Thursday, August 12th 2010

AMD Spurs Software Development to Benefit from Heterogeneous Compute Architectures
AMD today announced the availability of the ATI Stream Software Development Kit (SDK) v2.2 with full OpenCL 1.1 support to provide developers with the tools they need to build incredible next-generation applications. By taking advantage of both CPU and GPU processing power in a given system, applications like 3D video, HD video chat and multi-display 3D gaming are possible.
A development platform created by AMD, the ATI Stream SDK v2.2 brings a wide range of tools to the developer community including support for OpenCL 1.1, in addition to Ubuntu 10.04 and Red Hat Enterprise Linux 5.5. This updated development platform empowers the developer community to accelerate applications on the CPU and GPU quickly and efficiently. The new version is available for download here.
"Availability of the ATI Stream SDK v2.2 with OpenCL 1.1 support is a great example of how CPU and GPU technology continues to mature and usher in next-generation computing experiences, where voice, touch, gesture and facial recognition capabilities are common, everyday features," said Patricia Harrell, director of Stream Computing, AMD. "The enhancements in the ATI Stream SDK v2.2 are especially important due to the support for OpenCL 1.1, which is integral to the forthcoming AMD Fusion family of APUs. These tools allow the developer community to take advantage of heterogeneous computing architectures both today and tomorrow."
"As the premier provider of high-quality audio and video codec solutions, MainConcept has a shared vision with AMD to support open standards such as OpenCL," said Muzaffer Beygirci, managing director of MainConcept GmbH. "Our upcoming OpenCL H.264/AVC Encoder was built on ATI Stream technology and the OpenCL standard, and we believe ATI Stream-enabled GPU acceleration will be a compelling new feature in our customers' solutions."
ATI Stream SDK v2.2 delivers expanded support for operating systems, compilers, hardware and the latest industry standards, further streamlining application development and acceleration. New features include:
A development platform created by AMD, the ATI Stream SDK v2.2 brings a wide range of tools to the developer community including support for OpenCL 1.1, in addition to Ubuntu 10.04 and Red Hat Enterprise Linux 5.5. This updated development platform empowers the developer community to accelerate applications on the CPU and GPU quickly and efficiently. The new version is available for download here.
"Availability of the ATI Stream SDK v2.2 with OpenCL 1.1 support is a great example of how CPU and GPU technology continues to mature and usher in next-generation computing experiences, where voice, touch, gesture and facial recognition capabilities are common, everyday features," said Patricia Harrell, director of Stream Computing, AMD. "The enhancements in the ATI Stream SDK v2.2 are especially important due to the support for OpenCL 1.1, which is integral to the forthcoming AMD Fusion family of APUs. These tools allow the developer community to take advantage of heterogeneous computing architectures both today and tomorrow."
"As the premier provider of high-quality audio and video codec solutions, MainConcept has a shared vision with AMD to support open standards such as OpenCL," said Muzaffer Beygirci, managing director of MainConcept GmbH. "Our upcoming OpenCL H.264/AVC Encoder was built on ATI Stream technology and the OpenCL standard, and we believe ATI Stream-enabled GPU acceleration will be a compelling new feature in our customers' solutions."
ATI Stream SDK v2.2 delivers expanded support for operating systems, compilers, hardware and the latest industry standards, further streamlining application development and acceleration. New features include:
- Support for the OpenCL 1.1 specification, ratified in June 2010, providing developers with the ability to take full advantage of the new features OpenCL 1.1 brings
- Extended operating system support, including Ubuntu 10.04, Red Hat Enterprise Linux 5.5
- Increased hardware support including x86 CPUs with SSE 2 or later, allowing users to run OpenCL applications on a greater variety of CPUs
- Additional support for Linux and Windows compilers, including GNU Compiler Collection 4.1 or later, Microsoft Visual Studio (MSVS) 2010 Professional Edition and Minimalist GNU (GCC 4.4) for Windows (MinGW)
- Single-channel format support for OpenCL images, providing developers with greater flexibility to access the caches on the GPU
- Support for OpenCL and DirectX 10 interoperability increasing overall efficiency when displaying computed results to the user
- Support for additional double-precision floating point routines in OpenCL kernels to enable greater computational accuracy in HPC and scientific based applications.
12 Comments on AMD Spurs Software Development to Benefit from Heterogeneous Compute Architectures
(oh and you missed the S on the second last 'support')
It works fine when compiling using indirect OpenCL (through CUDA) and CUDA.
What's preventing wide usage of Stream is lack of direct support from AMD developers and that they're going for the OpenCL route, which is good, but again, lack of proper documentation and dev help is keeping them down.
Now, CUDA has the same "switch to OpenCL" goal, but there are dedicated NVIDIA devs who are willing to help port/convert/rewrite applications with enthusiast and budding programmers. This is the only reason why I have an 8600GT along with my Crossfire setup.
OpenCL is what would make Stream and CUDA a neutral ground. At the moment, CUDA's doing well in conversion. A situation very similar to what happened with OpenGL in the past (e.g. NVIDIA hardware extensions).
Thanks for clarification. I didn't really have any solid details on that. So it was an NVIDIA card that was used for PhysX.
I'm not really into PhysX, but it may seem that NVIDIA does truly use CUDA for PhysX if that's the case. I'll see if the PhysX code that game developers use access the video card the same way CUDA/OpenCL does.