CUDAEnsemble CLI output shows progress incorrectly

[zeyus]

Describe the bug
This is an extremely minor issue. When CUDAEnsemble outputs its progress to the console, it will say how many runs it has completed e.g. 4/10. The problem is the number on the left, seems to just be the index of the first missing log (or the result from one thread I'm not sure). So when I was doing runs, some exit early if the agents die, and with 4 threads let's say threads 3 and 4 finish, then log files 0 and 1 are missing, it will quickly report 2/10 complete then go to 0/10.

To Reproduce
Steps to reproduce the behaviour:

Make a CUDAEnsemble with multiple threads, set some later threads to terminate early.

Expected behaviour
The number should be the total simulation runs completed.

Configuration(please complete the following information):

• OS: Win 11
• CUDA version: 11.7
• GPU GTX 1070
• GPU Driver Version 516.59

[Robadob]

Yeah I'm aware of this, thought there was an issue, guess not.

Afaik, the progress lists the index of the last job to finish, so if they end out of order they were started the progress counter can go backwards.

[Robadob]

This is the offending line of code

FLAMEGPU2/src/flamegpu/sim/SimRunner.cu

Line 106 in e176d8a

fprintf(stdout, "\rCUDAEnsemble progress: %u/%u", run_id + 1, static_cast<unsigned int>(plans.size()));

The solution might be to replace run_id + 1 with next_run.get() - <number of runners> + 1.

The total number of runners would need to be passed to the SimRunner at construction though.