search

bespoke-silicon-group / bsg_manycore

Tile based architecture designed for computing efficiency, scalability and generality
Other
227 stars 58 forks source link

Add comments #107

Open taylor-bsg opened 5 years ago

taylor-bsg commented 5 years ago

https://github.com/bespoke-silicon-group/bsg_manycore/blob/5f5c3b975396e6e87e3c00688d9facb536ea07ef/testbenches/common/v/vanilla_core_profiler.v#L120

Add comments to clarify these questions from Chris:

Great! It might be good to have a more detailed dynamic instruction count breakdown? So basically how many of each type of instruction. The sum of these instruction counts should add up to be the total number of committed instructions?

Does num_ld include remote loads? Or is num_ld only for scratch pad accesses?

Maybe UW can provide an explanation of each stat?

This will be super useful!

Best Chris

Sent from my iPhone

taylor-bsg commented 5 years ago

It would also be useful to have comments clarifying the nature of stalls as well do they overlap, and what do they sum to?

drichmond commented 4 years ago

Did we reach a resolution on this? I see more comments in the file, but Chris and a student we're asking last night about what stall_md meant.

Do we need more documentation about the profiler in general?