gmarkomanolis
commented
4 years ago +1 for me
Closed adilger closed 4 years ago
gmarkomanolis
commented
4 years ago +1 for me
adilger
commented
4 years ago Actually, the script appended "-invalid" to the per-test results in this manner, which is why I changed it to be more similar to the old script.
That made it very clear which subtest was failing, which I thought wasn't as clear in the new output.
The main other difference compared to the old summary output was that it used to print "BW/IOPS phase N" or similar with each result, but I didn't think that was totally required, and it pushed the per-phase output over 80 columns for little benefit, so I preferred this new output.
Adding in "-invalid" to the per-phase output also pushed it over 80 columns, but since the test is not valid anyway, I didn't think that was critical.
Cheers, Andreas
On May 15, 2020, at 06:24, Julian Kunkel notifications@github.com wrote:
@JulianKunkel commented on this pull request.
In src/main.c:
@@ -330,7 +333,7 @@ int main(int argc, char ** argv){ PRINT_PAIR("hash", "%X\n", (int) score_hash);
dupprintf("[SCORE%s] Bandwidth %f GB/s : IOPS %f kiops : TOTAL %f\n",
- opt.is_valid_run ? "" : " INVALID",
- opt.is_valid_run ? "" : "-invalid", with the exception of this change all looks good. I reckon the script outputted INVALID in uppercase as well, that is why it is that way.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.
Add a constant for the 300s minimum runtime, rather than hard-coding it in several places in the code. It will eventually also needed for Student Cluster Competition.
Use it to print "-invalid" markers for any subtests that didn't run for a long enough time. That should never happen, but it makes the bad result(s) easier to see.
Signed-off-by: Andreas Dilger adilger@dilger.ca