AccelerEyes has released Jacket v1.1, the latest version of its GPU engine for Matlab, which allows standard Matlab code to run on Nvidia GPUs.
Highlights of this new release include support for double-precision arithmetic, which enables a higher-level of accuracy for applications requiring fine precision. There is also expanded type support, including support for logical, int32, uint32, etc, and a new Developer SDK enables integration of custom Cuda kernels into the Jacket runtime. This allows custom code to inherit Jacket’s optimisations for memory transfers, kernel executions, and system performance within Matlab.
The new release also offers expanded support for filtering functions, such as conv2, convn, filter2, etc, to include kernel sizes up to 10x10. A new licensing scheme to support Concurrent Network licenses has been introduced, along with expanded support for reductions functions, such as sum, min, max, any, all, find.