When displaying speeds > 10MB and < 100MB there is a significant error associated with it as it cuts off all fractional part of it. This error can approach 10%, for example: if there is a speed of 10.000 and 10.999 [MB] they will both be displayed as 10 when actual difference is 0.999, and an error is 9.99% ( |10.999-10.000|/10.000 = 0.0999 = 9.99% ). Similarly if those speed were 10.999 and 11.000 then they wold be displayed as 10 and 11, with a difference of 1 when actually it is 0.001. Again error would be 9.99% ( |1-0.001|/10 = 0.0999 = 9.99% ).
So I increased displayed accuracy for speeds > 10MB and < 100MB to one decimal point which reduces the error below 1%.
When displaying speeds > 10MB and < 100MB there is a significant error associated with it as it cuts off all fractional part of it. This error can approach 10%, for example: if there is a speed of 10.000 and 10.999 [MB] they will both be displayed as 10 when actual difference is 0.999, and an error is 9.99% ( |10.999-10.000|/10.000 = 0.0999 = 9.99% ). Similarly if those speed were 10.999 and 11.000 then they wold be displayed as 10 and 11, with a difference of 1 when actually it is 0.001. Again error would be 9.99% ( |1-0.001|/10 = 0.0999 = 9.99% ).
So I increased displayed accuracy for speeds > 10MB and < 100MB to one decimal point which reduces the error below 1%.
Here simple sample:
(from:)
$ ./lzbench -o1 -ezlib,1,5,9/bzip2,1,5,9 lzbench
(To:)
$ ./lzbench-print_10m0 -o1 -ezlib,1,5,9/bzip2,1,5,9 lzbench
Hope you find it reasonable and useful.