Hi Jan No, it's not based on the Xilinx core. It's a custom implementation, with the same interface as the existing CASPER 10G core (so should be a drop-in replacement for anyone wanting higher speeds in their designs).
We are still trying to navigate the murky legal waters to see if we can open-source it as part of the CASPER library for ROACH3. I'm hopeful that we'll be able to release it, but the bean-counters don't like the idea. Negotiations are ongoing... watch this space! Jason Manley CBF Manager SKA-SA Cell: +27 82 662 7726 Work: +27 21 506 7300 On 03 Aug 2015, at 10:10, Jan Wagner <[email protected]> wrote: > Hi Jason, > > thanks for your suggestions some time ago! I thought to come back to your > mention of 40G QSFP+ working on ROACH-3: do you know if this is based on > Xilinx' HSEC core (40G/100G), or if there will be a 'fourtygbe' Casper pcore? > > cheers, > Jan > > >Hi Jan > > > >At this time, there are no plans for a 40G card for ROACH2. > > >You could use four 10G streams into a switch and then from the switch into a > >40G card (we do this now), or you could try something like a quad-port 10G > >card: > http://www.chelsio.com/nic/t5-unified-wire-adapters/t540-cr/ > for direct > >connection to a ROACH-2. > > > >ROACH-3 prototypes have demonstrated working 40G QSFP+, but this will only > >be > >available on ROACH-3, not ROACH-2. > > > > >Jason Manley > >CBF Manager > >SKA-SA >
