Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
8 years ago

Why does this design not fit and how to reason about congestion

Hi all, I'm trying to optimize a simple program which computes a correlation matrix. To promote reuse/parallelism the design proceeds in blocks. However, I do not get the design to route for la...
Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
8 years ago

Thanks for the input. I've attached the quartus_sh_compile.sh for the second kernel below. I'm afraid I've deleted it for the first kernel, but I will rerun this tonight, and send it later.

I'm using an Arria 10 with Quartus 16.0 (no 16.1 yet for Nalla board).

What exactly is the definition of congestion here? How do I prevent it?

I'll also try a NDRange kernel, the problem with that is that to devide the computation some extra computations involving squareroots are needed.

Thanks in advance for the help!