I have a question that is very basic. How can I tell if BOINC is even seeing my GPUs? So far, I'm not seeing any GPU work units. I am repurposing an old bitcoin mining rig. It has some lame built in graphics, and then two very nice AMD GPUs. I have enabled GPU use all the time, but how do I know it's working? And if I do see GPU work units, how do I know which GPUs are being used? I'd like to be able to specifically tell it use both AMDs, and don't use the integrated.

7.4.36 BOINC Manager running on Windows 7, btw.

And I just chcked the event log. It does seem to see both of my GPUs in there. The AMD ones, not the integrated (which is fine). But why am I still getting "GPU Missing" errors from Milkyway@Home and Collatz?

I am running AMD driver updates right now just in case that is the problem. But somehow, I doubt it. And for what it's worth, my litecoin miner had no problems using both GPUs for a year.

Updating to the newest AMD drivers had no effect.

Please post your Boinc startup from the Event Log.

Claggy

I agree with Claggy, we need a lot more information from you than what you've posted thus far. You've so far only told us the BOINC version, your operating system, that you have 2 AMD GPUs and which projects you expect work from.

Nothing about the actual make and model of the GPUs in question;
which version of the drivers;
where you got the drivers from, AMD or Windows;
how BOINC is installed, user install or service;
if these projects are up and have work for the applications;
whether or not you have set in the project preferences that these GPUs can be used;
and for which science applications;

I have now added the steps program of when requesting help on these forum to this sub-forum as well, and think that the BOINC start up messages are the minimum you still need to tell us about. Probably wise to tell us what AMD GPUs they are as well, in case BOINC recognizes them as a grouped object.

---

3/11/2015 5:01:49 PM | | cc_config.xml not found - using defaults
3/11/2015 5:01:51 PM | | Starting BOINC client version 7.4.36 for windows_x86_64
3/11/2015 5:01:51 PM | | log flags: file_xfer, sched_ops, task
3/11/2015 5:01:51 PM | | Libraries: libcurl/7.39.0 OpenSSL/1.0.1j zlib/1.2.8
3/11/2015 5:01:51 PM | | Data directory: C:\ProgramData\BOINC
3/11/2015 5:01:51 PM | | Running under account johnm
3/11/2015 5:01:51 PM | | CAL: ATI GPU 0: AMD Radeon HD 7850/7870 series (Pitcairn) (CAL version 1.4.1848, 2048MB, 2008MB available, 6240 GFLOPS peak)
3/11/2015 5:01:51 PM | | CAL: ATI GPU 1: AMD Radeon HD 7850/7870 series (Pitcairn) (CAL version 1.4.1848, 2048MB, 2008MB available, 6240 GFLOPS peak)
3/11/2015 5:01:51 PM | | OpenCL: AMD/ATI GPU 0: AMD Radeon HD 7850/7870 series (Pitcairn) (driver version 1642.5 (VM), device version OpenCL 1.2 AMD-APP (1642.5), 2048MB, 2008MB available, 6240 GFLOPS peak)
3/11/2015 5:01:51 PM | | OpenCL: AMD/ATI GPU 1: AMD Radeon HD 7850/7870 series (Pitcairn) (driver version 1642.5 (VM), device version OpenCL 1.2 AMD-APP (1642.5), 2048MB, 2008MB available, 6240 GFLOPS peak)
3/11/2015 5:01:51 PM | | OpenCL CPU: Intel(R) Celeron(R) CPU G540 @ 2.50GHz (OpenCL driver vendor: Intel(R) Corporation, driver version 1.1, device version OpenCL 1.1 (Build 37149.37214))
3/11/2015 5:01:51 PM | | OpenCL CPU: Intel(R) Celeron(R) CPU G540 @ 2.50GHz (OpenCL driver vendor: Advanced Micro Devices, Inc., driver version 1642.5 (sse2), device version OpenCL 1.2 AMD-APP (1642.5))
3/11/2015 5:01:51 PM | | Host name: Coin-Miner-1
3/11/2015 5:01:51 PM | | Processor: 2 GenuineIntel Intel(R) Celeron(R) CPU G540 @ 2.50GHz [Family 6 Model 42 Stepping 7]
3/11/2015 5:01:51 PM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 popcnt syscall nx lm vmx tm2 pbe
3/11/2015 5:01:51 PM | | OS: Microsoft Windows 7: Ultimate x64 Edition, Service Pack 1, (06.01.7601.00)
3/11/2015 5:01:51 PM | | Memory: 3.70 GB physical, 7.39 GB virtual
3/11/2015 5:01:51 PM | | Disk: 111.69 GB total, 80.81 GB free
3/11/2015 5:01:51 PM | | Local time is UTC -4 hours
3/11/2015 5:01:51 PM | Collatz Conjecture | URL http://boinc.thesonntags.com/collatz/; Computer ID 180341; resource share 100
3/11/2015 5:01:51 PM | Milkyway@Home | URL http://milkyway.cs.rpi.edu/milkyway/; Computer ID 606817; resource share 100
3/11/2015 5:01:51 PM | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7482970; resource share 100
3/11/2015 5:01:51 PM | SETI@home | General prefs: from SETI@home (last modified 21-Apr-2007 22:42:54)
3/11/2015 5:01:51 PM | SETI@home | Computer location: home
3/11/2015 5:01:51 PM | SETI@home | General prefs: no separate prefs for home; using your defaults
3/11/2015 5:01:51 PM | | Reading preferences override file
3/11/2015 5:01:51 PM | | Preferences:
3/11/2015 5:01:51 PM | | max memory usage when active: 1893.46MB
3/11/2015 5:01:51 PM | | max memory usage when idle: 3408.22MB
3/11/2015 5:01:51 PM | | max disk usage: 80.75GB
3/11/2015 5:01:51 PM | | (to change preferences, visit a project web site or select Preferences in the Manager)
3/11/2015 5:01:51 PM | | Not using a proxy
3/11/2015 5:02:17 PM | | Remote desktop in use; disabling GPU tasks
3/11/2015 5:02:17 PM | Collatz Conjecture | Sending scheduler request: To report completed tasks.
3/11/2015 5:02:17 PM | Collatz Conjecture | Reporting 1 completed tasks
3/11/2015 5:02:17 PM | Collatz Conjecture | Requesting new tasks for CPU
3/11/2015 5:02:20 PM | Collatz Conjecture | Scheduler request completed: got 0 new tasks
3/11/2015 5:02:20 PM | Collatz Conjecture | Server error: feeder not running

Two AMD R9 270s with 2GB of DDR5 each
The latest drivers which I believe are 14.2, downloaded from AMD's site
BOINC is installed as a regular GUI application.
The projects are up and do have work. I have GPU tasks downoaded and ready to go, but they all give the "GPU missing" message in the Tasks tab, under status.
In preferences, I have it set to use GPU all the time without limitations.

You're not using remote desktop by any chance?

Claggy

3/11/2015 5:02:17 PM | | Remote desktop in use; disabling GPU tasks

As Claggy said...
When you use remote desktop in Windows, Windows switches to using a built-in driver for remote desktop. This driver is just a basic screen driver that lacks GPGPU components, and thus then the GPU stops doing calculations.

Simplest fix if you must use a remote desktop option is to use a third party application that uses the driver you installed, such as VNC (not free) or the freeware thightVNC.

3/11/2015 5:02:17 PM | Collatz Conjecture | Sending scheduler request: To report completed tasks.
3/11/2015 5:02:17 PM | Collatz Conjecture | Reporting 1 completed tasks
3/11/2015 5:02:17 PM | Collatz Conjecture | Requesting new tasks for CPU
3/11/2015 5:02:20 PM | Collatz Conjecture | Scheduler request completed: got 0 new tasks
3/11/2015 5:02:20 PM | Collatz Conjecture | Server error: feeder not running

This shows that Collatz is down.

---

I do use windows remote desktop. This is a headless former-coin miner. I'll see if I can get VNC going. Would I have to do anything to 'disable' the windows remote desktop, and activate the normal GPU drivers? Once I get VNC going, that is?

Success! Connected via VNC, and I turned off remote desktop in some system setting for good measure, and now I have GPU work units running!

Would I have to do anything to 'disable' the windows remote desktop, and activate the normal GPU drivers? Once I get VNC going, that is?

I know you have it working, but just for anyone else wondering about this... no, as soon as Windows stops using remote desktop, it'll switch back to the original driver that you installed. BOINC will then follow suit within 10 seconds, and restart calculations.

---

So my tight vnc server isn't working so well. I lose contact after a few minutes. So just to be clear, if I connect with windows remote desktop... as soon as I disconnect from that screen sharing, the computer will automatically switch back over to the "accelerated" graphics drivers, and BOINC will automatically start doing GPU work again? So I'll never be able to SEE it doing GPU work, but it will be any time I'm not looking? Windows remote desktop was very reliable, this VNC stuff is torture on this particular machine.

How 'remote' is your server, in networking terms? If you access it across a WAN, then RDP/VNC is probably your only option. But if you have LAN access, you could use RDP to enable Controlling BOINC remotely, and then BOINC's own inbuilt networking capability to verify that it's working.

What a pain in the ass this is turning into.
So the VNC server worked when it worked, but it kept freezing up the machine solid.
So then I switched back to MS remote desktop, uninstalled the VNC server. No more system freezes and I can connect better. But now I'm getting on GPU work. When I disconnect from remote desktop, the machine never starts doing any GPU work. I can tell this because I have the box on a watt meter and I can clearly see when it's doing work just on the CPU, on CPU and one GPU and on CPU and both GPUs. It stays not doing any GPU work with MS remote desktop just enabled. I really want to find aliens but I guess it's not going to happen.

Turns out it wasn't VNC that was crashing the machine, it was BOINC. It can only run for a few minutes, maybe half hour at most with BOINC running before it freezes solid. Without BOINC running, its perfectly stable. :-/

GPU over heating?