CULA 1.3 Released

The CULA developers at EM Photonics are pleased to announced the release of CULA 1.3.

CULA is a GPU-accelerated linear algebra library that utilizes the NVIDIA CUDA parallel computing architecture to dramatically improve the computation speed of sophisticated mathematics. CULA is designed from the ground up to be simple to use and can easily integrated into existing code bases. Available as pre-compiled libraries for both 64- and 32-bit versions of Windows, Linux, and Mac OS X, CULA brings GPU accelerated LAPACK to wide variety of systems.

CULA is based on the familiar LAPACK interface and features accelerated routines from the families of:

  • System solvers
  • Least squares solvers
  • Orthogonal factorizations
  • Eigenvalue and Eigenvector solvers
  • Singular value decompositions

Version 1.3 brings speedups to all of our linear solvers including:

  • General system solve
  • Positive definite system solve
  • Least squares system solve

Our 1.3 release is our final release before CULA 2.0 which will introduce:

  • CUDA 3.0 support
  • Initial Fermi optimizations
  • New routines

Please visit our webpage to learn more about CULA’s performance, our full function list, read our developer’s blog, or ask a question on our forums.

Just a quick update to note that CULA 1.3a is available with some small bug fixes related to symmetric eigenvalues.

We have also posted our first Fermi (Tesla C2050) benchmarks on our developers blogs.

And finally, a Fermi compatible CULA 2.0 should be available shortly.