Customers running Bright Cluster Manager can install the new, preconfigured CUDA package directly from Bright Computing's Yum repository using a simple command. Bright Computing's sophisticated provisioning system will then take care of distributing the relevant elements to the GPU nodes.
This latest CUDA package can co-exist with CUDA packages previously released by Bright Computing, and users can easily switch between CUDA versions, due to Bright Cluster Manager's preconfigured Environment Modules. Furthermore, the CUDA driver is automatically compiled at boot time against the running Linux kernel, ensuring compatibility with the kernel at all times.
Bright Computing's preconfigured CUDA 3.0 package includes the following:
- CUDA 3.0 Environment Module — allows easy switching between CUDA 3.0 and previous CUDA versions;
- Support for new GPUs based on Fermi architecture — including ECC, optimized double precision, support for linear algebra libraries such as BLAS and LAPACK, the CUDA-GDB debugger and Visual Profiler;
- C++ support — delivering improved productivity with class and template inheritance;
- GPGPU/Graphics interoperability — delivering OpenGL for both CUDA and OpenCL;
- Improved developer tools for Linux — including the new CUDA Memory Checker that reports misalignment and out-of-bounds errors.