Hello, new here. Just a hobbyist experimenting with opencl. Any help would be greatly appreciated.
I recently purchased the Gizmo 2 SBC because I was interested in its OpenCL capability. The Gizmo 2 spec specifys that it has an AMD GX-210HA w/ Radeon HD Graphics. This is an integrated gpu on the same die as the cpu that supports OpenCL 1.2.
I downloaded and installed AMD APP SDK 2.9.1. I ran the clinfo utility to profile the system. Apparently, the sdk sees the integrated gpu and cpu as two different devices. What's even more strange is that for the gpu device, it says the Device OpenCL version is 2.0! But for the cpu device, it lists the OpenCL version as 1.2 (as expected). According to AMD APP SDK 2.9.1 FAQ, this sdk supports Opencl 1.2 , not 2.0.
2. Which versions of the OpenCL™ standard does this SDK support?
AMD APP SDK 2.9.1 supports the development of applications using the OpenCL™ Specification v 1.2.
So I tried some of the included samples in the sdk and discovered something surprising. Most of the programs choose the gpu device by default which is actually slower than choosing the cpu device (which contains the integrated gpu)! Again, this situation is tricky because the cpu and gpu are physically on the same device, so I think the sdk is getting confused somehow. But for sure, when choosing the cpu device to run a program, it runs much faster than choosing the gpu device. However, because it is a dual core cpu, I don't know if the speedup is coming from using both cores of the cpu or if it really is using the integrated gpu portion of the die.
Can anyone provide some insight? Are there any tips/suggestions on how to install/configure OpenCL for a system where the gpu is integrated with the cpu (no discrete gpu). Why/how did I somehow get OpenCL 2.0 when neither device supports it and the sdk I installed, amd app sdk 2.9.1, specifically specified it supports OpenCL 1.2 - not 2.0 ( I did not install any previous versions of OpenCL or amd app sdk) ?
In developing OpenCL programs, I could just explicitly choose to use the cpu device which I know will run faster than the "gpu", but it would be nice to let the OpenCL correctly profile the system and choose the fastest device. My next move is to try to install an older sdk (2.9) and hopefully there will be no listing of OpenCL 2.0 anywhere on the system. Any help would be greatly appreciated.
From running clinfo:
test@test-Gizmo2:/opt/AMDAPPSDK-2.9-1/bin/x86_64$ ./clinfo
Number of platforms: 1
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 2.0 AMD-APP (1702.3)
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
Vendor ID: 1002h
Board name: AMD Radeon(TM) HD 8210E
Device Topology: PCI[ B#0, D#1, F#0 ]
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: 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: 300Mhz
Address bits: 64
Max memory allocation: 134217728
Image support: Yes
Max number of images read arguments: 128
Max number of images write arguments: 64
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: 517996544
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: 1
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: 0x00007fd0bde93630
Name: Kalindi
Vendor: Advanced Micro Devices, Inc.
Device OpenCL C version: OpenCL C 2.0
Driver version: 1702.3 (VM)
Profile: FULL_PROFILE
Version: OpenCL 2.0 AMD-APP (1702.3)
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 cl_khr_spir cl_khr_subgroups cl_khr_gl_event cl_khr_depth_images
Device Type: CL_DEVICE_TYPE_CPU
Vendor ID: 1002h
Board name:
Max compute units: 2
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: 800Mhz
Address bits: 64
Max memory allocation: 762167296
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: 762167296
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: 0x00007fd0bde93630
Name: AMD GX-210HA SOC with Radeon(tm) HD Graphics
Vendor: AuthenticAMD
Device OpenCL C version: OpenCL C 1.2
Driver version: 1702.3 (sse2,avx)
Profile: FULL_PROFILE
Version: OpenCL 1.2 AMD-APP (1702.3)
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 cl_khr_spir cl_khr_gl_event
Solved! Go to Solution.
SDK 2.9.1 supports OCL 1.2. This means that the header file and .lib file that came with the SDK supports OCL 1.2
clinfo reports the capability provided by the OCL runtime/driver. What version of Catalyst driver did you install? AMD started supporting OCL 2.0 in their driver since Catalyst Omega (14.12)
Even with the runtime/driver that supports OCL2.0, you will need the right header file (SDK3.0) to use OCL2.0 capability.
SDK 2.9.1 supports OCL 1.2. This means that the header file and .lib file that came with the SDK supports OCL 1.2
clinfo reports the capability provided by the OCL runtime/driver. What version of Catalyst driver did you install? AMD started supporting OCL 2.0 in their driver since Catalyst Omega (14.12)
Even with the runtime/driver that supports OCL2.0, you will need the right header file (SDK3.0) to use OCL2.0 capability.
Hi,
I was trying to download AMD APP SDK 3.0 Beta from the official site (http://developer.amd.com/tools-and-sdks/opencl-zone/amd-accelerated-parallel-processing-app-sdk/) but failed. I need Windows7-64bit version and the only thing available for Windows is the client Manager which does not work.
Need help please.
Marco
You need to download the AMDSDKInstallManager.exe and it will automatically download and install the suitable SDK package. Please disable the "pop-up" block option (if any) when try to download the install manager.
Regards,
I did it before. But to me it does not work. I tried also on other computers. I run the Install Manager and I choose to download the APP SDK fro later use. No process is going to start, no green bar, the only thing I download is a "AMD-SDKInstaller-v1.3.158-Beta-windows-F-x86" application file of 0 bytes (sometimes, it depends from something that I really don't know, it downloads this "AMD-SDKInstaller-v1.3.158-Beta-windows-W-x64" file, guess for 64bit machines, but without having specified something in particular, of 0 bytes).
Trying to launch it, obviously it doesn't work. "Not a valid Win32 application".
@binghy,
The AMDSDKInstallManager.exe does not support proxy servers as of now, so if you are behind a proxy, then it will fail to download. As an alternate, use the following URLs to download the APP SDK 3.0.0 Beta installers:
Adding support for proxy servers is planned for a future release.
Regards,
Manish
Thank you so much.
Regards,
Marco