Matrox Supersight0 pages

Matrox Supersight e2
High-performance computing (HPC) platform for
computationally-demanding industrial imaging.
Benefits
Tackle extreme applications with
certainty using a high-performance
compute cluster platform
Harness the full power of today’s
multi-core CPU, GPU, and FPGA
technology to provide image
processing offload and acceleration
Eliminate I/O bottlenecks with a
unique PCIe® 2.0 switched fabric
backplane architecture
Directly interface to external process
equipment through the integrated
Gigabit Ethernet, RS-232/485 and USB
2.0 connectivity
Minimize the need for revalidation by
utilizing a lifecycle managed platform
with consistent long term availability
Simplify system integration by using
an integrated platform from a single
vendor and pre-qualified third-party
components
Solve applications rather than
develop underlying tools by
leveraging standard Microsoft®
development tools, Matrox Imaging
Library (MIL)
HPC for imaging
Matrox Supersight brings together the latest technologies from multi-core CPUs
to GPUs to FPGAs in a single pre-validated high-performance computing (HPC)
platform, allowing OEMs to focus on developing applications with cutting-edge
performance instead of integrating components. Matrox Supersight is fully supported
by Matrox Imaging Library (MIL), an established collection of software tools for
developing industrial imaging applications, in order to extract the maximum
performance from the platform and deliver a complete solution in a timely manner.
Backed by a carefully managed lifecycle and long term availability, the Matrox
Supersight is a solid foundation for your next computationally-demanding application.
Processing trinity
Achieving maximum performance requires using the right technology in the right place
within the application. FPGAs show great strength with image pre-processing tasks,
general purpose CPUs are ideal for image processing and analysis, and GPUs provide
substantial acceleration for image processing primitives. Matrox Supersight enables
developers to put together all of these technologies into a complete HPC platform with
enough compute density to surpass blade server offerings.
Consistent long-term availability
Carefully selected components are coupled with strict change control to ensure
consistent long-term supply. Longevity of a stable supply lets OEMs maximize return
on the original investment without incurring the additional costs associated with
repeated validation of constantly-changing mainstream commercial platforms.
Distinctive switched fabric
A unique PCIe 2.0 backplane provides the switched fabric to group processing
elements (CPUs, GPUs and FPGAs) into compute clusters. Utilizing the point-to-point
full-duplex nature of PCIe , bus traffic from compute clusters is isolated ensuring
optimal performance with increasing accelerator count. Unlike classical segmented
backplanes, the Matrox Supersight backplane, through the Distributed
MIL (DMIL) communication API, enables segments/clusters to look
like a unified system.
®
®