Re: [petsc-dev] sm_70

2020-09-27 Thread Barry Smith
https://gitlab.com/petsc/petsc/-/merge_requests/3252 > On Sep 27, 2020, at 11:55 AM, Zhang, Hong wrote: > > > >> On Sep 25, 2020, at 8:09 PM, Barry Smith > > wrote: >> >> >> Configure by default should find

Re: [petsc-dev] sm_70

2020-09-27 Thread Barry Smith
> On Sep 27, 2020, at 11:55 AM, Zhang, Hong wrote: > > > >> On Sep 25, 2020, at 8:09 PM, Barry Smith > > wrote: >> >> >> Configure by default should find out the available GPU and build for that >> sm_* it should not require the user to set this (how the heck is

Re: [petsc-dev] sm_70

2020-09-27 Thread Zhang, Hong via petsc-dev
On Sep 25, 2020, at 8:09 PM, Barry Smith mailto:bsm...@petsc.dev>> wrote: Configure by default should find out the available GPU and build for that sm_* it should not require the user to set this (how the heck is the user going to know what to set?) If I remember correctly there is a uti

Re: [petsc-dev] sm_70

2020-09-26 Thread Mark Adams
That was a login node. Here is a compute node on SUMMIT: 12:45 130 adams/feature-mat-cuda *= ~/petsc/src/mat/tutorials$ nvidia-smi Sat Sep 26 12:45:28 2020 +-+ | NVIDIA-SMI 418.116.00 Driver Version: 418.116.00 CUDA Ve

Re: [petsc-dev] sm_70

2020-09-26 Thread Mark Adams
0:30 adams/feature-mat-cuda *= ~/petsc$ nvidia-smi Sat Sep 26 12:42:37 2020 +-+ | NVIDIA-SMI 418.116.00 Driver Version: 418.116.00 CUDA Version: 10.1 | |---+--+-

Re: [petsc-dev] sm_70

2020-09-25 Thread Stefano Zampini
I have added the configure option --with-cuda-gencodearch=XX to pass down such information to the external packages build. We are not currently using it in the library itself. Il Sab 26 Set 2020, 06:08 Barry Smith ha scritto: > > > On Sep 25, 2020, at 8:36 PM, Jacob Faibussowitsch > wrote: > >

Re: [petsc-dev] sm_70

2020-09-25 Thread Barry Smith
> On Sep 25, 2020, at 8:36 PM, Jacob Faibussowitsch wrote: > >> Configure by default should find out the available GPU and build for that >> sm_* it should not require the user to set this (how the heck is the user >> going to know what to set?) If I remember correctly there is a utility

Re: [petsc-dev] sm_70

2020-09-25 Thread Jacob Faibussowitsch
> Configure by default should find out the available GPU and build for that > sm_* it should not require the user to set this (how the heck is the user > going to know what to set?) If I remember correctly there is a utility > available that gives this information. For CUDA I believe the too

Re: [petsc-dev] sm_70

2020-09-25 Thread Barry Smith
Configure by default should find out the available GPU and build for that sm_* it should not require the user to set this (how the heck is the user going to know what to set?) If I remember correctly there is a utility available that gives this information. For generic builds like in pa

Re: [petsc-dev] sm_70

2020-09-25 Thread Mark Adams
'--CUDAFLAGS=-arch=sm_70', seems to fix this. On Fri, Sep 25, 2020 at 6:31 PM Mark Adams wrote: > I see kokkos and hyper have a sm_70 flag, but I don't see one for PETSc. > > It looks like you have to specify this to get modern atomics to work in > Cuda. I get: > > /ccs/home/adams/petsc/incl

[petsc-dev] sm_70

2020-09-25 Thread Mark Adams
I see kokkos and hyper have a sm_70 flag, but I don't see one for PETSc. It looks like you have to specify this to get modern atomics to work in Cuda. I get: /ccs/home/adams/petsc/include/petscaijdevice.h(99): error: no instance of overloaded function "atomicAdd" matches the argument list