The clinfo program can be used to query and dump out information about the platform's OpenCL implementation. The following is a sample of the output that it produces:
OpenCL Platforms:
Platform ID : 134570448
Profile : FULL_PROFILE
Version : OpenCL 1.0 CUDA 3.0.1
Name : NVIDIA CUDA
Vendor : NVIDIA Corporation
Extension Suffix : NV
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
OpenCL Devices:
Device ID : 134570528
Platform ID : 134570448
Vendor ID : 4318
Type : GPU
Profile : FULL_PROFILE
Version : OpenCL 1.0 CUDA
Driver Version : 195.36.15
Language Version : OpenCL 1.0
Name : GeForce GTX 275
Vendor : NVIDIA Corporation
Available : true
Compute Units : 30
Clock Frequency : 1404 MHz
Address Bits : 32
Byte Order : Little Endian
Max Work Size : 512x512x64
Max Items/Group : 512
Local Memory : 16 kB
Global Memory : 895 MB
Global Cache Type : None
Global Cache Size : 0 kB
Cache Line Size : 0
Max Alloc Size : 223 MB
Max Constant Size : 64 kB
Max Constant Args : 9
Separate Local : true
Alignment : 32 (min 16)
Max Param Bytes : 4352
Error Correction : false
Unified Memory : false
Float Support : InfinityNaN RoundNearest RoundZero RoundInfinity FusedMultiplyAdd
Double Support : Denom InfinityNaN RoundNearest RoundZero RoundInfinity FusedMultiplyAdd
Half Float Support: No
Image Support : 2D, 3D
Max 2D Image Size : 8192x8192
Max Samplers : 16
Max Read Images : 128
Max Write Images : 8
Max 3D Image : 2048x2048x2048
Writable 3D Images: false
Has Compiler : true
Native Kernels : false
Out of Order Exec : true
Preferred Vector Sizes:
char1, short1, int1, long1, float1, double1
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_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
Return to Home
