cancel
Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for 
Search instead for 
Did you mean: 

OpenCL

george72
Adept I
‎08-08-2018 04:31 AM

OpenCL linker hangs & terminates application on R9 200

After shipping our application, some users with AMD R9 200 series cards report the application hangs up and then quits. After studying log files and minidumps it seems the issue is the with OpenCL linker on those systems as they show the application is executing a clLinkProgram command at the time of the "hang".

As we handle and log both C++ and structured exceptions and the log doesn't show any of these, our conclusion is the linker somehow terminates the entire application (we have seen this behavior before with Intel OpenCL drivers).

The issue occurs on the following system:

Microsoft Windows 10 Home, Version 1803, OS Build 17134, Current Version 6.3

Intel(R) Core(TM) i7-4790 CPU @ 3.60GHz, 8130 MB RAM

CL Info:

(1.0)   CL_PLATFORM_NAME              : AMD Accelerated Parallel Processing
(1.0)   CL_PLATFORM_VERSION           : OpenCL 2.1 AMD-APP (2580.6)
(1.0)   CL_PLATFORM_PROFILE           : FULL_PROFILE
(1.0)   CL_PLATFORM_VENDOR            : Advanced Micro Devices, Inc.
    CL_PLATFORM_EXTENSIONS:
        cl_khr_icd
        cl_khr_d3d10_sharing
        cl_khr_d3d11_sharing
        cl_khr_dx9_media_sharing
        cl_amd_event_callback
        cl_amd_offline_devices
      

CL_DEVICE[0]

(1.0)   CL_DEVICE_TYPE                                : GPU
(1.0)   CL_DEVICE_NAME                                : Tahiti
(1.0)   CL_DEVICE_VERSION                             : OpenCL 1.2 AMD-APP (2580.6)
(1.0)   CL_DEVICE_PROFILE                             : FULL_PROFILE
(1.0)   CL_DEVICE_VENDOR                              : Advanced Micro Devices, Inc.
(1.0)   CL_DEVICE_VENDOR_ID                           : 4098
(1.0)   CL_DRIVER_VERSION                             : 2580.6
        CL_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_amd_device_attribute_query
            cl_amd_vec3
            cl_amd_printf
            cl_amd_media_ops
            cl_amd_media_ops2
            cl_amd_popcnt
            cl_khr_d3d10_sharing
            cl_khr_d3d11_sharing
            cl_khr_dx9_media_sharing
            cl_khr_image2d_from_buffer
            cl_khr_spir
            cl_khr_gl_event
            cl_amd_liquid_flash
          
(AMD)   CL_DEVICE_PROFILING_TIMER_OFFSET_AMD          : 1532807364555193883
(AMD)   CL_DEVICE_TOPOLOGY_AMD
(AMD)     type                                        : CL_DEVICE_TOPOLOGY_TYPE_PCIE_AMD
(AMD)     bus                                         : 1
(AMD)     device                                      : 0
(AMD)     function                                    : 0
(AMD)   CL_DEVICE_BOARD_NAME_AMD                      : AMD Radeon R9 200 Series
(AMD)   CL_DEVICE_GLOBAL_FREE_MEMORY_AMD              : 3095289
(AMD)   CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD           : 4
(AMD)   CL_DEVICE_SIMD_WIDTH_AMD                      : 16
(AMD)   CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD          : 1
(AMD)   CL_DEVICE_WAVEFRONT_WIDTH_AMD                 : 64
(AMD)   CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD             : 12
(AMD)   CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD        : 16
(AMD)   CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD   : 256
(AMD)   CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD : 65536
(AMD)   CL_DEVICE_LOCAL_MEM_BANKS_AMD                 : 32
(AMD)   CL_DEVICE_THREAD_TRACE_SUPPORTED_AMD          : 0
(AMD)   CL_DEVICE_GFXIP_MAJOR_AMD                     : 6
(AMD)   CL_DEVICE_GFXIP_MINOR_AMD                     : 0
(AMD)   CL_DEVICE_AVAILABLE_ASYNC_QUEUES_AMD          : 2
(AMD)   CL_DEVICE_PREFERRED_WORK_GROUP_SIZE_AMD       : 256
(AMD)   CL_DEVICE_MAX_WORK_GROUP_SIZE_AMD             : 1024
(AMD)   CL_DEVICE_PREFERRED_CONSTANT_BUFFER_SIZE_AMD  : 16384
(1.0)   CL_DEVICE_MAX_COMPUTE_UNITS                   : 28
(1.0)   CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS            : 3
(1.0)   CL_DEVICE_MAX_WORK_ITEM_SIZES                 : [1024, 1024, 1024]
(1.0)   CL_DEVICE_MAX_WORK_GROUP_SIZE                 : 256
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR         : 4
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT        : 2
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT          : 1
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG         : 1
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT        : 1
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE       : 1
(1.1)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF         : 1
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR            : 4
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT           : 2
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_INT             : 1
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG            : 1
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT           : 1
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE          : 1
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF            : 1
(1.0)   CL_DEVICE_MAX_CLOCK_FREQUENCY                 : 940 MHz
(1.0)   CL_DEVICE_ADDRESS_BITS                        : 32
(1.0)   CL_DEVICE_MAX_MEM_ALLOC_SIZE                  : 2509871513 bytes
(1.0)   CL_DEVICE_IMAGE_SUPPORT                       : 1
(1.0)   CL_DEVICE_MAX_READ_IMAGE_ARGS                 : 128
(1.0)   CL_DEVICE_MAX_WRITE_IMAGE_ARGS                : 8
(1.0)   CL_DEVICE_IMAGE2D_MAX_WIDTH                   : 16384 pixels
(1.0)   CL_DEVICE_IMAGE2D_MAX_HEIGHT                  : 16384 pixels
(1.0)   CL_DEVICE_IMAGE3D_MAX_WIDTH                   : 2048 pixels
(1.0)   CL_DEVICE_IMAGE3D_MAX_HEIGHT                  : 2048 pixels
(1.0)   CL_DEVICE_IMAGE3D_MAX_DEPTH                   : 2048 pixels
(1.2)   CL_DEVICE_IMAGE_MAX_BUFFER_SIZE               : 134217728 pixels
(1.2)   CL_DEVICE_IMAGE_MAX_ARRAY_SIZE                : 2048
(1.0)   CL_DEVICE_MAX_SAMPLERS                        : 16
(1.0)   CL_DEVICE_MAX_PARAMETER_SIZE                  : 1024 bytes
(1.0)   CL_DEVICE_MEM_BASE_ADDR_ALIGN                 : 2048 bits
(1.0)   CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE            : 128 bytes
(1.0)   CL_DEVICE_SINGLE_FP_CONFIG                    : CL_FP_INF_NAN CL_FP_ROUND_TO_NEAREST CL_FP_ROUND_TO_ZERO CL_FP_ROUND_TO_INF CL_FP_FMA CL_FP_CORRECTLY_ROUNDED_DIVIDE_SQRT
(1.2)   CL_DEVICE_DOUBLE_FP_CONFIG                    : CL_FP_DENORM CL_FP_INF_NAN CL_FP_ROUND_TO_NEAREST CL_FP_ROUND_TO_ZERO CL_FP_ROUND_TO_INF CL_FP_FMA
(1.0)   CL_DEVICE_GLOBAL_MEM_CACHE_TYPE               : CL_READ_WRITE_CACHE
(1.0)   CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE           : 64 bytes
(1.0)   CL_DEVICE_GLOBAL_MEM_CACHE_SIZE               : 16384 bytes
(1.0)   CL_DEVICE_GLOBAL_MEM_SIZE                     : 3221225472 bytes
(1.0)   CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE            : 65536 bytes
(1.0)   CL_DEVICE_MAX_CONSTANT_ARGS                   : 8
(1.0)   CL_DEVICE_LOCAL_MEM_TYPE                      : CL_LOCAL
(1.0)   CL_DEVICE_LOCAL_MEM_SIZE                      : 32768 bytes
(1.0)   CL_DEVICE_ERROR_CORRECTION_SUPPORT            : 0
(1.1)   CL_DEVICE_HOST_UNIFIED_MEMORY                 : 0
(1.0)   CL_DEVICE_PROFILING_TIMER_RESOLUTION          : 1 ns
(1.0)   CL_DEVICE_ENDIAN_LITTLE                       : 1
(1.0)   CL_DEVICE_AVAILABLE                           : 1
(1.0)   CL_DEVICE_COMPILER_AVAILABLE                  : 1
(1.2)   CL_DEVICE_LINKER_AVAILABLE                    : 1
(1.0)   CL_DEVICE_EXECUTION_CAPABILITIES              : CL_EXEC_KERNEL
(1.0)   CL_DEVICE_QUEUE_PROPERTIES                    : CL_QUEUE_PROFILING_ENABLE
(1.1)   CL_DEVICE_OPENCL_C_VERSION                    : OpenCL C 1.2
(1.2)   CL_DEVICE_PRINTF_BUFFER_SIZE                  : 4194304 bytes
(1.2)   CL_DEVICE_PREFERRED_INTEROP_USER_SYNC         : 1
(1.2)   CL_DEVICE_PARENT_DEVICE                       : none (device is root)
(1.2)   CL_DEVICE_PARTITION_MAX_SUB_DEVICES           : 28
(1.2)   CL_DEVICE_PARTITION_PROPERTIES                :
(1.2)   CL_DEVICE_PARTITION_AFFINITY_DOMAIN           :
(1.2)   CL_DEVICE_PARTITION_TYPE                      :
(1.2)   CL_DEVICE_REFERENCE_COUNT                     : 1

CL_DEVICE[1]

(1.0)   CL_DEVICE_TYPE                                : CPU
(1.0)   CL_DEVICE_NAME                                : Intel(R) Core(TM) i7-4790 CPU @ 3.60GHz
(1.0)   CL_DEVICE_VERSION                             : OpenCL 1.2 AMD-APP (2580.6)
(1.0)   CL_DEVICE_PROFILE                             : FULL_PROFILE
(1.0)   CL_DEVICE_VENDOR                              : GenuineIntel
(1.0)   CL_DEVICE_VENDOR_ID                           : 4098
(1.0)   CL_DRIVER_VERSION                             : 2580.6 (sse2,avx)
        CL_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_d3d10_sharing
            cl_khr_spir
            cl_khr_gl_event
          
(1.0)   CL_DEVICE_MAX_COMPUTE_UNITS                   : 8
(1.0)   CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS            : 3
(1.0)   CL_DEVICE_MAX_WORK_ITEM_SIZES                 : [1024, 1024, 1024]
(1.0)   CL_DEVICE_MAX_WORK_GROUP_SIZE                 : 1024
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR         : 16
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT        : 8
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT          : 4
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG         : 2
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT        : 8
(1.0)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE       : 4
(1.1)   CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF         : 4
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR            : 16
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT           : 8
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_INT             : 4
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG            : 2
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT           : 8
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE          : 4
(1.1)   CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF            : 4
(1.0)   CL_DEVICE_MAX_CLOCK_FREQUENCY                 : 3592 MHz
(1.0)   CL_DEVICE_ADDRESS_BITS                        : 64
(1.0)   CL_DEVICE_MAX_MEM_ALLOC_SIZE                  : 2147483648 bytes
(1.0)   CL_DEVICE_IMAGE_SUPPORT                       : 1
(1.0)   CL_DEVICE_MAX_READ_IMAGE_ARGS                 : 128
(1.0)   CL_DEVICE_MAX_WRITE_IMAGE_ARGS                : 64
(1.0)   CL_DEVICE_IMAGE2D_MAX_WIDTH                   : 8192 pixels
(1.0)   CL_DEVICE_IMAGE2D_MAX_HEIGHT                  : 8192 pixels
(1.0)   CL_DEVICE_IMAGE3D_MAX_WIDTH                   : 2048 pixels
(1.0)   CL_DEVICE_IMAGE3D_MAX_HEIGHT                  : 2048 pixels
(1.0)   CL_DEVICE_IMAGE3D_MAX_DEPTH                   : 2048 pixels
(1.2)   CL_DEVICE_IMAGE_MAX_BUFFER_SIZE               : 65536 pixels
(1.2)   CL_DEVICE_IMAGE_MAX_ARRAY_SIZE                : 2048
(1.0)   CL_DEVICE_MAX_SAMPLERS                        : 16
(1.0)   CL_DEVICE_MAX_PARAMETER_SIZE                  : 4096 bytes
(1.0)   CL_DEVICE_MEM_BASE_ADDR_ALIGN                 : 1024 bits
(1.0)   CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE            : 128 bytes
(1.0)   CL_DEVICE_SINGLE_FP_CONFIG                    : CL_FP_DENORM CL_FP_INF_NAN CL_FP_ROUND_TO_NEAREST CL_FP_ROUND_TO_ZERO CL_FP_ROUND_TO_INF CL_FP_FMA CL_FP_CORRECTLY_ROUNDED_DIVIDE_SQRT
(1.2)   CL_DEVICE_DOUBLE_FP_CONFIG                    : CL_FP_DENORM CL_FP_INF_NAN CL_FP_ROUND_TO_NEAREST CL_FP_ROUND_TO_ZERO CL_FP_ROUND_TO_INF CL_FP_FMA
(1.0)   CL_DEVICE_GLOBAL_MEM_CACHE_TYPE               : CL_READ_WRITE_CACHE
(1.0)   CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE           : 64 bytes
(1.0)   CL_DEVICE_GLOBAL_MEM_CACHE_SIZE               : 32768 bytes
(1.0)   CL_DEVICE_GLOBAL_MEM_SIZE                     : 8525291520 bytes
(1.0)   CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE            : 65536 bytes
(1.0)   CL_DEVICE_MAX_CONSTANT_ARGS                   : 8
(1.0)   CL_DEVICE_LOCAL_MEM_TYPE                      : CL_GLOBAL
(1.0)   CL_DEVICE_LOCAL_MEM_SIZE                      : 32768 bytes
(1.0)   CL_DEVICE_ERROR_CORRECTION_SUPPORT            : 0
(1.1)   CL_DEVICE_HOST_UNIFIED_MEMORY                 : 1
(1.0)   CL_DEVICE_PROFILING_TIMER_RESOLUTION          : 285 ns
(1.0)   CL_DEVICE_ENDIAN_LITTLE                       : 1
(1.0)   CL_DEVICE_AVAILABLE                           : 1
(1.0)   CL_DEVICE_COMPILER_AVAILABLE                  : 1
(1.2)   CL_DEVICE_LINKER_AVAILABLE                    : 1
(1.0)   CL_DEVICE_EXECUTION_CAPABILITIES              : CL_EXEC_KERNEL CL_EXEC_NATIVE_KERNEL
(1.0)   CL_DEVICE_QUEUE_PROPERTIES                    : CL_QUEUE_PROFILING_ENABLE
(1.1)   CL_DEVICE_OPENCL_C_VERSION                    : OpenCL C 1.2
(1.2)   CL_DEVICE_PRINTF_BUFFER_SIZE                  : 65536 bytes
(1.2)   CL_DEVICE_PREFERRED_INTEROP_USER_SYNC         : 1
(1.2)   CL_DEVICE_PARENT_DEVICE                       : none (device is root)
(1.2)   CL_DEVICE_PARTITION_MAX_SUB_DEVICES           : 8
(1.2)   CL_DEVICE_PARTITION_PROPERTIES                : CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN
(1.2)   CL_DEVICE_PARTITION_AFFINITY_DOMAIN           : CL_DEVICE_AFFINITY_DOMAIN_L3_CACHE CL_DEVICE_AFFINITY_DOMAIN_L2_CACHE CL_DEVICE_AFFINITY_DOMAIN_L1_CACHE CL_DEVICE_AFFINITY_DOMAIN_NEXT_PARTITIONABLE
(1.2)   CL_DEVICE_PARTITION_TYPE                      :
(1.2)   CL_DEVICE_REFERENCE_COUNT                     : 1
0 Likes
6 Replies
dipak
Big Boss
‎08-08-2018 04:58 AM

Hi george72 ,

Can you please provide a minimal test-case to reproduce the issue?

Thanks.

0 Likes
george72
In response to dipak
Adept I
‎08-08-2018 05:43 AM

Hi dipak,

To reproduce the issue:

1. download the following archive: http://militaryoperationshq.com/MilOps%20Benchmark%20-%20AMD%20Link%20Failure.7z and unpack it

2. run "MilOps Benchmark.exe"

3. during the compilation/linking step you will see this screen, on the system where it fails it hangs here and then terminates

pastedImage_3.png

6. if everything goes well, you should see the main menu:

pastedImage_2.png

NOTE: due to missing resources, you cannot do anything else except quit.

Based on the CL information, we think the issue occurs on the R9 280 video card, we have tested on R9 290 cards and there it works as expected.

0 Likes
dipak
In response to george72
Big Boss
‎08-08-2018 06:18 AM

Hi George,

Thank you for providing the repro and sharing the information.

Based on the CL information, we think the issue occurs on the R9 280 video card, we have tested on R9 290 cards and there it works as expected.

The issue might be related to SI cards, that why it occurs on R9 280 (SI) but not on R9 290 (CI). Anyway, we will check and get back to you.

BTW, Adrenalin 18.8.1 is available now and if you haven't checked with this driver, please try and share your observation.

0 Likes
george72
In response to dipak
Adept I
‎08-13-2018 09:28 AM

dipak wrote:

BTW, Adrenalin 18.8.1 is available now and if you haven't checked with this driver, please try and share your observation.

Just to let you know: the issue also occurs with Adrenalin 18.8.1 drivers

0 Likes
dipak
In response to george72
Big Boss
‎08-16-2018 06:44 AM

Thanks for confirming this. The concerned team is investigating the issue. If I get any update from the team, I'll let you know.

0 Likes
dipak
In response to george72
Big Boss
‎08-10-2018 04:01 AM

Hi,

I've opened a ticket against this issue. Once I've any update, I'll get back to you.

Thanks.   

0 Likes