I noticed the following two issues with the current Benchmark/SW setup in Spatz when working on some kernel improvement.
- The printf of floating values in 32b configuration is broken. The compiler will map it to 64b and cause an illegal instruction
- Even if the test fails (returns -1), the CI will still say the test completed successfully and mark it as passed.
I will try to fix these two issues in the PR I opened for benchmark improvement
I noticed the following two issues with the current Benchmark/SW setup in Spatz when working on some kernel improvement.
I will try to fix these two issues in the PR I opened for benchmark improvement