I'm creating this issue just to get a discussion going about the odd 'nop' timing in the fpgperf.py run. I suggest removing it in #203.

From what I've seen, the 'nop' timing is the same for each toolchain, simply running of the true command in bash, and reporting something around 0.003 or 0.004 seconds. This is the code in toolchain.py lines 53-54 which every toolchain inherits from:

with Timed(self, 'nop'):
            subprocess.check_call("true", shell=True, cwd=self.out_dir)

I wondered if this was an attempt to make sure that different builds being compared are running with the same processing speed, but this seems to me like it would be an ineffective way to do that. My other thought is that it may be a placeholder for something that hasn't yet been implemented but in that case, we could remove it until it is functional.

Can we get any clarification about the purpose of 'nop' in the reported runtimes? If not, can it be removed with the merge of #203?