Summary OPs wrong calculation

[mirimirima]

I performed a very basic test using 1MiB as transfer size and this led to same numbers for bw and OPs in the summary of the test while the bw and IOPs values in the output of each single test were different. Attached you can find the output of the test.
IOR-RW-Single_File-c_71-8GiB-s_1_2024-02-17_15:28:01.txt

Looking at the source code I noticed that:

• for the summary output there is one unique formula used for both the bw and the ops
  r->val[i] = ((double) (point->aggFileSizeForBW)) / transfer_size / vals[i];
  (https://github.com/hpc/ior/blob/bfc922f22543b63a324ca7ef9f92c67dc24d1bdd/src/ior-output.c#L497C17-L498C55)
  and to obtain the bw the transfer_size is set to 1 in the code (

  ior/src/ior-output.c

  Line 520 in bfc922f

  static struct results *bw_values(const int reps, IOR_results_t *measured,

  ) while to obtain the OPs the transfer_size is equal to the transfer given as input (

  ior/src/ior-output.c

  Line 526 in bfc922f

  static struct results *ops_values(const int reps, IOR_results_t *measured,

  )
• for the output of each single test on the contrary there are two different formula used since the bw is evaluated with the totalTime while the IOPs is evaluated with the accessTime:
  bw = (double)point->aggFileSizeForBW / totalTime; iops = (point->aggFileSizeForBW / params->transferSize) / accessTime;
  (https://github.com/hpc/ior/blob/bfc922f22543b63a324ca7ef9f92c67dc24d1bdd/src/ior.c#L779C9-L783C78)