Multi-threaded analysis now uses multicore models for real-world emulation

Technology News |
By eeNews Europe

Unlike vfAnalyst, the new tool uses models of the actual target processor, to give a better representation of the performance boost the user can expect from parallelising their code.

Today vfThreaded supports the multicore variants of Intel’s i5 and Atom processors and the official release of vfThreaded comes a few weeks after Vector Fabrics became an Intel partner.

According to Marco Jacobs, marketing director for Vector Fabrics, vfThreaded targets a different market to vfEmbedded by focusing on the Windows-based applications running on an Intel platform, explaining that parallel programming can return as much as 16x the performance and 2x lower power.

The tool works by first analysing the code to find ‘hot spots’ and dependencies, then allows the user to partition using the visual interface, before implementing the changes made at the code level and then running the code on the emulated processor models. Once a satisfactory performance point has been found, the tool then delivers a ‘recipe’ showing what changes the software engineers should implement.

As with the company’s other offerings, vfThreaded runs under a ‘Software as a Service’ model in an Amazon-hosted cloud-based server, on a monthly subscription. A free trial is also available.


Linked Articles
eeNews Power