CoreConverter -processor problem

[chrisjj]

This

start "" CoreConverter.exe -infile=\temp\1.wma -outfile=\temp\1.wav -processor="0" -convert_to="Wave"
         CoreConverter.exe -infile=\temp\2.wma -outfile=\temp\2.wav -processor="1" -convert_to="Wave"

is no quicker to complete than

start "" CoreConverter.exe -infile=\temp\1.wma -outfile=\temp\1.wav -processor="0" -convert_to="Wave"
         CoreConverter.exe -infile=\temp\2.wma -outfile=\temp\2.wav -processor="0" -convert_to="Wave"

(same core number).

And the second command runs slower until the first has completed.

Both suggest that -processor= is not working. That both conversions are running on the the same processor (core).

This is a four-core CPU running Windows 7 64-bit Pro.

Can anyone suggest the cause, and a remedy? I want to do parallel conversions from a batch file.

Thank you.

[Spoon]

Re: CoreConverter -processor problem

You should try without processor, Windows its self will allocate to cpu.

[chrisjj]

Re: CoreConverter -processor problem

You should try without processor, Windows its self will allocate to cpu.

Thanks. Removing -processor does reduce execution time, confirming Windows is allocating core.

But the reduction is only about 20%. This is I guess because of the non-fully-paralleled file read/write time - seen as the difference between a single conversion's reported conversion time of 1.6s and measured execution time of 2.4s.

Also I find:

The first case (-processor=0/1) takes 5.33s, with conversion reported as 2.2s and 3.2s.
The second case (-processor=0/0) also takes 5.33s, also with conversion reported as 2.2s and 3.2s.

I think this indicates that -processor is failing to take effect here. Experiment shows 0 is being interpreted as 1. I take it the documentation https://www.dbpoweramp.com/developer-cli-encoder.htm "n is 1=CPU1 2=CPU2, etc" means the first core is 0 not 1.

-processor=1/1 takes 5.33s, with conversion reported as 2.2s and 3.2s.
-processor=1/2 takes ~4.00s, with conversion reported as 1.6s and 1.6s.
No -processor=1/2 takes again ~4.00s, with conversion reported as 16.s and 1.6s.

I think this confirms -processor is working.

On this 4-core CPU, -processor="5" seems to have no effect. So I take it out-of-range core number causes -processor to be ignored.

This is on Intel(R) Core(TM) i5-3550 CPU @ 3.30GHz .