Gpu Systems accelerates
computations using modern accelerators such as CPUs, GPUs and future devices while simplifiying implementations.
Today most computers, handhelds and mobile phones include one or several processing devices - CPUs or GPUs.
Creating Software Across Platforms and Devices
Our focus lies in the development of next generation APIs and software for modern processor architectures, such as CPUs and GPUs. GPU technology today delivers unparalleled performance but is difficult to harness. Moreover, a GPU and CPU architecture is very different from each other whereas computations are to be expressed uniformly across both.
Working with our product Libra SDK, results in more speed to your program at the same time targetting both the CPU and the GPU. Your program also needs writing less code compared to other existing solutions. Our customers will benefit greatly from this new cross-platform, cross-processor opportunity.
Custom library development with Libra SDK:
gVar add(gVar array1, gVar array2)
return array1 + array2;
A portable function to access the power of CUDA, OpenCL, OpenGL (GPU) and x86/x64 (CPU) devices.
Libra introductory app source code - Hello World
Libra with CUDA devices producing less source code than native CUDA
Libra with OpenCL devices producing less source code than native OpenCL
See more code Examples
See FAQ for more info
Libra SDK - GPGPU/Heterogeneous software development kit - available for major operating systems and compute devices.
Introduction to generic CPU/GPU computing in a C/C++ environment.
Libra SDK matlab support released 27/10/2010
Bringing CPU & GPU computing to matlab environment.
Libra 1.2 SDK Showcase 01/06/2010 AMD booth - ISC '10 - Germany
Enabling accelerated computing through the combination of GPU and CPU technologies.
Libra SDK 1.2 version - 08/06/2010
Simplify and accelerate your computations using Libra 1.2 SDK. Libra SDK enables C/C++ programming for both CPUs and GPUs including sample programs and documentation. Technology overview
Libra accelerates and simplifies
3D technology - 05/09/2008
A disparity/depth estimation algorithm was easily implemented with Libra 1.0 SDK. The performance of the previous c++/openMP implementation was significally improved by using Libra 1.0 SDK.