Advantage of having checksum on each ask and bid
Each bid and ask should be signed by the client software of node B. So when node B is sending a bid or an ask to node A, node B will have to refer to a previously established contract, and the software will also have to sign the bid or ask. If a bid or ask is cancelled, a cancel message is created and signed by private key.
Node A specifies in the contract with B the maximum number of messages or RPC that B is allowed to send per second.
If node A receives a flood of messages, but these have the wrong checksums, they are simply thrown away. Perhaps it would be possible to have the GPU do the checking, since this work seems to be of the type that GPUs are good at. This would allow even a modest computer with a GPU to do significant amount of checksum or message-checking.