I'd like to use the llvm-mca tool to view execution timelines for instruction sequences in ARM Cortex-M cores, e.g. the Cortex-M4. I can see that there is support for in-order cores, but I'm not sure if these cores are supported by llvm-mca. Concretely, I'm looking for the specific command-line options that I need to pass in order to analyze code for this architecture. Here's what I tried in Compiler Explorer, but I'm getting an error and I'm out of ideas on how to proceed.
Asked
Active
Viewed 92 times
1
swineone
- 2,296
- 1
- 18
- 32
-
since on all processors execution time is primarily determined by the fetch/memory system and not the core itself, how is this modeled in this tool? At least with cortex-m's it is chip specific and not as dynamic as typical cortex-a applications, but still how would it be possible to make a generic tool (that is of any value)? – old_timer Mar 11 '23 at 19:55
-
what results are you expecting from such a tool, if it existed? (if it were possible) – old_timer Mar 11 '23 at 19:55