ChipEnet.com - Ethernet Attached FPGA Acceleration

Gigabit, 100Mb  Ethernet MAC  IP cores for Xilinx FPGAs

<------- Gigabit Ethernet MAC core optimized for Streaming Data

ChipEnet Ethernet MAC IP cores enables easy control and interface of FPGA
based accelerators. FPGAs offer quick implementation and high
performance execution of special data/signal processing tasks such as
video data compression/decompresion, data encryption/decryption, or
molecular dynamics simulation. While current CPUs do run at multi-GHz
rates, even the newest CPUs can only handle a small number of threads of
execution in parallel. In contrast, a FPGA can handle hundreds or
thousands of threads in parallel, albeit at a lower clock rate. Without the
hefty NRE charges and long lead times of ASICs, FPGAs allow quick
implementation ( and easy incremental improvement ) of special algorithms
that cannot be or are difficult to execute with standard desktops or servers.

But being able to execute the special algorithms quickly is not enough.
Where does the data come form and where does the processed data go ?
How is the FPGA accelerator controlled and initialized ? These types of  IO
and management functions are best handled by standard desktops or
servers. A desktop or server running Unix, Linux, Windows, or Mac OS
offers a far more powerful management environment than an embedded
processor running a small embedded OS can ever achieve.

The next question is then, How do you connect a FPGA accelerator to a
standard desktop or server ?  This is where Ethernet comes in. Using
Ethernet MAC IP cores to interface to FPGA accelerators offers many
advantages. Simple programming, no drivers necessary, and the option to
share the FPGA accelerator with other desktop or server on the network
are key advantages. Read more here. The other option of using PCI or PCI
Express bus to connect to a desktop or server requires far more effort,
driver needed for every OS and continuous updates for every OS changes,
and more implementation constraints due to PCI board size and power
limitations.

Need more bandwidth ?  Take a look at the 4 Gigabit ChipEnet Virtual
Ethernet MAC core. read more

Check out the Linux packet processing max limits experiment using the
ChipEnet SM1000 Ethernet MAC core. read more

New Ethernet Switch Fabric IP core with Multicast,
Isochronous support  coming !

This is a very low latency, sub 100 ns, low gate count, scalable switch
fabric IP core with support for Multicast and Isochronous packet streams.
Link speeds up to 4Gb/s.

IP Cores for
FPGA