E@H GPU tasks error out after newMesa (GPU) drivers installed.

jay
jay
Joined: 25 Jan 07
Posts: 99
Credit: 84044023
RAC: 0
Topic 225778

Greetings.

 

Some times, when I do a linx/Ubuntu update and upgrade, I hit a snag.

Sometimes a compile flag was set differently and things don't work.

Other times a library is in an unexpected place.

This time, I suspect that my perspective is confused...   :-)

 

This line from the task results has me confused:

OpenCL compiling FAILED! : -11 . Error message: fatal error: cannot open file '/usr/local//usr/lib/clc/verde-amdgcn-mesa-mesa3d.bc': No such file or directory.

My first thought that a path-variable is mis-formed...

I do have:

/usr/lib/clc/verde-amdgcn-mesa-mesa3d.bc

so how did /usr/local/  and /usr/lib/clc/ get munged together?   ??????

I took a look at what I just installed today: (from /var/log/apt)

Start-Date: 2021-08-03  17:33:44
Requested-By: jay (1000)
Install: libllvm12:amd64 (1:12.0.0-3ubuntu1~21.04.1, automatic),
libclang-cpp12:amd64 (1:12.0.0-3ubuntu1~21.04.1, automatic)
Upgrade: mesa-opencl-icd:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2),
libglx-mesa0:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2),
libgbm1:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2),
wireless-regdb:amd64 (2020.11.20-0ubuntu1, 2021.07.14-0ubuntu1~21.04.1),
libxatracker2:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2),
mesa-va-drivers:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2),
libgl1-mesa-dri:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2),
libegl1-mesa-dev:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2),
mesa-vulkan-drivers:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2),
libglapi-mesa:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2),
libegl-mesa0:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2),
mesa-vdpau-drivers:amd64 (21.0.1-2, 21.0.3-0ubuntu0.2)
Remove: libclang-cpp11:amd64 (1:11.0.1-2ubuntu4)
End-Date: 2021-08-03  17:34:03


I guess I should state that my GPU was happily crunching beforehand.

Please help.

THANK YOU!!

Jay

 

=====

PS here is data from my boinc log:

Tue 03 Aug 2021 05:59:11 PM EDT |  | Starting BOINC client version 7.16.16 for x86_64-pc-linux-gnu
Tue 03 Aug 2021 05:59:11 PM EDT |  | log flags: file_xfer, sched_ops, task
Tue 03 Aug 2021 05:59:11 PM EDT |  | Libraries: libcurl/7.74.0 OpenSSL/1.1.1j zlib/1.2.11 brotli/1.0.9 libidn2/2.3.0 libpsl/0.21.0 (+libidn2/2.3.0) libssh/0.9.5/openssl/zlib nghttp2/1.43.0 librtmp/2.3
Tue 03 Aug 2021 05:59:11 PM EDT |  | Data directory: /var/lib/boinc-client
Tue 03 Aug 2021 05:59:11 PM EDT |  | OpenCL: AMD/ATI GPU 0: AMD VERDE (DRM 2.50.0, 5.11.0-25-generic, LLVM 12.0.0) (driver version 21.0.3, device version OpenCL 1.1 Mesa 21.0.3, 2048MB, 2048MB available, 512 GFLOPS peak)
Tue 03 Aug 2021 05:59:11 PM EDT |  | libc: Ubuntu GLIBC 2.33-0ubuntu5 version 2.33
Tue 03 Aug 2021 05:59:11 PM EDT |  | Host name: pc-14
Tue 03 Aug 2021 05:59:11 PM EDT |  | Processor: 8 AuthenticAMD AMD FX(tm)-8150 Eight-Core Processor [Family 21 Model 1 Stepping 2]
Tue 03 Aug 2021 05:59:11 PM EDT |  | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf pni pclmulqdq monitor ssse3 cx16 sse4_1 sse4_2 popcnt aes xsave avx lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs xop skinit wdt fma4 nodeid_msr topoext perfctr_core perfctr_nb cpb hw_pstate ssbd ibpb vmmcall arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold
Tue 03 Aug 2021 05:59:11 PM EDT |  | OS: Linux Ubuntu: Ubuntu 21.04 [5.11.0-25-generic|libc 2.33 (Ubuntu GLIBC 2.33-0ubuntu5)]
Tue 03 Aug 2021 05:59:11 PM EDT |  | Memory: 11.60 GB physical, 9.31 GB virtual
Tue 03 Aug 2021 05:59:11 PM EDT |  | Disk: 91.17 GB total, 83.29 GB free
Tue 03 Aug 2021 05:59:11 PM EDT |  | Local time is UTC -4 hours

 

Other stuff:

https://einsteinathome.org/host/12201025

and

https://einsteinathome.org/task/1151791002  is part of https://einsteinathome.org/workunit/566238901

  ( One of a few that errored out. I aborted the rest.)

 

Thanks again,

Jay

 

 

 

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3953
Credit: 46802462642
RAC: 64221291

Mesa drivers are not

Mesa drivers are not supported. You need to install the driver package from AMD or the ROCm drivers. 

_________________________________________________________________________

jay
jay
Joined: 25 Jan 07
Posts: 99
Credit: 84044023
RAC: 0

@ Ian and Steve:  For many

@ Ian and Steve:  For many years, I have used the drivers listed above.

Question 1: What OS, chipset, and drivers are you using?

 

The current AMD drivers have a limited scope of the hardware that they support..

I don't think either ROCM or off-the-AMD-shelf drivers support my card (see above.)

Ubuntu has had a good history of supplying the OpenCL-ICD drivers ( with a few hiccups.)

 

Question 2: Did you have an explanation of the mangled path?

 

Jay

 

ps:

I do not have the chips mentioned below:

 

 

ROCm officially supports AMD GPUs that use following chips:

  • GFX8 GPUs. “Fiji” chips, such as on the AMD Radeon R9 Fury X and Radeon Instinct MI8. “Polaris 10” chips, such as on the AMD Radeon RX 580 and Radeon Instinct MI6. ...
  • GFX9 GPUs. “Vega 10” chips, such as on the AMD Radeon RX Vega 64 and Radeon Instinct MI25.
Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5872
Credit: 117573309920
RAC: 35249888

jay wrote:... This line from

jay wrote:

... This line from the task results has me confused:

OpenCL compiling FAILED! : -11 . Error message: fatal error: cannot open file '/usr/local//usr/lib/clc/verde-amdgcn-mesa-mesa3d.bc': No such file or directory.

My first thought that a path-variable is mis-formed...

I do have:

/usr/lib/clc/verde-amdgcn-mesa-mesa3d.bc

so how did /usr/local/  and /usr/lib/clc/ get munged together?   ??????

Most probably a packaging error.  You should try to bring it to the attention of the packager.  It's nothing to do with the claim that, "Mesa drivers are not supported."

If you look at the stderr output on the website for a 'good' result from before the upgrade, it clearly says:-

Using OpenCL platform provided by: Mesa

It says exactly the same for a failed result so Mesa is not the problem.  If you look at the next couple of lines for both 'good' and 'bad' results after the above line, the only difference is the upgraded version of LLVM - 12.0.0 instead of 11.0.1.  Then for a 'good' result you get, "OpenCL device has FP64 support" whereas for the 'bad' result you get, "OpenCL compiling FAILED!", followed by the obviously mangled path.

That path would likely have been created by some script in the package itself and is not something you did or can fix.  You need to bring it to the attention of the package maintainer.  Chances are it will be a show stopper for lots of people so it should get fixed pretty quickly.

 

Cheers,
Gary.

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3953
Credit: 46802462642
RAC: 64221291

What GPU do you have exactly?

What GPU do you have exactly? "Verde" is just a vague reference to the generation. There are many different GPUs under this generation that fit your system specs.

_________________________________________________________________________

Comment viewing options

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