Closed Tobi29 closed 2 years ago
Hi. I'm extremely interested in getting a Cortex-A72 model for OSACA. Is there anything I can do to help this pull request get merged?
Merging #60 (ab61554) into master (6e99954) will not change coverage. The diff coverage is
n/a
.
@@ Coverage Diff @@
## master #60 +/- ##
=======================================
Coverage 90.62% 90.62%
=======================================
Files 13 13
Lines 2401 2401
=======================================
Hits 2176 2176
Misses 225 225
Impacted Files | Coverage Δ | |
---|---|---|
osaca/osaca.py | 84.45% <ø> (ø) |
|
osaca/semantics/hw_model.py | 79.22% <ø> (ø) |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 6e99954...ab61554. Read the comment docs.
This adds a model for the ARM Cortex-A72 found in the Raspberry Pi 4. Additionally, the required files for regenerating the model are included as well, in particular:
osaca/data/a72/mapping_pmevo.json
: PMEvo mapping, used forport_pressure
entriesosaca/data/a72/template.yml
: Template containing basic information and a few instructions that cannot be handled automatically or yield bad resultsosaca/data/pmevo_importer.py
: Script to convert PMEvo mapping together with template into final modelIn case changes were done the model can be generated with:
However this will not give correct latencies due to asmbench not being used. To do that also pass
--asmbench
to the script, that will adjust the latency (and throughput) in the output model. Obviously that requires actually running on a RasPi and the conversion takes a lot longer (a few minutes usually). The includedosaca/data/a72.yml
was created with--asmbench
, so these values are adjusted already.