Add Coremark.

[yvantor]

This PR integrates the Coremark benchmark into Cheshire.
Task list:

• Build Coremark together with the other Cheshire tests.
• Add Coremark execution as a post FPGA deployment CI test on the Genesys2.
• Update the docs to modifify CHS_XILINX_FLASH_BIN into CHS_XILINX_FLASH_IMG (I have updated the source files but don't know how to properly change the doc website).
• Migrate Cheshire-specific files to avoid Coremark fork.

Discussion points:

• Is there a specific reason for using O2 compiler options instead of O3? This is just for curiosity since the performance increase on Coremark is negligible (0.01 Coremark/MHz).
• The current integrated Coremark is a personal fork. I don't know if it might make sense to fork it into pulp-platform to support other pulp targets other than Cheshire in the future.

[paulsc96]

This is a draft, but also ready for review. Is there a mistake here? If not, how should I interpret this?

[yvantor]

This is a draft, but also ready for review. Is there a mistake here? If not, how should I interpret this?

I left it as draft because I was considering adding a CI step that launches coremark on an FPGA as a proof that the test is successful and returns a consistent result. Do you think it would make sense?

[yvantor]

Dependent on the nonfree related PR

[yvantor]

@paulsc96 I think this one is ready for being reviewed.

[paulsc96]

It looks fine overall, but I will have to think the details through.

One important question is: is it feasible to avoid a fork of Coremark? If so, we should. If not, we may aswell move some of the coremark-specific build stuff there.

[yvantor]

It looks fine overall, but I will have to think the details through.

One important question is: is it feasible to avoid a fork of Coremark? If so, we should. If not, we may aswell move some of the coremark-specific build stuff there.

I think avoiding the fork whould be feasible if we move some stuff here, in the end they are not many files... I can spend some time on this later this week

[yvantor]

It looks fine overall, but I will have to think the details through.
One important question is: is it feasible to avoid a fork of Coremark? If so, we should. If not, we may aswell move some of the coremark-specific build stuff there.

I think avoiding the fork whould be feasible if we move some stuff here, in the end they are not many files... I can spend some time on this later this week

@paulsc96 this is complete now.