Dears, thanks for your patience.

I updated the acemd3 apps.

Also, I verified that there were very few results by apps from the old CUDA version, so I won't be re-deploying them. In other words, apps now require CUDA 10 (Linux) and CUDA 10.1 (Windows). In terms of drivers versions:

CUDA 10.1 (10.1.105) >= 418.39 for Windows
CUDA 10.0 (10.0.130) >= 410.48 for Linux

https://docs.nvidia.com/deploy/cuda-compatibility/index.html

do you have an ETA on when we can expect some CUDA 11.1 apps with Ampere support? I'm a little surprised you didnt recompile with CUDA 11.1 while you were already rebuilding the app for this most recent issue.

---

do you have an ETA on when we can expect some CUDA 11.1 apps with Ampere support? I'm a little surprised you didnt recompile with CUDA 11.1 while you were already rebuilding the app for this most recent issue.

Won't be long. Awaiting official releases from underlying libraries.

Please post here if you manage to get WUs.

First Windows task has downloaded and is running cleanly on GTX 1660 Super.

4lziA01_320_0-TONI_MDADex2sl-11-50-RND7994_1

Linux reports 'no tasks available'.

09/10/2020 14:44:40 | GPUGRID | No tasks are available for New version of ACEMD

on Linux, my systems are only seeing "no tasks available" during the schedule request.

---

Oh no here we go again

My Win10 machine is being told 'No tasks available"

First Windows task has downloaded and is running cleanly on GTX 1660 Super.

4lziA01_320_0-TONI_MDADex2sl-11-50-RND7994_1

Linux reports 'no tasks available'.

09/10/2020 14:44:40 | GPUGRID | No tasks are available for New version of ACEMD

This should provide an answer to those asking what's so difficult.

Got one task on the Windows 10 PC with GTX 1650, but the expected time seems to be too long.
Tullio

---

I think expected times estimates were broken by past failures. Size/time should be reasonable.

... the expected time seems to be too long.
Tullio

That's normal with a new application version, and it's the fault of the client - nothing for Toni to worry about.

It'll slowly get it right over time.

Also can't get WUs.
Host: Windows 10 LTSC x64 / GTX 1060 6GB / Driver 451.67

| GPUGRID | Sending scheduler request: To fetch work.
| GPUGRID | Requesting new tasks for CPU and NVIDIA GPU
| GPUGRID | Scheduler request completed: got 0 new tasks
| GPUGRID | No tasks sent
| GPUGRID | No tasks are available for Long runs (8-12 hours on fastest card)
| GPUGRID | No tasks are available for New version of ACEMD

Actually a couple of linux WUs went through. I hope it sorts out by itself.

It's probably just taking time to re-prepare all the tasks for the new app version, and we're all grabbing them faster than they become available.

My first Windows task has just returned and was validated.

Getting "no work units" here on a GTX 960. I checked my driver version... 390.138.

*sigh* I hope Ubuntu 18.04LTS can cough up a newer driver on its own. Otherwise it's Einstein for now.

Edit: There is indeed a 450 driver for 18.04LTS in the automatic installer. Now I'm waiting for some long-running work units on WCG to get to their next checkpoint so I can reboot.

My Linux (Mint) has driver: 450.66, and was running Cuda 10 before the interruption. I'll keep watching and waiting for new tasks.

No WUs sent - 4 Windows 10 machines, varying driver versions all above 452.xx

This isn't because we reported so many errored WUs when the project restarted is it? (No task available)

Firat task completed in about 6000 s.
Tullio

---