LONDON – Samplify Systems Inc. (Santa Clara, Calif.), a startup fabless chip company with expertise in compression, has announced it can provide application acceleration technology called APAX that can more than double the performance of multicore processors.
It would appear the technology is based on data compression applied to large data sets that are often part of the challenge thrown at multicore processors and processor arrays. Initially the technology has been applied to x86-compatible CPUs and GPUs from Nvidia Corp. and Advanced Micro Devices Inc.
The sustained performance of multicore processors reaches only a fraction of their potential because of memory and I/O bandwidth constraints, said Al Wegener, founder and CTO of Samplify Systems in a statement. "For the first time, we are offering a technology that can reduce the bandwidth requirements in a multitude of memory-bound numerical, graphical, and imaging applications," he said.
APAX works by speeding up the distribution of integer and floating-point data and is also applicable to graphics and images. Samplify said on its website that the use of APAX can produce mathematically identical results to non-APAX systems while providing between 35 percent improvement to a more than doubling in performance. Alternatively, the technology should allow the clock frequency to be reduced to produce adequate performance at reduced power consumption.
The technology works for both supercomputing and data center applications as well as for processors inside smartphones and mobile devices, Samplify said. The company said it plans to deliver APAX as both software for high-performance computing applications and as circuit intellectual property for inclusion in SoCs and FPGA families with embedded processors.
APAX is available as a standalone executable for file-based I/O and also as a DLL (Windows) or a shared object (Linux) which can be compiled into an HPC application.
APAX can also accelerate memory subsystems, storage subsystems, I/O, and network interfaces by integrating it as hardware IP into the memory controller, storage controller, Southbridge, and network interface card, respectively.
The integration of multicore processors, along with GPUs and video codecs, into application processors creates similar memory and I/O bottlenecks to those experienced in HPC applications, Samplify asserted. APAX, when integrated as a hardware IP block within an application processor, can accelerate application performance on multicore processors while simultaneously improving graphics performance and video playback, Samplify said.
APAX is available as software for Intel x86/x64 CPUs on both Linux and Windows 7 for HPC and cloud computing applications. Samplify's APAX technology will be available as Verilog RTL for SoCs and FPGAs in Q3 2012, the company said.
Multicore graphics firm LucidLogix Technologies Ltd. (Kfar Netter, Israel) has announced the
existence of XLR8, which is virtualization software designed to improve
the performance of graphics processing units (GPUs). LucidLogix claims its GPU virtualization software can improve the graphical
response performance, visual quality and battery efficiency of mobile
devices including smartphones, tablet and notebook computers by up to
Samplify, founded in 2006, entered the Silicon 60, EE Times' list of
emerging startup companies at version 9.0 in October 2009. The latest
edition of the Silicon 60 is version 12.5, which is the subject of a
detailed technology and employment digital edition which can be accessed
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.