Computation errors on OpenCL on Radeon

Forums › Problems and Bug Reports

Caterpillar
Caterpillar
Joined: 8 Jul 10
Posts: 12
Credit: 144552638
RAC: 0
6 Jun 2017 10:14:50 UTC
Topic 208119
(moderation:
    )

    Hi, I am the BOINC maintainer for Fedora / RHEL / CentOS.

    Before talking about computation errors on OpenCL on Radeon, I have to introduce you in the whole procedure I have followed.

    Recently, I investigated to figure out why boinc-client, while running as a service, could not detect videocard for GPU calculus. In order to fix this problem I had to add Group=video to boinc-client systemd unit file. At the moment I haven't yet pushed the change into stable repositories, but I will do it soon.

    AMDGPU-Pro drivers are not available for Fedora, so in order to use OpenCL propertary driver, you have to download AMDGPU-Pro drivers for CentOS 7, unpack libdrm, libAMDOpenCL, libkms RPMs and put them under

     

    /opt/amdgpu-pro/

     So you will have those driver files under

    /opt/amdgpu-pro/lib64

     Don't forget the file amdocl64.icd has to be copied in the path

    /etc/OpenCL/vendors/

    Ok once you have done, simply add

    Environment=LD_LIBRARY_PATH=/opt/amdgpu-pro/lib64

    to boinc working unit file. Set SELinux to permissive, or create a new custom rule to let boinc-client use such path for OpenCL drivers. Then restart boinc-client service.

    Ok so now boinc-client is able to see OpenCL AMDGPU-Pro device and use it. By the way, all Einstein@home working units fail. Milkyway working units instead, are working perfectly.

    Here under I attach some boinc-client logs and the boinc-client systemd unit file

     

     

    mar 06 giu 2017 10:39:48 CEST | Einstein@Home | project resumed by user
mar 06 giu 2017 10:39:52 CEST | Einstein@Home | Sending scheduler request: To report completed tasks.
mar 06 giu 2017 10:39:52 CEST | Einstein@Home | Reporting 2 completed tasks
mar 06 giu 2017 10:39:52 CEST | Einstein@Home | Requesting new tasks for AMD/ATI GPU
mar 06 giu 2017 10:39:53 CEST | Einstein@Home | Computation for task LATeah0030L_1052.0_0_0.0_14738720_0 finished
mar 06 giu 2017 10:39:53 CEST | Einstein@Home | Output file LATeah0030L_1052.0_0_0.0_14738720_0_0 for task LATeah0030L_1052.0_0_0.0_14738720_0 absent
mar 06 giu 2017 10:39:53 CEST | Einstein@Home | Output file LATeah0030L_1052.0_0_0.0_14738720_0_1 for task LATeah0030L_1052.0_0_0.0_14738720_0 absent
mar 06 giu 2017 10:39:53 CEST | Einstein@Home | Starting task LATeah0030L_1052.0_0_0.0_14846650_0
mar 06 giu 2017 10:39:54 CEST | Einstein@Home | Scheduler request completed: got 5 new tasks
mar 06 giu 2017 10:39:55 CEST | Einstein@Home | Computation for task LATeah0030L_1052.0_0_0.0_14846650_0 finished
mar 06 giu 2017 10:39:55 CEST | Einstein@Home | Output file LATeah0030L_1052.0_0_0.0_14846650_0_0 for task LATeah0030L_1052.0_0_0.0_14846650_0 absent
mar 06 giu 2017 10:39:55 CEST | Einstein@Home | Output file LATeah0030L_1052.0_0_0.0_14846650_0_1 for task LATeah0030L_1052.0_0_0.0_14846650_0 absent
mar 06 giu 2017 10:39:56 CEST | Einstein@Home | Started download of templates_LATeah0030L_1156_19851590.dat
mar 06 giu 2017 10:39:56 CEST | Einstein@Home | Started download of templates_LATeah0030L_1156_19431165.dat
mar 06 giu 2017 10:39:58 CEST | Einstein@Home | Finished download of templates_LATeah0030L_1156_19431165.dat
mar 06 giu 2017 10:39:58 CEST | Einstein@Home | Started download of templates_LATeah0030L_1156_19432420.dat
mar 06 giu 2017 10:39:58 CEST | Einstein@Home | Starting task LATeah0030L_1156.0_0_0.0_19431165_1
mar 06 giu 2017 10:39:59 CEST | Einstein@Home | Finished download of templates_LATeah0030L_1156_19851590.dat
mar 06 giu 2017 10:39:59 CEST | Einstein@Home | Finished download of templates_LATeah0030L_1156_19432420.dat
mar 06 giu 2017 10:39:59 CEST | Einstein@Home | Started download of templates_LATeah0030L_1156_21105335.dat
mar 06 giu 2017 10:39:59 CEST | Einstein@Home | Started download of templates_LATeah0030L_1156_21134200.dat
mar 06 giu 2017 10:40:00 CEST | Einstein@Home | Finished download of templates_LATeah0030L_1156_21105335.dat
mar 06 giu 2017 10:40:00 CEST | Einstein@Home | Finished download of templates_LATeah0030L_1156_21134200.dat
mar 06 giu 2017 10:40:01 CEST | Einstein@Home | Computation for task LATeah0030L_1156.0_0_0.0_19431165_1 finished
mar 06 giu 2017 10:40:01 CEST | Einstein@Home | Output file LATeah0030L_1156.0_0_0.0_19431165_1_0 for task LATeah0030L_1156.0_0_0.0_19431165_1 absent
mar 06 giu 2017 10:40:01 CEST | Einstein@Home | Output file LATeah0030L_1156.0_0_0.0_19431165_1_1 for task LATeah0030L_1156.0_0_0.0_19431165_1 absent
mar 06 giu 2017 10:40:01 CEST | Einstein@Home | Starting task LATeah0030L_1156.0_0_0.0_19851590_0
mar 06 giu 2017 10:40:03 CEST | Einstein@Home | project suspended by user
mar 06 giu 2017 10:40:04 CEST | Einstein@Home | Computation for task LATeah0030L_1156.0_0_0.0_19851590_0 finished
mar 06 giu 2017 10:40:04 CEST | Einstein@Home | Output file LATeah0030L_1156.0_0_0.0_19851590_0_0 for task LATeah0030L_1156.0_0_0.0_19851590_0 absent
mar 06 giu 2017 10:40:04 CEST | Einstein@Home | Output file LATeah0030L_1156.0_0_0.0_19851590_0_1 for task LATeah0030L_1156.0_0_0.0_19851590_0 absent

     

    [Unit]
Description=Berkeley Open Infrastructure Network Computing Client
Documentation=man:boinc(1)
After=network-online.target

    [Service]
    Type=forking
    Nice=10
    User=boinc
    WorkingDirectory=/var/lib/boinc
    ExecStart=/usr/bin/boinc_client --daemon --start_delay 1
    ExecStop=/usr/bin/boinccmd --quit
    ExecReload=/usr/bin/boinccmd --read_cc_config
    ExecStopPost=/bin/rm -f /var/lib/boinc/lockfile
    IOSchedulingClass=idle
    Environment=LD_LIBRARY_PATH=/opt/amdgpu-pro/lib64
    SupplementaryGroups=video

    [Install]
    WantedBy=multi-user.target

    Holmis
    Joined: 4 Jan 05
    Posts: 1118
    Credit: 1055935564
    RAC: 0

    Because your computers are

    6 Jun 2017 17:02:37 UTC
    Message 158625
    (moderation:
      )

      Because your computers are hidden there is no way to look at the "stderr output" of a fail tasks to get a hint as to why it failed. Boincs event log doesn't give any hint as to why it failed, only that it did.
      Please either unhide your computers or post a link to one of the failed tasks, like this:

      https://einsteinathome.org/task/653309030

      Caterpillar
      Caterpillar
      Joined: 8 Jul 10
      Posts: 12
      Credit: 144552638
      RAC: 0

      Some of

      7 Jun 2017 14:05:40 UTC
      Message 158652
      (moderation:
        )

        Some of them

        https://einsteinathome.org/task/653178780

        https://einsteinathome.org/task/653176693

        https://einsteinathome.org/task/653178730

        Thank you

        Holmis
        Joined: 4 Jan 05
        Posts: 1118
        Credit: 1055935564
        RAC: 0

        Don't thank me yet , I'm not

        7 Jun 2017 18:40:46 UTC
        Message 158656
        (moderation:
          )

          Don't thank me yet Wink, I'm not a Linux user so can't give any specific help but maybe someone else might help out now that we have more info to go on.

          It seems all of the linked tasks fail with the following errors (all copied from different parts of the task page and stderr outpu):

          Exit status: 6 (0x00000006) Unknown error code

          process exited with code 6 (0x6, -250)

          free(): invalid pointer: 0x00007f8a49b84780

          -- signal handler called: signal 6

          Maybe one of the devs can shed some light on what "signal 6" means?

          As you've been playing around with getting the GPU to run tasks while Boinc is installed in service mode I would suspect that there's still something amiss with the driver installation or the science applications ability to use OpenCL.

          Caterpillar
          Caterpillar
          Joined: 8 Jul 10
          Posts: 12
          Credit: 144552638
          RAC: 0

          https://stackoverflow.com/que

          7 Jun 2017 18:53:17 UTC
          Message 158657
          (moderation:
            )

            https://stackoverflow.com/questions/3413166/when-does-a-process-get-sigabrt-signal-6

            AgentB
            AgentB
            Joined: 17 Mar 12
            Posts: 915
            Credit: 513211304
            RAC: 0

            Something does not look right

            7 Jun 2017 20:12:01 UTC
            Message 158659 in response to message 158657
            (moderation:
              )

              Something does not look right in the backtrace - well to my untrained eye.  Whenever i see LD= being needed something is wrong with the Force....

              here i see this

              /lib64/libMesaOpenCL.so.1

              suggesting the library calls from the Mesa OpenCL libraries but later

              /opt/amdgpu-pro/lib64/libOpenCL.so.1(+0x2733)[0x7f8a4deb3733]

              suggesting calls from AMDGPUPRO OpenCL libraries.

              I'd suggest a close look at why that has happened... ldd and ldconfig and see what the executables are needing.

              Perhaps clinfo might be revealing.

              good luck...

               

              Caterpillar
              Caterpillar
              Joined: 8 Jul 10
              Posts: 12
              Credit: 144552638
              RAC: 0

              AgentB

              7 Jun 2017 21:06:34 UTC
              Message 158663 in response to message 158659
              (moderation:
                )

                AgentB wrote:

                /lib64/libMesaOpenCL.so.1

                suggesting the library calls from the Mesa OpenCL libraries but later

                /opt/amdgpu-pro/lib64/libOpenCL.so.1(+0x2733)[0x7f8a4deb3733

                 

                The reason is in first message

                Gary Roberts
                Gary Roberts
                Moderator
                Joined: 9 Feb 05
                Posts: 5850
                Credit: 110027676131
                RAC: 22474641

                Caterpillar wrote:AMDGPU-Pro

                9 Jun 2017 1:25:36 UTC
                Message 158699
                (moderation:
                  )

                  Caterpillar wrote:

                  AMDGPU-Pro drivers are not available for Fedora, so in order to use OpenCL propertary driver, you have to download AMDGPU-Pro drivers for CentOS 7, unpack libdrm, libAMDOpenCL, libkms RPMs and put them under

                   

                  /opt/amdgpu-pro/

                  I use Linux (PCLinuxOS) on a large number of machines, virtually all with usable GPUs these days.  I'm not a programmer, just a self taught user and I wouldn't regard my knowledge of how things really work as being all that great.  Often by trial and error I get things to work without really understanding the finer details of how and why.

                  About 4 months ago I decided to investigate what was needed to get an RX460 GPU running.  My distro has the amdgpu free driver (not -pro) so the card is detected and usable for producing a GUI but of course not detected by BOINC.  I downloaded the -pro driver from AMD and unpacked it all and perused the install script.  I identified what I thought would be needed from the whole package to provide OpenCL functionality and installed the bits pretty much as you describe above.  I was pleasantly surprised to find that BOINC was able to detect the card and use it and the results were validated just fine.

                  I now have a total of 12 RX460 cards installed mainly in older machines and all working fine.  I put all the bits on a file share and wrote a small install script to automate the whole setup process.  There was some instability with the original amdgpu driver (the graphics would crash after about a week of running) but an update to the driver seems to have solved that.  Here is a link to one of the oldies that I upgraded with the RX460s.  It has a Q6600 quad core CPU and it has been crunching (CPU only until the upgrade) 24/7 for almost 9 years.  I had intended retiring a bunch of similar vintage machines but the RX460 upgrade has given them a new lease on life.  I've restricted these hosts to just one CPU task alongside 2 concurrent GPU tasks to keep the power use down a bit.  I'm still using the same PSUs as before when the machines crunched 4 CPU tasks.  The Asus RX460 cards I'm using don't have a PCIe power connector.

                  I don't know why tasks are failing with your setup.  The main reason I'm posting is to suggest that what you are doing should work, since it does for me.  There must be something with your setup that is interfering.  My distro doesn't use systemd (and never will according to the developers).  I use the Berkeley shell archive for BOINC since my distro doesn't (and wont) maintain it in their repo.  All bar one of my machines run 7.2.42.  I have downloaded and built 7.6.33 from source and it runs fine too (on one machine).  One of these days I might get around to deploying it over the entire fleet.  The changes in the user interface between the two are sufficiently and annoyingly different to discourage a piecemeal approach so I don't have much enthusiasm to get started.  I'm a firm believer in, "If it ain't broke, don't fix it" :-).

                   

                  Cheers,
                  Gary.

                  Caterpillar
                  Caterpillar
                  Joined: 8 Jul 10
                  Posts: 12
                  Credit: 144552638
                  RAC: 0

                  ping

                  12 Jun 2017 21:56:16 UTC
                  Message 158779
                  (moderation:
                    )

                    ping

                    Caterpillar
                    Caterpillar
                    Joined: 8 Jul 10
                    Posts: 12
                    Credit: 144552638
                    RAC: 0

                    $

                    15 Jun 2017 13:53:03 UTC
                    Message 158886
                    (moderation:
                      )

                      $ LD_LIBRARY_PATH=/opt/amdgpu-pro/lib64 clinfo Number of platforms                               3   Platform Name                                   Clover   Platform Vendor                                 Mesa   Platform Version                                OpenCL 1.1 Mesa 17.0.5   Platform Profile                                FULL_PROFILE   Platform Extensions                             cl_khr_icd   Platform Extensions function suffix             MESA

                        Platform Name                                   Portable Computing Language
                        Platform Vendor                                 The pocl project
                        Platform Version                                OpenCL 2.0 pocl 0.14-pre, LLVM 3.9.1
                        Platform Profile                                FULL_PROFILE
                        Platform Extensions                             cl_khr_icd
                        Platform Extensions function suffix             POCL

                        Platform Name                                   AMD Accelerated Parallel Processing
                        Platform Vendor                                 Advanced Micro Devices, Inc.
                        Platform Version                                OpenCL 2.0 AMD-APP (2348.3)
                        Platform Profile                                FULL_PROFILE
                        Platform Extensions                             cl_khr_icd cl_amd_event_callback cl_amd_offline_devices
                        Platform Extensions function suffix             AMD

                        Platform Name                                   Clover
                      Number of devices                                 1
                        Device Name                                     AMD POLARIS10 (DRM 3.10.0 / 4.11.3-202.fc25.x86_64, LLVM 3.9.1)
                        Device Vendor                                   AMD
                        Device Vendor ID                                0x1002
                        Device Version                                  OpenCL 1.1 Mesa 17.0.5
                        Driver Version                                  17.0.5
                        Device OpenCL C Version                         OpenCL C 1.1
                        Device Type                                     GPU
                        Device Available                                Yes
                        Device Profile                                  FULL_PROFILE
                        Max compute units                               36
                        Max clock frequency                             1288MHz
                        Max work item dimensions                        3
                        Max work item sizes                             256x256x256
                        Max work group size                             256
                        Compiler Available                              Yes
                      === CL_PROGRAM_BUILD_LOG ===
                      fatal error: cannot open file '/usr/lib64/clc//polaris10-amdgcn--.bc': No such file or directory
                        Preferred work group size multiple              <getWGsizes:500: build program : error -11>
                        Preferred / native vector sizes                 
                          char                                                16 / 16      
                          short                                                8 / 8       
                          int                                                  4 / 4       
                          long                                                 2 / 2       
                          half                                                 0 / 0        (n/a)
                          float                                                4 / 4       
                          double                                               2 / 2        (cl_khr_fp64)
                        Half-precision Floating-point support           (n/a)
                        Single-precision Floating-point support         (core)
                          Denormals                                     No
                          Infinity and NANs                             Yes
                          Round to nearest                              Yes
                          Round to zero                                 No
                          Round to infinity                             No
                          IEEE754-2008 fused multiply-add               No
                          Support is emulated in software               No
                          Correctly-rounded divide and sqrt operations  No
                        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                              8588079104 (7.998GiB)
                        Error Correction support                        No
                        Max memory allocation                           6011655372 (5.599GiB)
                        Unified memory for Host and Device              Yes
                        Minimum alignment for any data type             128 bytes
                        Alignment of base address                       1024 bits (128 bytes)
                        Global Memory cache type                        None
                        Image support                                   No
                        Local memory type                               Local
                        Local memory size                               32768 (32KiB)
                        Max constant buffer size                        2147483647 (2GiB)
                        Max number of constant args                     16
                        Max size of kernel argument                     1024
                        Queue properties                                
                          Out-of-order execution                        No
                          Profiling                                     Yes
                        Profiling timer resolution                      0ns
                        Execution capabilities                          
                          Run OpenCL kernels                            Yes
                          Run native kernels                            No
                        Device 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                                   Portable Computing Language
                      Number of devices                                 1
                        Device Name                                     pthread-AMD Phenom(tm) II X4 965 Processor
                        Device Vendor                                   AuthenticAMD
                        Device Vendor ID                                0x1002
                        Device Version                                  OpenCL 2.0 pocl HSTR: pthread-x86_64-unknown-linux-gnu-haswell
                        Driver Version                                  0.14-pre
                        Device OpenCL C Version                         OpenCL C 2.0
                        Device Type                                     CPU, Default
                        Device Available                                Yes
                        Device Profile                                  FULL_PROFILE
                        Max compute units                               4
                        Max clock frequency                             3400MHz
                        Device Partition                                (core)
                          Max number of sub-devices                     4
                          Supported partition types                     equally, by counts
                        Max work item dimensions                        3
                        Max work item sizes                             4096x4096x4096
                        Max work group size                             4096
                        Compiler Available                              Yes
                        Linker Available                                Yes
                        Preferred work group size multiple              8
                        Preferred / native vector sizes                 
                          char                                                16 / 16      
                          short                                                8 / 8       
                          int                                                  4 / 4       
                          long                                                 2 / 2       
                          half                                                 8 / 8        (n/a)
                          float                                                4 / 4       
                          double                                               2 / 2        (cl_khr_fp64)
                        Half-precision Floating-point support           (n/a)
                        Single-precision Floating-point support         (core)
                          Denormals                                     No
                          Infinity and NANs                             Yes
                          Round to nearest                              Yes
                          Round to zero                                 No
                          Round to infinity                             No
                          IEEE754-2008 fused multiply-add               No
                          Support is emulated in software               No
                          Correctly-rounded divide and sqrt operations  No
                        Double-precision Floating-point support         (cl_khr_fp64)
                          Denormals                                     No
                          Infinity and NANs                             Yes
                          Round to nearest                              Yes
                          Round to zero                                 No
                          Round to infinity                             No
                          IEEE754-2008 fused multiply-add               No
                          Support is emulated in software               No
                          Correctly-rounded divide and sqrt operations  No
                        Address bits                                    64, Little-Endian
                        Global memory size                              18970021888 (17.67GiB)
                        Error Correction support                        No
                        Max memory allocation                           18970021888 (17.67GiB)
                        Unified memory for Host and Device              Yes
                        Shared Virtual Memory (SVM) capabilities        (core)
                          Coarse-grained buffer sharing                 Yes
                          Fine-grained buffer sharing                   Yes
                          Fine-grained system sharing                   No
                          Atomics                                       Yes
                        Minimum alignment for any data type             128 bytes
                        Alignment of base address                       1024 bits (128 bytes)
                        Preferred alignment for atomics                 
                          SVM                                           0 bytes
                          Global                                        0 bytes
                          Local                                         0 bytes
                        Max size for global variable                    0
                        Preferred total size of global vars             0
                        Global Memory cache type                        Read/Write
                        Global Memory cache size                        65536 (64KiB)
                        Global Memory cache line                        64 bytes
                        Image support                                   Yes
                          Max number of samplers per kernel             16
                          Max size for 1D images from buffer            1185626368 pixels
                          Max 1D or 2D image array size                 2048 images
                          Max 2D image size                             32768x32768 pixels
                          Max 3D image size                             2048x2048x2048 pixels
                          Max number of read image args                 128
                          Max number of write image args                128
                          Max number of read/write image args           128
                        Max number of pipe args                         16
                        Max active pipe reservations                    1
                        Max pipe packet size                            1024
                        Local memory type                               Global
                        Local memory size                               18970021888 (17.67GiB)
                        Max constant buffer size                        18970021888 (17.67GiB)
                        Max number of constant args                     8
                        Max size of kernel argument                     1024
                        Queue properties (on host)                      
                          Out-of-order execution                        No
                          Profiling                                     Yes
                        Queue properties (on device)                    
                          Out-of-order execution                        Yes
                          Profiling                                     Yes
                          Preferred size                                16384 (16KiB)
                          Max size                                      262144 (256KiB)
                        Max queues on device                            1
                        Max events on device                            1024
                        Prefer user sync for interop                    Yes
                        Profiling timer resolution                      1ns
                        Execution capabilities                          
                          Run OpenCL kernels                            Yes
                          Run native kernels                            Yes
                          SPIR versions                                 1.2
                        printf() buffer size                            1048576 (1024KiB)
                        Built-in kernels                                
                        Device Extensions                               cl_khr_byte_addressable_store 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_spir cl_khr_int64 cl_khr_fp64 cl_khr_int64_base_atomics cl_khr_int64_extended_atomics

                        Platform Name                                   AMD Accelerated Parallel Processing
                      Number of devices                                 2
                        Device Name                                     Ellesmere
                        Device Vendor                                   Advanced Micro Devices, Inc.
                        Device Vendor ID                                0x1002
                        Device Version                                  OpenCL 1.2 AMD-APP (2348.3)
                        Driver Version                                  2348.3
                        Device OpenCL C Version                         OpenCL C 1.2
                        Device Type                                     GPU
                        Device Available                                Yes
                        Device Profile                                  FULL_PROFILE
                        Device Board Name (AMD)                         AMD Radeon (TM) RX 480 Graphics
                        Device Topology (AMD)                           PCI-E, 01:00.0
                        Max compute units                               36
                        SIMD per compute unit (AMD)                     4
                        SIMD width (AMD)                                16
                        SIMD instruction width (AMD)                    1
                        Max clock frequency                             1288MHz
                        Graphics IP (AMD)                               8.0
                        Device Partition                                (core)
                          Max number of sub-devices                     36
                          Supported partition types                     none specified
                        Max work item dimensions                        3
                        Max work item sizes                             256x256x256
                        Max work group size                             256
                        Compiler Available                              Yes
                        Linker Available                                Yes
                        Preferred work group size multiple              64
                        Wavefront width (AMD)                           64
                        Preferred / native vector sizes                 
                          char                                                 4 / 4       
                          short                                                2 / 2       
                          int                                                  1 / 1       
                          long                                                 1 / 1       
                          half                                                 1 / 1        (cl_khr_fp16)
                          float                                                1 / 1       
                          double                                               1 / 1        (cl_khr_fp64)
                        Half-precision Floating-point support           (cl_khr_fp16)
                          Denormals                                     No
                          Infinity and NANs                             No
                          Round to nearest                              No
                          Round to zero                                 No
                          Round to infinity                             No
                          IEEE754-2008 fused multiply-add               No
                          Support is emulated in software               No
                          Correctly-rounded divide and sqrt operations  No
                        Single-precision Floating-point support         (core)
                          Denormals                                     No
                          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                              8254726144 (7.688GiB)
                        Global free memory (AMD)                        8045444 (7.673GiB)
                        Global memory channels (AMD)                    8
                        Global memory banks per channel (AMD)           16
                        Global memory bank width (AMD)                  256 bytes
                        Error Correction support                        No
                        Max memory allocation                           4244635648 (3.953GiB)
                        Unified memory for Host and Device              No
                        Minimum alignment for any data type             128 bytes
                        Alignment of base address                       2048 bits (256 bytes)
                        Global Memory cache type                        Read/Write
                        Global Memory cache size                        16384 (16KiB)
                        Global Memory cache line                        64 bytes
                        Image support                                   Yes
                          Max number of samplers per kernel             16
                          Max size for 1D images from buffer            134217728 pixels
                          Max 1D or 2D image array size                 2048 images
                          Base address alignment for 2D image buffers   256 bytes
                          Pitch alignment for 2D image buffers          256 bytes
                          Max 2D image size                             16384x16384 pixels
                          Max 3D image size                             2048x2048x2048 pixels
                          Max number of read image args                 128
                          Max number of write image args                8
                        Local memory type                               Local
                        Local memory size                               32768 (32KiB)
                        Local memory syze per CU (AMD)                  65536 (64KiB)
                        Local memory banks (AMD)                        32
                        Max constant buffer size                        4244635648 (3.953GiB)
                        Max number of constant args                     8
                        Max size of kernel argument                     1024
                        Queue properties                                
                          Out-of-order execution                        No
                          Profiling                                     Yes
                        Prefer user sync for interop                    Yes
                        Profiling timer resolution                      1ns
                        Profiling timer offset since Epoch (AMD)        1497511456057028485ns (Thu Jun 15 09:24:16 2017)
                        Execution capabilities                          
                          Run OpenCL kernels                            Yes
                          Run native kernels                            No
                          Thread trace supported (AMD)                  Yes
                          SPIR versions                                 1.2
                        printf() buffer size                            1048576 (1024KiB)
                        Built-in kernels                                
                        Device 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_fp16 cl_khr_gl_sharing cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_image2d_from_buffer cl_khr_spir cl_khr_gl_event

                        Device Name                                     AMD Phenom(tm) II X4 965 Processor
                        Device Vendor                                   AuthenticAMD
                        Device Vendor ID                                0x1002
                        Device Version                                  OpenCL 1.2 AMD-APP (2348.3)
                        Driver Version                                  2348.3 (sse2)
                        Device OpenCL C Version                         OpenCL C 1.2
                        Device Type                                     CPU
                        Device Available                                Yes
                        Device Profile                                  FULL_PROFILE
                        Device Board Name (AMD)                         
                        Device Topology (AMD)                           (n/a)
                        Max compute units                               4
                        Max clock frequency                             3400MHz
                        Device Partition                                (core, cl_ext_device_fission)
                          Max number of sub-devices                     4
                          Supported partition types                     equally, by counts, by affinity domain
                          Supported affinity domains                    L3 cache, L2 cache, L1 cache, next partitionable
                          Supported partition types (ext)               equally, by counts, by affinity domain
                          Supported affinity domains (ext)              L3 cache, L2 cache, L1 cache, next fissionable
                        Max work item dimensions                        3
                        Max work item sizes                             1024x1024x1024
                        Max work group size                             1024
                        Compiler Available                              Yes
                        Linker Available                                Yes
                        Preferred work group size multiple              1
                        Preferred / native vector sizes                 
                          char                                                16 / 16      
                          short                                                8 / 8       
                          int                                                  4 / 4       
                          long                                                 2 / 2       
                          half                                                 2 / 2        (n/a)
                          float                                                4 / 4       
                          double                                               2 / 2        (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                              16822538240 (15.67GiB)
                        Error Correction support                        No
                        Max memory allocation                           4205634560 (3.917GiB)
                        Unified memory for Host and Device              Yes
                        Minimum alignment for any data type             128 bytes
                        Alignment of base address                       1024 bits (128 bytes)
                        Global Memory cache type                        Read/Write
                        Global Memory cache size                        65536 (64KiB)
                        Global Memory cache line                        64 bytes
                        Image support                                   Yes
                          Max number of samplers per kernel             16
                          Max size for 1D images from buffer            65536 pixels
                          Max 1D or 2D image array size                 2048 images
                          Max 2D image size                             8192x8192 pixels
                          Max 3D image size                             2048x2048x2048 pixels
                          Max number of read image args                 128
                          Max number of write image args                64
                        Local memory type                               Global
                        Local memory size                               32768 (32KiB)
                        Max constant buffer size                        65536 (64KiB)
                        Max number of constant args                     8
                        Max size of kernel argument                     4096 (4KiB)
                        Queue properties                                
                          Out-of-order execution                        No
                          Profiling                                     Yes
                        Prefer user sync for interop                    Yes
                        Profiling timer resolution                      1ns
                        Profiling timer offset since Epoch (AMD)        1497511456057028485ns (Thu Jun 15 09:24:16 2017)
                        Execution capabilities                          
                          Run OpenCL kernels                            Yes
                          Run native kernels                            Yes
                          SPIR versions                                 1.2
                        printf() buffer size                            65536 (64KiB)
                        Built-in kernels                                
                        Device 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

                      NULL platform behavior
                        clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...)  No platform
                        clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...)   No platform
                        clCreateContext(NULL, ...) [default]            No platform
                        clCreateContext(NULL, ...) [other]              Success [MESA]
                        clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU)  No platform
                        clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU)  No platform
                        clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR)  No platform
                        clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM)  No platform
                        clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL)  No platform

                       

                      Paul
                      Paul
                      Joined: 3 May 07
                      Posts: 121
                      Credit: 1654643650
                      RAC: 22442

                      Hey CATERPILLAR, thanks for

                      28 Aug 2017 6:22:45 UTC
                      Message 161391
                      (moderation:
                        )

                        Hey CATERPILLAR, thanks for all your work on Fedora.  I've been crunching on Fedora since 1999.

                        I'm having the same problem (for over a year), but with the OSS AMDGPU drivers.  I would love to help troubleshoot from my end.  If E@H folks or you can point me in the right direction, I'd be happy to try something.

                        Comment viewing options

                        Select your preferred way to display the comments and click "Save settings" to activate your changes.

                        Forums › Problems and Bug Reports