There are a few possibilities and none will be as good as some of our other products with better memory interfaces. There is also a question of how much memory is actually required. I have seen a wide range of numbers from people in this forum so that is something to established.

Possibility1

CM1 contains logic only and works with a host processor that has software and memory. Bandwidth of USB will be a major limit.

Possibility2

1,2, or 3 FPGAs on CM1 are used as memory. We get about 700KB per FPGA when used as memory.

Possibility3

The up/down interface is used to support a custom memory board. There are 18 wires available to make this connection which are never going to compete with a direct DDR3 interface but the add-on board could be a very large memory size. 100GB would be possible but maybe not cheap.

Outside of these possibilites there are further hybrids of these ideas e.g. use one FPGA as a cache with larger memory as add on board.

Bear in mind the original Litecoin point was made in a different tread in the context of people thinking ASIC would kill CM1 Bitcoin mining profitability and this was something else to do with these boards it is never going to be the best solution possible. Almost any FPGA board will have Litecoin capability but not necessarily huge performance.

If you want Litecoin miner we probably already have better boards already in the product range for this purpose although prices may scare you as they never priced/designed for this market. They just happen to be possible better fits to the requirement. So far all of this conjecture. Most people have looked at what software does to do this processing and there may be entirely better ways in a FPGA than just converting a software function into FPGA hardware functions. Such radical thinking actually needs to be done by someone who knows FPGAs very well and probably software pretty well too so it is not going to be a simple or quick task to do this sort of analysis/design quickly. I am not going to promise any timelines but it is one task we are hoping to get some time on. However our design team are very loaded in the 2-3 month span and probably no way that will happen here in the short term.