[gmx-users] Could not build OpenCL program, error was CL_INVALID_BUILD_OPTIONS

[Mark Abraham]

Hi,

What changed between when it used to work and now it does not? GROMACS
version upgrade? System update? GPU driver upgrade? Phase of the moon? :-)
There's supposed to be output between LOG_START and LOG_END if the
compilation ran, but it seems to not have reached that point. Yet the
preprocessor options seem well formed.

Does a single-GPU run work, e.g. mdrun -ntmpi 1?

Mark

On Mon, Feb 20, 2017 at 11:37 AM Uliano Guerrini <uliano.guerrini at unimi.it>
wrote:

> Hi,
>
> I have gromacs 2016.2 (2016.1 was the same) installed on a MacPro with a
> dual AMD D500 graphic card. Mdrun used to work fine with OPENCL giving some
> nice (~3x) speed increase until the last simulation gave me this:
>
> […]
> Command line:
>   mdrun -v -s md.tpr
>
>
> Back Off! I just backed up md.log to ./#md.log.9#
>
> Running on 1 node with total 12 cores, 12 logical cores, 2 compatible GPUs
> Hardware detected:
>   CPU info:
>     Vendor: Intel
>     Brand:  Intel(R) Xeon(R) CPU E5-1650 v2 @ 3.50GHz
>     SIMD instructions most likely to fit this hardware: AVX_256
>     SIMD instructions selected at GROMACS compile time: AVX_256
>
>   Hardware topology: Only logical processor count
>   GPU info:
>     Number of GPUs detected: 2
>     #0: name: AMD Radeon HD - FirePro D500 Compute Engine, vendor: AMD,
> device version: OpenCL 1.2 , stat: compatible
>     #1: name: AMD Radeon HD - FirePro D500 Compute Engine, vendor: AMD,
> device version: OpenCL 1.2 , stat: compatible
>
> Reading file md.tpr, VERSION 2016.2 (single precision)
> Changing nstlist from 10 to 40, rlist from 1.2 to 1.247
>
> Using 2 MPI threads
> Using 6 OpenMP threads per tMPI thread
>
> 2 compatible GPUs are present, with IDs 0,1
> 2 GPUs auto-selected for this run.
> Mapping of GPU IDs to the 2 PP ranks in this node: 0,1
>
> Compilation of source file /opt/gromacs-2016.2/share/gromacs/opencl/
> nbnxn_ocl_kernels.cl failed!
> -- Used build options: -DWARP_SIZE_TEST=64 -D_AMD_SOURCE_
> -DGMX_OCL_FASTGEN_ADD_TWINCUT -DEL_EWALD_ANA -DEELNAME=_ElecEw
> -DLJ_COMB_LB  -DVDWNAME=_VdwLJCombLB -DCENTRAL=22
> -DNBNXN_GPU_NCLUSTER_PER_SUPERCLUSTER=8 -DNBNXN_GPU_CLUSTER_SIZE=8
> -DNBNXN_GPU_JGROUP_SIZE=4 -DNBNXN_MIN_RSQ=3.82e-07f -DIATYPE_SHMEM
> -cl-fast-relaxed-math -I/opt/gromacs-2016.2/share/gromacs/opencl
> --------------LOG START---------------
> ---------------LOG END----------------
>
> -------------------------------------------------------
> Program:     mdrun, version 2016.2
> Source file: src/gromacs/gpu_utils/ocl_compiler.cpp (line 482)
> Function:    cl_program gmx::ocl::compileProgram(FILE *, const std::string
> &, const std::string &, cl_context, cl_device_id, ocl_vendor_id_t)
> MPI rank:    0 (out of 2)
>
> Internal error (bug):
> Failed to compile NBNXN kernels for GPU #AMD Radeon HD - FirePro D500
> Compute
> Engine
>   Could not build OpenCL program, error was CL_INVALID_BUILD_OPTIONS
>
> For more information and tips for troubleshooting, please check the GROMACS
> website at http://www.gromacs.org/Documentation/Errors
> -------------------------------------------------------
> uliano at nero:~/Desktop/S1PR1/S1PR1+ML056$
>
> Any hint on what went wrong and on what could I do to have this working?
>
> best,
>
> uliano
>
>
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.