App version wants Nvidia? Now using AMD rx470

Stu D.
Stu D.
Joined: 25 Aug 15
Posts: 25
Credit: 684979946
RAC: 58869
Topic 206217

Hi, I changed the GPU card from Nvidia to AMD in computer z600. All tasks are for the AMD/ATI.

Perferences are set to, no nvidia, yes ATI.

Event log says app version is missing Nvidia, Any fixes? Thanks, Stu

 

3/18/2017 9:31:34 AM | | cc_config.xml not found - using defaults
3/18/2017 9:31:34 AM | | Starting BOINC client version 7.6.33 for windows_x86_64
3/18/2017 9:31:34 AM | | log flags: file_xfer, sched_ops, task
3/18/2017 9:31:34 AM | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2g zlib/1.2.8
3/18/2017 9:31:34 AM | | Data directory: C:\ProgramData\BOINC
3/18/2017 9:31:34 AM | | Running under account User
3/18/2017 9:31:34 AM | | OpenCL: AMD/ATI GPU 0: Radeon (TM) RX 470 Graphics (driver version 2264.11, device version OpenCL 2.0 AMD-APP (2264.11), 4096MB, 4096MB available, 5120 GFLOPS peak)
3/18/2017 9:31:35 AM | | app version refers to missing GPU type NVIDIA
3/18/2017 9:31:35 AM | Einstein@Home | Application uses missing NVIDIA GPU
3/18/2017 9:31:35 AM | | app version refers to missing GPU type NVIDIA
3/18/2017 9:31:35 AM | Einstein@Home | Application uses missing NVIDIA GPU
3/18/2017 9:31:35 AM | | app version refers to missing GPU type NVIDIA
3/18/2017 9:31:35 AM | Einstein@Home | Application uses missing NVIDIA GPU
3/18/2017 9:31:35 AM | | Host name: Z600
3/18/2017 9:31:35 AM | | Processor: 8 GenuineIntel Intel(R) Xeon(R) CPU E5620 @ 2.40GHz [Family 6 Model 44 Stepping 2]
3/18/2017 9:31:35 AM | | 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 aes syscall nx lm vmx smx tm2 dca pbe
3/18/2017 9:31:35 AM | | OS: Microsoft Windows 7: Professional x64 Edition, Service Pack 1, (06.01.7601.00)
3/18/2017 9:31:35 AM | | Memory: 5.98 GB physical, 11.97 GB virtual
3/18/2017 9:31:35 AM | | Disk: 465.66 GB total, 404.36 GB free
3/18/2017 9:31:35 AM | | Local time is UTC -4 hours
3/18/2017 9:31:35 AM | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 12346108; resource share 100
3/18/2017 9:31:35 AM | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7979788; resource share 100
3/18/2017 9:31:35 AM | Einstein@Home | General prefs: from Einstein@Home (last modified ---)
3/18/2017 9:31:35 AM | Einstein@Home | Computer location: work
3/18/2017 9:31:35 AM | | General prefs: using separate prefs for work
3/18/2017 9:31:35 AM | | Reading preferences override file
3/18/2017 9:31:35 AM | | Preferences:
3/18/2017 9:31:35 AM | | max memory usage when active: 5820.86MB
3/18/2017 9:31:35 AM | | max memory usage when idle: 6127.22MB
3/18/2017 9:31:35 AM | | max disk usage: 100.00GB
3/18/2017 9:31:35 AM | | (to change preferences, visit a project web site or select Preferences in the Manager)

Holmis
Joined: 4 Jan 05
Posts: 1118
Credit: 1055935564
RAC: 0

Abort all Nvidia tasks and

Abort all Nvidia tasks and download new ones for your AMD card or reinstall the Nvidia card and finish them.
Technically it's the same tasks but you use a different application and plan class to process them, the easiest fix is to abort and download new tasks.

Next time you change GPU brand set "No new tasks" and run down your cache.

Stu D.
Stu D.
Joined: 25 Aug 15
Posts: 25
Credit: 684979946
RAC: 58869

Hi, Thank you for the reply

Hi,

Thank you for the reply Holmis,

I thought of that a head of time. I did do the No new tasks the day before I switched. The remaining few GPU Nvidia Tasks I aborted. (couldn't wait to get to the new amd card in ;)

However, I did not abort any cpu tasks. I figured cpu wasn't dependant on which gpu is installed, but I believe those cpu tasks errored out and are gone

All tasks are newly download, after the amd card, but the log says it wants Nvidia??

Thank you again,

Stu

Gavin
Gavin
Joined: 21 Sep 10
Posts: 191
Credit: 40643367835
RAC: 1587258

Do a project reset to clear

Do a project reset to clear out any no longer needed Einstein related guff... The message is likely erroneous, Boinc probably thinks you still have a NVidia card hiding somewhere, especially if you didn't fully uninstall the NV drivers before swapping cards.
Whatever AMD work you have in your cache now will be re-sent, theres nothing to lose doing this :-)

Gav.

 

Stu D.
Stu D.
Joined: 25 Aug 15
Posts: 25
Credit: 684979946
RAC: 58869

Hi Gav, Thank you for your

Hi Gav,

Thank you for your reply, the reset did it. No more referencing a missing Nvidia gpu, the event log looks good!

Thank you again,

Stu

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5842
Credit: 109379456192
RAC: 35975186

Even though this issue is

Even though this issue is resolved, I thought I would add some comments so that anybody else who happens to browse this thread will be a little less concerned by the "missing GPU" messages.

Ideally, if a GPU change is planned, the work cache settings should be reduced right down ahead of time so that if something goes wrong, very few tasks will be lost.  Then, if NNT (no new tasks) is set at the appropriate time before the changeover, perhaps there will be no affected tasks left.  Before shutting down to change the hardware, abort AND report any remaining tasks that rely on the particular item being removed - in this case just NVIDIA GPU tasks.  There is no reason to be concerned about CPU tasks as they wont be adversely affected by a GPU change.  To prevent any issues while installing new drivers, BOINC should be set NOT to restart when the machine is rebooted.

After the change with the new hardware in place and new drivers correctly working, it's time to fire up BOINC.  The event log posted earlier is exactly what you should see if things are working correctly.  The particular entry you want to see is the one showing that the new GPU is correctly recognised along with the OpenCL capability.  You look for both those bits in the message.

Of course there will be an informational message about the fact that there is no longer an NVIDIA GPU.  It's not an error message, it's just information that shows that BOINC has recognised the change.  Notice there are three messages.  That simply means there were three separate entries for different app versions that relied on the NVIDIA GPU.  That's not surprising because there have been a number of versions (including beta versions) in recent times.  If you browse BOINC's state file (client_state.xml) you will probably find literally dozens of <app_version> .... </app_version> blocks in there which refer to all the previous versions of all the different apps that have been used on the machine.  This is no big deal.

The information about the missing GPU is given only at startup.  Once the client talks to the server and informs about the new GPU, appropriate new tasks should immediately be sent for it.  The CPU tasks should simply restart where they left off at the time of the shutdown.  I know this because I have changed several machines recently to a different brand of GPU and not had any issues with either CPU or GPU tasks.

I suspect (but don't know for sure) that you will see the missing GPU information each time BOINC is restarted.  It may be that the exchanges between client and server may eventually get rid of the message.  I don't know because I've never given it a chance to go that far.  Before restarting BOINC after a GPU change, I simply edit the state file to remove any <app_version> blocks that refer to the former GPU, seeing as they are definitely not needed anymore.  I'm not encouraging anyone else to do this as there are always risks in tampering with things you don't fully understand.  I'd be interested in knowing if anybody has seen whether or not you keep getting the message each time BOINC is restarted even after the new hardware has well and truly settled in.  I'll check myself next time I change a GPU.

The final point I'd like to make is that whilst a project 'reset' can cure many ills, it is often a 'sledgehammer to crack a peanut' and it does come with some consequences.   If you have to do a reset, you will trash everything, including all the CPU tasks that should have been quite OK.  You will lose all currently downloaded files, like apps, large data files, associated static files, templates, etc.  Everything will need to be downloaded afresh.  For many, this may be relatively insignificant in the overall scheme of things but for people on slow links with restrictive quotas it may be an issue.

 

Cheers,
Gary.

mmonnin
mmonnin
Joined: 29 May 16
Posts: 291
Credit: 3229203990
RAC: 1146385

App_config.xml files get

App_config.xml files get deleted as well with a reset. I forgot about that the one time I did that on another project.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.