Error compiling suitesparse package on Pop!_OS 19.10

43 views
Skip to first unread message

Olivier Guillon

unread,
Dec 19, 2019, 3:59:28 PM12/19/19
to sage-devel
Hello,
while trying to compile the last develop version of sagemath, my Pop!_OS machine failed to build suitesparse package
it seems like the CUDA toolchain on Pop!_OS 19.10 does not accept gcc compiler with version superior to 7. I got the following error :

-------------------------------------------------------------------------------------------------------------------------------------------------------
In file included from /usr/lib/cuda/include/cuda_runtime.h:83,
                 from <command-line>:
/usr/lib/cuda/include/crt/host_config.h:129:2: error: #error -- unsupported GNU version! gcc versions later than 7 are not supported!
  129 | #error -- unsupported GNU version! gcc versions later than 7 are not supported!
-------------------------------------------------------------------------------------------------------------------------------------------------------

Any clue for a workaround ?

Thanks.

Olivier Guillon

suitesparse-5.6.0.log

François Bissey

unread,
Dec 19, 2019, 4:57:02 PM12/19/19
to sage-devel
To be honest, I thought turning on using the GPU required manual intervention. Obviously it was turned on automatically for you. 

I'll get a ticket to fix this.

In the meantime do
CUDA=no make suitesparse
and it should build

François

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/4e7502b3-60c9-4d44-9589-0731fdd5a66f%40googlegroups.com.

Olivier Guillon

unread,
Dec 19, 2019, 5:10:33 PM12/19/19
to sage-devel
Thanks a lot for the quick answer !!! Will try it and report result. Actually I was in the process of building the package with the gcc version that came with CUDA (gcc6), but I must admit my total lack of positive outcome so far...

Olivier

Le jeudi 19 décembre 2019 22:57:02 UTC+1, François Bissey a écrit :
To be honest, I thought turning on using the GPU required manual intervention. Obviously it was turned on automatically for you. 

I'll get a ticket to fix this.

In the meantime do
CUDA=no make suitesparse
and it should build

François

On Fri, Dec 20, 2019 at 9:59 AM Olivier Guillon <olivier...@gmail.com> wrote:
Hello,
while trying to compile the last develop version of sagemath, my Pop!_OS machine failed to build suitesparse package
it seems like the CUDA toolchain on Pop!_OS 19.10 does not accept gcc compiler with version superior to 7. I got the following error :

-------------------------------------------------------------------------------------------------------------------------------------------------------
In file included from /usr/lib/cuda/include/cuda_runtime.h:83,
                 from <command-line>:
/usr/lib/cuda/include/crt/host_config.h:129:2: error: #error -- unsupported GNU version! gcc versions later than 7 are not supported!
  129 | #error -- unsupported GNU version! gcc versions later than 7 are not supported!
-------------------------------------------------------------------------------------------------------------------------------------------------------

Any clue for a workaround ?

Thanks.

Olivier Guillon

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-...@googlegroups.com.

Olivier Guillon

unread,
Dec 19, 2019, 6:24:39 PM12/19/19
to sage-devel
So "CUDA=no make suitesparse" didn't do the trick. I had to purge the whole CUDA toolchain to finally get a build, then reinstall CUDA...and now sage is up and running !
I should have mentionned that the physical machine is a dell xps 15 (9560) with hybrid graphics and Pop!OS 19.10 which untill now really did the job flawlessly,
but I'm seriously thinking jumping back to Archlinux where I didn't get any such collision between sage and cuda.
Once again thanks for answering so quickly, I might still be struggling with my idea  of building it with gcc6 (started building the whole source with it, which just brought more problems...).

Olivier
(olivier dot guillon at gmail dot com)
Reply all
Reply to author
Forward
0 new messages