AccelerEyes has released Jacket 1.1.1, its GPU engine for MATLAB. This release includes random-number generation with grand(), and double-precision support for FFT functions. All Jacket trials have been reset, so if you've trialed Jacket in the past, you now have another chance to trial again. This version of Jacket includes the following enhancements:
- fft is now supported for gdouble inputs.
- grandn - GPU version of randn.
- grand - GPU version of rand.
- sort(X) is now supported.
- cumsum(X) is now supported for vectors.
- cumprod(X) is now supported for vectors.
- permute, ipermute is now supported.
- repmat is now supported for complex inputs.
- The gcache command now allows users to manipulate the cache by flushing, saving or loading it.
- The command 'ghelp' displays information about functions supported by Jacket on the MATLAB console.
Jacket 1.1.1 requires CUDA 2.3. The CUDA drivers and toolkit are available for download. For more on CUDA, see the article series CUDA, Supercomputing for the Masses.
According to AccelerEyes, Jacket 1.2 will be released shortly and include support for running Jacket across multiple GPUs. This will work in conjunction with MATLAB's Parallel Computing Toolbox (PCT). For every GPU in the system (including both workstations and clusters), a PCT Worker will be assigned which will have the ability to offload computations from the CPU to the GPU. This will enable maximal performance under any given hardware configuration.
