Comment 8 for bug 1499996

Works! So, wouldn't it be a good idea making nvidia-opencl-icd-* at least recommend to install nvidia-modprobe? Other people won't get confused like me trying to install the nvidia-libopencl1-*, too.

thopiekar@home:~$ clinfo
Number of platforms 1
  Platform Name NVIDIA CUDA
  Platform Vendor NVIDIA Corporation
  Platform Version OpenCL 1.2 CUDA 7.5.18
  Platform Profile FULL_PROFILE
  Platform Extensions cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_copy_opts
  Platform Extensions function suffix NV

  Platform Name NVIDIA CUDA
Number of devices 1
  Device Name GeForce GT 720
  Device Vendor NVIDIA Corporation
  Device Vendor ID 0x10de
  Device Version OpenCL 1.2 CUDA
  Driver Version 352.41
  Device OpenCL C Version OpenCL C 1.2
  Device Type GPU
  Device Profile FULL_PROFILE
  Device Topology (NV) PCI-E, 02:00.0
  Max compute units 1
  Max clock frequency 797MHz
  Compute Capability (NV) 3.5
  Device Partition (core)
    Max number of sub-devices 1
    Supported partition types None
  Max work item dimensions 3
  Max work item sizes 1024x1024x64
  Max work group size 1024
  Preferred work group size multiple 32
  Warp size (NV) 32
  Preferred / native vector sizes
    char 1 / 1
    short 1 / 1
    int 1 / 1
    long 1 / 1
    half 0 / 0 (n/a)
    float 1 / 1
    double 1 / 1 (cl_khr_fp64)
  Half-precision Floating-point support (n/a)
  Single-precision Floating-point support (core)
    Denormals Yes
    Infinity and NANs Yes
    Round to nearest Yes
    Round to zero Yes
    Round to infinity Yes
    IEEE754-2008 fused multiply-add Yes
    Support is emulated in software No
    Correctly-rounded divide and sqrt operations Yes
  Double-precision Floating-point support (cl_khr_fp64)
    Denormals Yes
    Infinity and NANs Yes
    Round to nearest Yes
    Round to zero Yes
    Round to infinity Yes
    IEEE754-2008 fused multiply-add Yes
    Support is emulated in software No
    Correctly-rounded divide and sqrt operations No
  Address bits 64, Little-Endian
  Global memory size 2140099968 (1.993GiB)
  Error Correction support No
  Max memory allocation 535024992 (510.2MiB)
  Unified memory for Host and Device No
  Integrated memory (NV) No
  Minimum alignment for any data type 128 bytes
  Alignment of base address 4096 bits (512 bytes)
  Global Memory cache type Read/Write
  Global Memory cache size 16384
  Global Memory cache line 128 bytes
  Image support Yes
    Max number of samplers per kernel 32
    Max size for 1D images from buffer 134217728 pixels
    Max 1D or 2D image array size 2048 images
    Max 2D image size 16384x16384 pixels
    Max 3D image size 4096x4096x4096 pixels
    Max number of read image args 256
    Max number of write image args 16
  Local memory type Local
  Local memory size 49152 (48KiB)
  Registers per block (NV) 65536
  Max constant buffer size 65536 (64KiB)
  Max number of constant args 9
  Max size of kernel argument 4352 (4.25KiB)
  Queue properties
    Out-of-order execution Yes
    Profiling Yes
  Prefer user sync for interop No
  Profiling timer resolution 1000ns
  Execution capabilities
    Run OpenCL kernels Yes
    Run native kernels No
    Kernel execution timeout (NV) Yes
  Concurrent copy and kernel execution (NV) Yes
    Number of async copy engines 1
  printf() buffer size 1048576 (1024KiB)
  Built-in kernels
  Device Available Yes
  Compiler Available Yes
  Linker Available Yes
  Device Extensions cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_copy_opts cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_fp64

NULL platform behavior
  clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) NVIDIA CUDA
  clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...) Success [NV]
  clCreateContext(NULL, ...) [default] Success [NV]
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) No platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR) No devices found in platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM) No devices found in platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL) No platform

ICD loader properties
  ICD loader Name OpenCL ICD Loader
  ICD loader Vendor OCL Icd free software
  ICD loader Version 2.2.7
  ICD loader Profile OpenCL 1.2