How to understand Presto Benchmark driver results.

[PranayMunshi1982]

Hi Friends,
I am running Presto Benchmark driver. I am using Presto 318.
Running benchmark driver for TPCDS queries. My data is in HDFS and using ORC file format.
I managed to run the benchmark driver successfully but I can not understand

Q1 Why driver is running queries 3 to 4 times.

Q2. In result of benchmark driver I can see some output but I am not able to understand those terms used in output.
I am looking for some description of following terms -

wallTimeP50, wallTimeMean, wallTimeStd, processCpuTimeP50, processCpuTimeMean, processCpuTimeStd, queryCpuTimeP50, queryCpuTimeMean, queryCpuTimeStd,

Which points I should consider in benchmarking, the above one or the points which appear inside the query results in PRESTO UI?

How can we say that these bench-marking is useful, because in real world many queries run parallely but in case of bench-marking we are running queries sequentially.

@dain