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

[Uliano Guerrini]

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