Hi everybody,
this is driving me crazy: I have 2 separate XFX HD7900 Devices installed on Ubuntu 12.04.2. clinfo definitely shows two GPU devices available, plus one CPU device. But clGetDeviceIDs will always only return a single GPU available in the system. The same programm running on our main evaluation station (Debian Wheezy, 2x NVIDIA GTX 580) works fine, reporting 2 GPU Devices. On the workstation of my collegue our software also finds the available 2x GPU devices (NVIDIA, Ubuntu 12.04 too). But my AMD setup seems screwed. Any of the sample of the AMD APP SDK, only report a single GPU.
I encountered this problem using regular Ubuntu 12.04.2 fglrx driver. Today, I also removed the ubuntu fglrx driver and manually installed the latest catalyst 13.11 beta V9.4 catalyst driver and latest AMD APP SDK. Still the same issue.
Here are some relevant outputs:
lspci | grep VGA:
01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Tahiti XT [Radeon HD 7970]
02:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Tahiti XT [Radeon HD 7970]
clinfo:
Number of platforms: 1
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 1.2 AMD-APP (1214.3)
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: 2
Device Type: CL_DEVICE_TYPE_GPU
Device ID: 4098
Board name: AMD Radeon HD 7900 Series
Device Topology: PCI[ B#1, D#0, F#0 ]
Max compute units: 32
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: 4
Preferred vector width short: 2
Preferred vector width int: 1
Preferred vector width long: 1
Preferred vector width float: 1
Preferred vector width double: 1
Native vector width char: 4
Native vector width short: 2
Native vector width int: 1
Native vector width long: 1
Native vector width float: 1
Native vector width double: 1
Max clock frequency: 1000Mhz
Address bits: 32
Max memory allocation: 1073741824
Image support: Yes
Max number of images read arguments: 128
Max number of images write arguments: 8
Max image 2D width: 16384
Max image 2D height: 16384
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: 1024
Alignment (bits) of base address: 2048
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: 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: 2872049664
Constant buffer size: 65536
Max number of constant args: 8
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: 0
Profiling timer resolution: 1
Device endianess: Little
Available: Yes
Compiler available: Yes
Execution capabilities:
Execute OpenCL kernels: Yes
Execute native function: No
Queue properties:
Out-of-Order: No
Profiling : Yes
Platform ID: 0x00007fa2c3939fc0
Name: Tahiti
Vendor: Advanced Micro Devices, Inc.
Device OpenCL C version: OpenCL C 1.2
Driver version: 1214.3 (VM)
Profile: FULL_PROFILE
Version: OpenCL 1.2 AMD-APP (1214.3)
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_atomic_counters_32 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
Device Type: CL_DEVICE_TYPE_CPU
Device ID: 4098
Board name:
Max compute units: 8
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: 1600Mhz
Address bits: 64
Max memory allocation: 8423577600
Image support: Yes
Max number of images read arguments: 128
Max number of images write arguments: 8
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: 33694310400
Constant buffer size: 65536
Max number of constant args: 8
Local memory type: Global
Local memory size: 32768
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 properties:
Out-of-Order: No
Profiling : Yes
Platform ID: 0x00007fa2c3939fc0
Name: Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz
Vendor: GenuineIntel
Device OpenCL C version: OpenCL C 1.2
Driver version: 1214.3 (sse2,avx)
Profile: FULL_PROFILE
Version: OpenCL 1.2 AMD-APP (1214.3)
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
env | grep LIB
LIBRARY_PATH=:/opt/AMDAPP/lib/x86_64/
LD_LIBRARY_PATH=:/opt/AMDAPP/lib/x86_64/
One of the AMD APP examples that will use GPU+CPU show only one GPU
/opt/AMDAPP/samples/opencl/bin/x86_64/DeviceFission
Platform 0 : Advanced Micro Devices, Inc.
Platform found : Advanced Micro Devices, Inc.
Selected Platform Vendor : Advanced Micro Devices, Inc.
Device 0 : Tahiti Device ID is 0xf8bae0
Device 1 : Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz Device ID is 0x128d3f0
One of the AMD APP examples that will only use GPU:
/opt/AMDAPP/samples/opencl/bin/x86_64/SobelFilter
Platform 0 : Advanced Micro Devices, Inc.
Platform found : Advanced Micro Devices, Inc.
Selected Platform Vendor : Advanced Micro Devices, Inc.
Device 0 : Tahiti Device ID is 0xf1cde0
Executing kernel for 1 iterations
Please clarify the following
1. In the above clinfo output, where you are seeing 2 GPUs, its reported only one right.
2. Is this is the same case if you install both different GPUs, one from AMD and one from other vendor like intel.
Could you please try this and give me the exact senario so that i can set the system with that configuration and test
The clinfo output shows:
Platform Name: AMD Accelerated Parallel Processing
Number of devices: 2
Device Type: CL_DEVICE_TYPE_GPU
I was assuming the "2" is related to "CL_DEVICE_TYPE_GPU", so detected 2 GPU devices. Now that I read it, it might be possible that the 2 is related to the whole platform.
Hower, please note that the "lspci" linux command output that I pasted definietely shows 2 devices with different BusIDs, so the two devices seems to be recognized by the kernel. I can't replace one of the cards with another one, as I don't have a spare GPU lying around and those NVIDIA installed in collegues computer are required for work.
that number show total number of OpenCL devices. one of them is CPU so there is error in detection of your second card. try run "export DISPLAY=:0" and then run clinfo again.
if that doesn't help then run "sudo amdconfig --adapter=all --initial" restart PC and run export DISPLAY=:0 and clinfo again.
Thank you, that worked! Now clinfo shows one more device and our program is faster by factor 2
But after running "sudo amdconfig --adapter=all --initial" and restarting, the gnome environment fell back to using classic mode, I guess because it couldn't use any 3D acceleration. I don't care about that and will use XFCE for now, but this should be improved in the long run.