4

On my laptop a have a discrete card that should support OpenCL 1.2 which I need to run the desired framework:

$ lspci | grep VGA
00:02.0 VGA compatible controller: Intel Corporation 2nd Generation Core Processor Family Integrated Graphics Controller (rev 09)
01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Seymour [Radeon HD 6400M/7400M Series] (rev ff)

I have installed opensource drivers as described in the answer to this question and also downloaded and installed AMD app sdk 3.0. Test code devices.c and hello.c compile and run, but everything reports OpenCL 1.1 MESA even after I replaced symbolic link for libOpenCL.so in /usr/lib/x86_64-linux-gnu/ with the one from SDK:

$ /opt/AMDAPPSDK-3.0/bin/x86_64/clinfo
Number of platforms:                             1
  Platform Profile:                              FULL_PROFILE
  Platform Version:                              OpenCL 1.1 MESA 11.2.0
  Platform Name:                                 Clover
  Platform Vendor:                               Mesa
  Platform Extensions:                           cl_khr_icd


  Platform Name:                                 Clover
Number of devices:                               1
  Device Type:                                   CL_DEVICE_TYPE_GPU
  Vendor ID:                                     1002h
  Max compute units:                             2
  Max work items dimensions:                     3
    Max work items[0]:                           256
    Max work items[1]:                           256
    Max work items[2]:                           256
  Max work group size:                           256
  Preferred vector width char:                   16
  Preferred vector width short:                  8
  Preferred vector width int:                    4
  Preferred vector width long:                   2
  Preferred vector width float:                  4
  Preferred vector width double:                 0
  Native vector width char:                      16
  Native vector width short:                     8
  Native vector width int:                       4
  Native vector width long:                      2
  Native vector width float:                     4
  Native vector width double:                    0
  Max clock frequency:                           750Mhz
  Address bits:                                  32
  Max memory allocation:                         268435456
  Image support:                                 No
  Max size of kernel argument:                   1024
  Alignment (bits) of base address:              1024
  Minimum alignment (bytes) for any datatype:    128
  Single precision floating point capability
    Denorms:                                     No
    Quiet NaNs:                                  Yes
    Round to nearest even:                       Yes
    Round to zero:                               No
    Round to +ve and infinity:                   No
    IEEE754-2008 fused multiply-add:             No
  Cache type:                                    None
  Cache line size:                               0
  Cache size:                                    0
  Global memory size:                            1073741824
  Constant buffer size:                          268435456
  Max number of constant args:                   13
  Local memory type:                             Scratchpad
  Local memory size:                             32768
  Kernel Preferred work group size multiple:     64
  Error correction support:                      0
  Unified memory for Host and Device:            1
  Profiling timer resolution:                    0
  Device endianess:                              Little
  Available:                                     Yes
  Compiler available:                            Yes
  Execution capabilities:                                
    Execute OpenCL kernels:                      Yes
    Execute native function:                     No
  Queue on Host properties:                              
    Out-of-Order:                                No
    Profiling :                                  Yes
  Platform ID:                                   0x7f9d5e667ac0
  Name:                                          AMD CAICOS (DRM 2.43.0, LLVM 3.8.0)
  Vendor:                                        AMD
  Device OpenCL C version:                       OpenCL C 1.1 
  Driver version:                                11.2.0
  Profile:                                       FULL_PROFILE
  Version:                                       OpenCL 1.1 MESA 11.2.0
  Extensions:                                    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_byte_addressable_store

Question: Is there any way to know if my device indeed doesn't support OpenCL 1.2 or if the support is present, enable it then? Thank you very much!

P.S. I haven't installed any drivers for discrete card, but it is said that I don't need them. I'm running X on the embedded card:

$ glxinfo | grep -i vendor
server glx vendor string: SGI
client glx vendor string: Mesa Project and SGI
    Vendor: Intel Open Source Technology Center (0x8086)
OpenGL vendor string: Intel Open Source Technology Center

P.S.2. When I uninstalled opencl-headers and mesa-opencl-icd, even /opt/AMDAPPSDK-3.0/bin/x86_64/clinfo stopped working (cannot get platform IDs).

Edit3: Fixed previous one, by creating link to 

sudo ln -s '/opt/AMDAPPSDK-3.0/lib/x86_64/sdk/libamdocl64.so' /usr/lib/x86_64-linux-gnu/libamdocl64.so

but now it reports CPU as OpenCL device!

$ clinfo
Number of platforms:                             1
  Platform Profile:                              FULL_PROFILE
  Platform Version:                              OpenCL 2.0 AMD-APP (1800.8)
  Platform Name:                                 AMD Accelerated Parallel Processing
  Platform Vendor:                               Advanced Micro Devices, Inc.
  Platform Extensions:                           cl_khr_icd cl_amd_event_callback cl_amd_offline_devices 


  Platform Name:                                 AMD Accelerated Parallel Processing
Number of devices:                               1
  Device Type:                                   CL_DEVICE_TYPE_CPU
  Vendor ID:                                     1002h
  Board name:                                    
  Max compute units:                             4
  Max work items dimensions:                     3
    Max work items[0]:                           1024
    Max work items[1]:                           1024
    Max work items[2]:                           1024
  Max work group size:                           1024
  Preferred vector width char:                   16
  Preferred vector width short:                  8
  Preferred vector width int:                    4
  Preferred vector width long:                   2
  Preferred vector width float:                  8
  Preferred vector width double:                 4
  Native vector width char:                      16
  Native vector width short:                     8
  Native vector width int:                       4
  Native vector width long:                      2
  Native vector width float:                     8
  Native vector width double:                    4
  Max clock frequency:                           1044Mhz
  Address bits:                                  64
  Max memory allocation:                         2147483648
  Image support:                                 Yes
  Max number of images read arguments:           128
  Max number of images write arguments:          64
  Max image 2D width:                            8192
  Max image 2D height:                           8192
  Max image 3D width:                            2048
  Max image 3D height:                           2048
  Max image 3D depth:                            2048
  Max samplers within kernel:                    16
  Max size of kernel argument:                   4096
  Alignment (bits) of base address:              1024
  Minimum alignment (bytes) for any datatype:    128
  Single precision floating point capability
    Denorms:                                     Yes
    Quiet NaNs:                                  Yes
    Round to nearest even:                       Yes
    Round to zero:                               Yes
    Round to +ve and infinity:                   Yes
    IEEE754-2008 fused multiply-add:             Yes
  Cache type:                                    Read/Write
  Cache line size:                               64
  Cache size:                                    32768
  Global memory size:                            8301789184
  Constant buffer size:                          65536
  Max number of constant args:                   8
  Local memory type:                             Global
  Local memory size:                             32768
  Max pipe arguments:                            16
  Max pipe active reservations:                  16
  Max pipe packet size:                          2147483648
  Max global variable size:                      1879048192
  Max global variable preferred total size:      1879048192
  Max read/write image args:                     64
  Max on device events:                          0
  Queue on device max size:                      0
  Max on device queues:                          0
  Queue on device preferred size:                0
  SVM capabilities:                              
    Coarse grain buffer:                         No
    Fine grain buffer:                           No
    Fine grain system:                           No
    Atomics:                                     No
  Preferred platform atomic alignment:           0
  Preferred global atomic alignment:             0
  Preferred local atomic alignment:              0
  Kernel Preferred work group size multiple:     1
  Error correction support:                      0
  Unified memory for Host and Device:            1
  Profiling timer resolution:                    1
  Device endianess:                              Little
  Available:                                     Yes
  Compiler available:                            Yes
  Execution capabilities:                                
    Execute OpenCL kernels:                      Yes
    Execute native function:                     Yes
  Queue on Host properties:                              
    Out-of-Order:                                No
    Profiling :                                  Yes
  Queue on Device properties:                            
    Out-of-Order:                                No
    Profiling :                                  No
  Platform ID:                                   0x7fbb72e8c430
  Name:                                          Intel(R) Core(TM) i5-2430M CPU @ 2.40GHz
  Vendor:                                        GenuineIntel
  Device OpenCL C version:                       OpenCL C 1.2 
  Driver version:                                1800.8 (sse2,avx)
  Profile:                                       FULL_PROFILE
  Version:                                       OpenCL 1.2 AMD-APP (1800.8)
  Extensions:                                    cl_khr_fp64 cl_amd_fp64 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_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_device_fission cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_spir cl_khr_gl_event

Edit 4: tried this version of fglrx driver for OpenCL on Ubuntu 16.04, clinfo just crashes.

Edit 5: Made this approach working on custom Debian 8 jessie LiveCD, it reports OpenCL 1.2 GPU (fglrx-driver + AMD APP SDK installed in chroot on VM (despite that fglrx failed to detect hardware and update Xorg, but DKMS module was installed) and packed into Live USB according to online guides), but failed to accomplish the same on Ubuntu 14.04 (Debian is not supported by the program I need). The problem on both my 16.04 and older Ubuntu seems to be:

01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Seymour [Radeon HD 6400M/7400M Series] (rev ff) (prog-if ff)
        !!! Unknown header type 7f
        Kernel driver in use: radeon
        Kernel modules: radeon

No, I can make fglrx to see AMD as current renderer (14.04) by setting environment variable DRI_PRIME=1, but clinfo still crashes there.

Now I've read about that fglrx is deprecated and new AMD drivers does not support Radeon xxxxM, while opensource radeon one does not support OpenCL.

Community
  • 1
Slowpoke
  • 218

1 Answers1

2

You partially solved my problem on A10-7850k APU. I ran:

sudo ln -s '/opt/AMDAPPSDK-3.0/lib/x86_64/sdk/libamdocl64.so' /usr/lib/x86_64-linux-gnu/libamdocl64.so

But now the CPU is on openCL 1.2 but the GPU is on 1.1. It looks like the GPU is using amdgpu open source drivers which were compiled with libclc, using opencl 1.1. I can't figure out how to force the graphics driver to use the AMD lib... Here's my clinfo output:

    Number of platforms:                 2
    Platform Profile:                FULL_PROFILE
    Platform Version:                OpenCL 1.1 Mesa 17.1.0-devel - padoka PPA
    Platform Name:               Clover
    Platform Vendor:                 Mesa
    Platform Extensions:                 cl_khr_icd
    Platform Profile:                FULL_PROFILE
    Platform Version:                OpenCL 2.0 AMD-APP (1800.8)
    Platform Name:               AMD Accelerated Parallel Processing
    Platform Vendor:                 Advanced Micro Devices, Inc.
    Platform Extensions:                 cl_khr_icd cl_amd_event_callback cl_amd_offline_devices 


  Platform Name:                 Clover
Number of devices:               1
  Device Type:                   CL_DEVICE_TYPE_GPU
  Vendor ID:                     1002h
  Max compute units:                 8
  Max work items dimensions:             3
    Max work items[0]:               256
    Max work items[1]:               256
    Max work items[2]:               256
  Max work group size:               256
  Preferred vector width char:           16
  Preferred vector width short:          8
  Preferred vector width int:            4
  Preferred vector width long:           2
  Preferred vector width float:          4
  Preferred vector width double:         2
  Native vector width char:          16
  Native vector width short:             8
  Native vector width int:           4
  Native vector width long:          2
  Native vector width float:             4
  Native vector width double:            2
  Max clock frequency:               720Mhz
  Address bits:                  64
  Max memory allocation:             1497105612
  Image support:                 No
  Max size of kernel argument:           1024
  Alignment (bits) of base address:      1024
  Minimum alignment (bytes) for any datatype:    128
  Single precision floating point capability
    Denorms:                     No
    Quiet NaNs:                  Yes
    Round to nearest even:           Yes
    Round to zero:               No
    Round to +ve and infinity:           No
    IEEE754-2008 fused multiply-add:         No
  Cache type:                    None
  Cache line size:               0
  Cache size:                    0
  Global memory size:                2138722304
  Constant buffer size:              1497105612
  Max number of constant args:           16
  Local memory type:                 Scratchpad
  Local memory size:                 32768
  Kernel Preferred work group size multiple:     64
  Error correction support:          0
  Unified memory for Host and Device:        1
  Profiling timer resolution:            0
  Device endianess:              Little
  Available:                     Yes
  Compiler available:                Yes
  Execution capabilities:                
    Execute OpenCL kernels:          Yes
    Execute native function:             No
  Queue on Host properties:              
    Out-of-Order:                No
    Profiling :                  Yes
  Platform ID:                   0x7fea637e5ee0
  Name:                      AMD KAVERI (DRM 2.43.0 / 4.4.0-64-generic, LLVM 5.0.0)
  Vendor:                    AMD
  Device OpenCL C version:           OpenCL C 1.1 
  Driver version:                17.1.0-devel - padoka PPA
  Profile:                   FULL_PROFILE
  Version:                   OpenCL 1.1 Mesa 17.1.0-devel - padoka PPA
  Extensions:                    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_byte_addressable_store cl_khr_fp64


  Platform Name:                 AMD Accelerated Parallel Processing
Number of devices:               1
  Device Type:                   CL_DEVICE_TYPE_CPU
  Vendor ID:                     1002h
  Board name:                    
  Max compute units:                 4
  Max work items dimensions:             3
    Max work items[0]:               1024
    Max work items[1]:               1024
    Max work items[2]:               1024
  Max work group size:               1024
  Preferred vector width char:           16
  Preferred vector width short:          8
  Preferred vector width int:            4
  Preferred vector width long:           2
  Preferred vector width float:          8
  Preferred vector width double:         4
  Native vector width char:          16
  Native vector width short:             8
  Native vector width int:           4
  Native vector width long:          2
  Native vector width float:             8
  Native vector width double:            4
  Max clock frequency:               1700Mhz
  Address bits:                  64
  Max memory allocation:             2147483648
  Image support:                 Yes
  Max number of images read arguments:       128
  Max number of images write arguments:      64
  Max image 2D width:                8192
  Max image 2D height:               8192
  Max image 3D width:                2048
  Max image 3D height:               2048
  Max image 3D depth:                2048
  Max samplers within kernel:            16
  Max size of kernel argument:           4096
  Alignment (bits) of base address:      1024
  Minimum alignment (bytes) for any datatype:    128
  Single precision floating point capability
    Denorms:                     Yes
    Quiet NaNs:                  Yes
    Round to nearest even:           Yes
    Round to zero:               Yes
    Round to +ve and infinity:           Yes
    IEEE754-2008 fused multiply-add:         Yes
  Cache type:                    Read/Write
  Cache line size:               64
  Cache size:                    16384
  Global memory size:                7321481216
  Constant buffer size:              65536
  Max number of constant args:           8
  Local memory type:                 Global
  Local memory size:                 32768
  Max pipe arguments:                16
  Max pipe active reservations:          16
  Max pipe packet size:              2147483648
  Max global variable size:          1879048192
  Max global variable preferred total size:  1879048192
  Max read/write image args:             64
  Max on device events:              0
  Queue on device max size:          0
  Max on device queues:              0
  Queue on device preferred size:        0
  SVM capabilities:              
    Coarse grain buffer:             No
    Fine grain buffer:               No
    Fine grain system:               No
    Atomics:                     No
  Preferred platform atomic alignment:       0
  Preferred global atomic alignment:         0
  Preferred local atomic alignment:      0
  Kernel Preferred work group size multiple:     1
  Error correction support:          0
  Unified memory for Host and Device:        1
  Profiling timer resolution:            1
  Device endianess:              Little
  Available:                     Yes
  Compiler available:                Yes
  Execution capabilities:                
    Execute OpenCL kernels:          Yes
    Execute native function:             Yes
  Queue on Host properties:              
    Out-of-Order:                No
    Profiling :                  Yes
  Queue on Device properties:                
    Out-of-Order:                No
    Profiling :                  No
  Platform ID:                   0x7fea585b9430
  Name:                      AMD A10-7850K APU with Radeon(TM) R7 Graphics
  Vendor:                    AuthenticAMD
  Device OpenCL C version:           OpenCL C 1.2 
  Driver version:                1800.8 (sse2,avx,fma4)
  Profile:                   FULL_PROFILE
  Version:                   OpenCL 1.2 AMD-APP (1800.8)
  Extensions:                    cl_khr_fp64 cl_amd_fp64 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_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_device_fission cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_spir cl_khr_gl_event ```
WhyteHorse
  • 21
  • Well, my Radeon HD 6400M/7400M is much older, so it works with old fglrx drivers which require older LInux kernel and provided that there is no opensource radeon driver in the system. However, it can probably work for you, as A10-7850k first appeared in 2014. I've managed to get it working on Debian 8, and it seems the same should be possible for, say, Ubuntu 14.04.4 (not 14.04.5 !) where older kernel is still used. It is possible to create custom live distribution for that by installing fglrx and AMD SDK in chroot, older lubuntu releases are still accessible. – Slowpoke Feb 22 '17 at 12:20
  • If amdgpu-pro(which should support opencl1.2) does not support your card and there is a possibility to rebuild amdgpu driver (I don't know), may be you can try to do it by uninstalling mesa OpenCL completely and making links for all .so files from AMD SDK to /usr/lib,/usr/lib/x86_64-linux-gnu/ and linking them for the build... I've also read somewhere that even OpenCL 1.1 implementation from mesa is not full-functional. – Slowpoke Feb 22 '17 at 12:23
  • P.S. AMD SDK needs fixing one more broken link to libOpenCL.soin its /opt/AMDAPPSDK-3.0/lib/x86_64/ folder, which points to /usr/lib instead of native libOpenCL.so.1 one in /opt/AMDAPPSDK-3.0/lib/x86_64/sdk/ (while there is nothing in /usr/lib or mesa one is present instead), but you probably know that already. But it is still better to try to get rid of mesa implementation completely and provide the links for all .so files into your /usr/lib,/usr/lib/x86_64-linux-gnu/, just to be sure that amdgpu is indeed not able to use it. – Slowpoke Feb 22 '17 at 12:31