User
Applications
|
High
Performance Computing
 High
Performance
Computation demands
highly parallel resources, and our PCI FPGA cards deliver. The
pairing of PowerPC processors with all the other resources of the
FPGA create an ideal platform for creating heterogeneous multi-core
solutions for your most difficult HPC problems. Multi-Gigabit serial
I/Os connected directly to the FPGA allow high bandwidth, low-latency
pipelining, clustering, and I/O. Our easy to use design environment
makes all this power easy to harness.
The
tremendous
internal bandwidth of
the FPGA combined with its distributed resources (including RAMs and
MAC units) allow for the creation of high efficient computational
units
that are busy all the time working on your solution. Each of the
cores that you create for your solution can be highly customized for
their task, leaving little to waste. Custom processing pipelines
allow you to efficiently manage data dependency issues that can
stall the performance of a normal processor.
The
ability to
directly connect cores
via the FPGA routing resources pushes synchronization time between
cores down to the nanosecond range. Data can flow directly between
cores (reducing the demand on shared memories) or can be buffered by
shared memories to smooth out the data flow. Your cores may be as
fine-grain or coarse-grain as your solution requires. FPGAs enable
you to design out the bottlenecks that limit the performance on
standard processors.
The
four
external multi-gigabit serial
SFP slots and four internal multi-gigabit serial connections allow
high bandwidth, low-latency communications between multiple cards
and I/O devices.
|
