How to reduce IC delay? Useful resources

Hello,

I have some basic questions:

My design includes a NIOS and an accelerator that is designed by myself. I designed a custom instruction to connect the accelerator and NIOS together. The custom instruction is connected to the accelerator by an Avalon Conduit interface.

The problem is that the accelerator is too large and the conduit's signals need to go to all parts of this accelerator. As a result, Fmax is too low for my design, and I can see that most of the delay is related to interconnection delay.

My question is that how I can reduce this delay? Is putting FIFO between the custom instruction and the accelerator a good idea? In that case, is there any way to add this FIFO in the Platform Designer?

I have attached the delay for the most critical path, and also the routing utilization in Chip Planner including the most critical path.

Also, I would be thankful if I can get some useful documents about Timing Closure, and how to reduce the delay in Quartus.

Thanks