easy_bf and easy_bf_fast do not have strictly the same outputs
Hi,
I was using the scheduler easy_bf
in my work, and wanted to switch to easy_bf_fast
for performance reasons.
However, while making some tests, I realized that they do not produce exactly the same outputs on the same inputs.
Steps to reproduce
The following folder contains every file needed to reproduce the bug (nix file, platform file, workload and robin file): mwe_easy_bf_fast.zip.
Enter the folder and run
nix-shell -A exp_env
robin easy_bf.yaml
robin easy_bf_fast.yaml
diff out/easy_bf/_jobs.csv out/easy_bf_fast/_jobs.csv
We see that job 58 gets executed earlier with easy_bf_fast
.
10 first lines of easy_bf _jobs.csv
```txt job_id,workload_name,profile,submission_time,requested_number_of_resources,requested_time,success,final_state,starting_time,execution_time,finish_time,waiting_time,turnaround_time,stretch,allocated_resources,consumed_energy,metadata 0,w0,delay_ft.B.1,0.000000,1,149.000000,1,COMPLETED_SUCCESSFULLY,0.000000,88.510000,88.510000,0.000000,88.510000,1.000000,0,-1.000000,"" 1,w0,delay_ft.C.2,4.709164,2,399.000000,1,COMPLETED_SUCCESSFULLY,88.510000,265.800000,354.310000,83.800836,349.600836,1.315278,0-1,-1.000000,"" 4,w0,delay_ft.B.1,54.195921,1,149.000000,1,COMPLETED_SUCCESSFULLY,354.310000,88.510000,442.820000,300.114079,388.624079,4.390736,1,-1.000000,"" 2,w0,delay_lu.B.1,16.003927,1,420.000000,1,COMPLETED_SUCCESSFULLY,354.310000,279.990000,634.300000,338.306073,618.296073,2.208279,0,-1.000000,"" 13,w0,delay_is.B.1,168.987800,1,70.000000,1,COMPLETED_SUCCESSFULLY,634.300000,9.257000,643.557000,465.312200,474.569200,51.265983,0,-1.000000,"" 5,w0,delay_lu.B.1,70.350749,1,420.000000,1,COMPLETED_SUCCESSFULLY,442.820000,279.990000,722.810000,372.469251,652.459251,2.330295,1,-1.000000,"" 20,w0,delay_ft.B.1,253.023655,1,149.000000,1,COMPLETED_SUCCESSFULLY,643.557000,88.510000,732.067000,390.533345,479.043345,5.412308,0,-1.000000,"" 7,w0,delay_is.C.2,121.609395,2,87.000000,1,COMPLETED_SUCCESSFULLY,732.067000,26.481000,758.548000,610.457605,636.938605,24.052664,0-1,-1.000000,"" 8,w0,delay_lu.C.2,126.478295,2,891.000000,1,COMPLETED_SUCCESSFULLY,758.548000,593.590000,1352.138000,632.069705,1225.659705,2.064825,0-1,-1.000000,"" ```10 first lines of easy_bf _jobs.csv
```txt job_id,workload_name,profile,submission_time,requested_number_of_resources,requested_time,success,final_state,starting_time,execution_time,finish_time,waiting_time,turnaround_time,stretch,allocated_resources,consumed_energy,metadata 0,w0,delay_ft.B.1,0.000000,1,149.000000,1,COMPLETED_SUCCESSFULLY,0.000000,88.510000,88.510000,0.000000,88.510000,1.000000,0,-1.000000,"" 1,w0,delay_ft.C.2,4.709164,2,399.000000,1,COMPLETED_SUCCESSFULLY,88.510000,265.800000,354.310000,83.800836,349.600836,1.315278,0-1,-1.000000,"" 4,w0,delay_ft.B.1,54.195921,1,149.000000,1,COMPLETED_SUCCESSFULLY,354.310000,88.510000,442.820000,300.114079,388.624079,4.390736,1,-1.000000,"" 2,w0,delay_lu.B.1,16.003927,1,420.000000,1,COMPLETED_SUCCESSFULLY,354.310000,279.990000,634.300000,338.306073,618.296073,2.208279,0,-1.000000,"" 13,w0,delay_is.B.1,168.987800,1,70.000000,1,COMPLETED_SUCCESSFULLY,634.300000,9.257000,643.557000,465.312200,474.569200,51.265983,0,-1.000000,"" 5,w0,delay_lu.B.1,70.350749,1,420.000000,1,COMPLETED_SUCCESSFULLY,442.820000,279.990000,722.810000,372.469251,652.459251,2.330295,1,-1.000000,"" 20,w0,delay_ft.B.1,253.023655,1,149.000000,1,COMPLETED_SUCCESSFULLY,643.557000,88.510000,732.067000,390.533345,479.043345,5.412308,0,-1.000000,"" 58,w0,delay_is.B.1,728.146984,1,70.000000,1,COMPLETED_SUCCESSFULLY,728.146984,9.257000,737.403984,0.000000,9.257000,1.000000,1,-1.000000,"" 7,w0,delay_is.C.2,121.609395,2,87.000000,1,COMPLETED_SUCCESSFULLY,737.403984,26.481000,763.884984,615.794589,642.275589,24.254204,0-1,-1.000000,"" ```Do you understand the reason? What confidence should I have that both the scheduler behave as they should?
Thank you for your help :)
Edited by Mema5