it worth running for many here (including me)

I'm seeing 97 to 99% GPU Utilization!

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

My gpu's were bought to work, not take vacations! Such a low gpu utilization means extremely low points are awarded, which will not make it worth running for many here (including me)

Chill Stoneageman, it's betatesting. To each his/her own ;)


_

stonage you can always opt out of beta testing if you are mainly point mining.

I received 23 of these units so far, all completed and validated. So enough of the betas. let get on with the real thing!

stonage you can always opt out of beta testing if you are mainly point mining.

OK. That'll leave just you to run the Linux betas then :)

If I was 'point mining', I wouldn't be here at all.

I just don't like seeing my gpus slacking. LOL

regards,
from a chilled cave dweller.

Actually, this is already the real thing.

We are using the beta queue because there we have only 4.2 applications and we can use some new features of the new application.

This is some initial equilibration that we were before doing locally, but that we cannot do for thousands of systems. All the runs are relatively short and sporadic.

When we can remove the cuda3.1 application, then we will be able to do it in the main queue.

gdf

Good! Let's get rid of cuda3.1 and move forward.

Why are some of these units still using cuba4.0? They take about 50% longer to complete than cuda4.2.

Because over 50% of the people are still using old drivers.
We cannot get rid of cuda3.1.

gdf

im pleased to hear in this thread cuda31 support gets not killed now over night ^^ i downgraded drivers on my one 24/7 gpugrid machine to get cuda31 back and, vola´, zero errors and much shorter Runtimes (up to 15k sec! on a not so slow 285gtx) again and would love to have it still some time ;)

---

DSKAG Austria Research Team: http://www.research.dskag.at