Author Topic: AMD Stream SDK 2.3  (Read 14291 times)

0 Members and 1 Guest are viewing this topic.


  • Global Moderator
  • Hero Member
  • *****
  • Posts: 4433
AMD Stream SDK 2.3
« on: December 13, 2010, 07:03:47 PM »
No changelog yet.

Direct downloads:

AMD CLINFO launched on desktop with Intel and NVIDIA SDK  :P

Number of platforms:             2
  Platform Profile:             FULL_PROFILE
  Platform Version:             OpenCL 1.1 WINDOWS
  Platform Name:             Intel OpenCL
  Platform Vendor:             Intel Corporation
  Platform Extensions:             cl_khr_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_gl_sharing cl_khr_byte_addressable_store cl_khr_icd
  Platform Profile:             FULL_PROFILE
  Platform Version:             OpenCL 1.0 CUDA 3.2.1
  Platform Name:             NVIDIA CUDA
  Platform Vendor:             NVIDIA Corporation
  Platform Extensions:             cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_d3d9_sharing cl_nv_d3d10_sharing cl_khr_d3d10_sharing cl_nv_d3d11_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll
ERROR: clgetDeviceInfo(-30)

  Platform Name:             Intel OpenCL
Number of devices:             1
  Device Type:                CL_DEVICE_TYPE_CPU
  Device ID:                32902
  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:          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:             3200Mhz
  Address bits:                18397247599280160
  Max memory allocation:          536838144
  Image support:             Yes
  Max number of images read arguments:       128
  Max number of images write arguments:       128
  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:          128
  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:                Yes
    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:                Read/Write
  Cache line size:             64
  Cache size:                6291456
  Global memory size:             2147352576
  Constant buffer size:             131072
  Max number of constant args:          128
  Local memory type:             Global
  Local memory size:             32768
ERROR: clBuildProgram(-11)


  • Global Moderator
  • Hero Member
  • *****
  • Posts: 4433
Re: AMD Stream SDK 2.3
« Reply #1 on: December 16, 2010, 05:28:01 PM »
What’s New in ATI Stream SDK 2.3

    * Improved OpenCL runtime performance:
          o Improved kernel launch times.
          o Improved PCIe transfer times.
          o Enabled DRMDMA for Evergreen GPUs, as well as AMD Radeon™ HD 6870 and AMD Radeon™ HD 6850 devices.

    * Increased size of staging buffers.
    * Enhanced Binary Image Format (BIF).
    * Support for UVD video hardware component through OpenCL (Windows 7).
    * Support for Northern Islands family of devices.
    * Support for AMD Radeon™ HD 6310 and AMD Radeon™ 6250 devices.
    * Support for OpenCL math libraries: FFT and BLAS-3, available for download at AMD Accelerated Parallel Processing Math Libraries.
    * Preview feature: An optimization pragma for unrolling loops.
    * Preview feature: Support for CPU/X86 image. This enables the support for Image formats, as described in the Khronos specification for OpenCL, to be run on the x86 CPU. It is enabled by the following environment variable in your application: CPU_IMAGE_SUPPORT.
    * ATI Stream Profiler 2.1 enhancements:
          o Timeline visualization and API Trace.
          o Support for 64-bit profiling.
          o Support for Linux (command line version).
          o Support for profiling DirectCompute applications.

    * Stream KernelAnalyzer (SKA) 1.7 enhancements:
          o Support for Northern Islands devices.
          o Support for Catalyst 10.9 to 10.12.

    * New samples:
          o MonteCarloAsianDP
          o DeviceFission
          o FluidSimulation2D

    * Optimized Histogram sample.
    * Added makefiles and implemented changes in samples code to compile with Intel compiler, MinGW + GCC, and MinGWx64 + GCC.
    * Various OpenCL™ compiler and runtime fixes and enhancements (see developer release notes for more details).
    * Various samples fixes and enhancements (see samples release notes for more details).

Supported devices to be found in "Getting started guide"
Either the developers forgot the FirePro V9800 or  their boss had none left over for them to play with  ;D

                                                                   Double-       Global     Compute Kernel
                     Product Line                Card              Precision      Buffer         (Shader)

                     AMD Radeon™ HD               6970                Yes           Yes             Yes
                                                  6950                Yes           Yes             Yes
                                                  6870                 No           Yes             Yes
                                                  6850                 No           Yes             Yes