Network and storage acceleration

Javier
Javier
Last updated 
Back to table of contents

As we can read in the section DOME MicroDataCenter, the cloudFPGA research team lead by François Abel took benefit of the IBM in-house DOME MicroDataCenter project to replace CPU-Modules with FPGA-Modules to go even further on the computing capabilities of the platform.

As we can see here:
IBM and ASTRON Unveil 64-bit Microserver Prototype
 
the DOME platform admits different types of nodes that fit the same form-factor:
where, in the specific case of the cloudFPGA platform, as presented till now, it is defined that all the modules are FPGA-Modules (see above, right).

However, there is nothing that stops us to have different setups where FPGA-Modules could me mixed with CPU- or storage-modules to cover a wider range of use-cases. A clear example would be to replace some FPGAs with storage nodes to achieve FPGA-based storage-acceleration. On the other hand, it has been proven that the cloudFPGA platform, with the current version of the FPGAs are very well supporting networking tasks as shown in the Networking performance in the Benchmarks section.

Putting all this together, computing, network and storage acceleration are just there within cloudFPGA, being the perfect platform supporting the vision Towards Distributed Adaptive Computing in the FPGAs in the datacenter section.