Closed WilliamWangPeng closed 3 years ago
It's been way too long since I've looked at this myself (over 10 years...), but I think it should be easy to deduce from the source code what those numbers mean exactly, see https://github.com/boegel/MICA/blob/master/mica_reg.cpp#L319 (I think)
Hi @WilliamWangPeng,
As @boegel mentioned, you can refer to the code to understand the output values. Feel free to have a look at the README.md file we provided in the repo. Also, If you use analysis_type: all
, it will emit almost ~100 metrics and we have provided a tablegen utility to generate a table of CSV for those. Have a look at line 10 to see the names:
https://github.com/boegel/MICA/blob/master/tableGen.sh#L10
Hi @WilliamWangPeng,
As @boegel mentioned, you can refer to the code to understand the output values. Feel free to have a look at the README.md file we provided in the repo. Also, If you use
analysis_type: all
, it will emit almost ~100 metrics and we have provided a tablegen utility to generate a table of CSV for those. Have a look at line 10 to see the names:https://github.com/boegel/MICA/blob/master/tableGen.sh#L10
- Amir
thank you so much dear @amirjamez
Another question is where to put the PID number in the tableGen.sh, so that I can use tableGen.sh to generate the a CSV file
for example, my PID number here is 19060
thank you
best regards
William
Line 24 of the script finds the unique PID of all analysis types so it concats the features of the same analysis into one csv.
Line 24 of the script finds the unique PID of all analysis types so it concats the features of the same analysis into one csv.
yes, I have noticed the line 24, I tried to replace the sed to 19060, but failed, and later I could just see the headers in micaTable.txt which has no data.
open another issue
@WilliamWangPeng pidList
returns a list of PIDs MICA generated. Every time you run MICA, the output files are named with a new PID. It looks for all output files having the same
PID, and later it iterates over them to generate the CSV file.
Hi dear author,
I have read the MICA article which says 2 values and 7 probabilities in register traffic as follwing shows:
and I have tested a set of data abot register traffic here:
I wonder which number is average number of register input operands per instruction, which is number of register reads per register write.
thank you
Best Regards to you
William