In order to get the total runtime down below Github's limit of 6 hours, I used --debug-single-run which runs each benchmark exactly once. This is likely not enough time for specialization or the JIT to warmup in most cases. This will need to be adjusted -- maybe not all the way back up to current levels, however.
I think the prerequisite here is to probably add more parallelism, see #7.
In order to get the total runtime down below Github's limit of 6 hours, I used
--debug-single-run
which runs each benchmark exactly once. This is likely not enough time for specialization or the JIT to warmup in most cases. This will need to be adjusted -- maybe not all the way back up to current levels, however.I think the prerequisite here is to probably add more parallelism, see #7.