Report tile utilization with DVFS info

[tancheng]

No description provided.

[MeowMJ]

It seems that the folder /test/kernels is only useful in ICEDdeom.py. You already have CGRA-Benchmark in your CGRA-Flow and the 'fileSource' of ICEDdemo.py can be replaced with any folder path. Do we need to keep the folder /test/kernels?

[tancheng]

I tried to think about a cleaner way. However, the script hard-coded the path and has the pre-assumption about the file names. In addition, it is also reasonable to keep a set of test.cpp within the test folder, so we won't heavily depend on the cgra-benchmark's content (in case it is updated by itself).

In the future, our verification of the cgra-mapper can leverage the test/kernels.

[tancheng]

Hi @HobbitQia, can u plz help take a look at the failed test? Does it look like this time the offset of 2 is unnecessary?

[HobbitQia]

Hi @HobbitQia, can u plz help take a look at the failed test? Does it look like this time the offset of 2 is unnecessary?

u r right. I checked the opcode count in line 650-660 and I thought their original opcodes are correct. So this time we should set this paramter to be 0.🤔

[HobbitQia]

I saw u change the directory names from nonlinear_test, idiv_test to nonlinear, idiv. However, the working directories we set in cmake.yml are still ${{github.workspace}}/test/nonlinear_test, ${{github.workspace}}/test/idiv_test, respectively. I think it's the reason that the github testing failed.

[tancheng]

Right, I recognized that but forgot to push my fix. Thanks!