does anyone know if one exists? i found an x64 compile for wavpack that is even slower than the stock. looking for one since i would like at least some speed improvement when encoding at x6.
I downloaded wavpack sources and compiled it with intel compiler 11.1.067. It is as fast as original .exe or (using different compiler options) crashes.
I downloaded wavpack sources and compiled it with intel compiler 11.1.067. It is as fast as original .exe or (using different compiler options) crashes.
Maybe there is a version for many-core processors???
I downloaded wavpack sources and compiled it with intel compiler 11.1.067. It is as fast as original .exe or (using different compiler options) crashes.
Maybe there is a version for many-core processors???
Run multiple instances at the same time.
Run multiple instances at the same time.
how
@for /F "delims=" %%S in ('dir /S /A-D /B *.wav') do @ (
@start /B /WAIT /LOW wavpack.exe -mihhb5x3 "%%~dpnS.wav" "%%~dpnS.wv"
@del "%%~dpnS.wav"
@echo ********************************************************************************
)
@for /F "delims=" %%S in ('dir /S /A-D /B *.flac') do @ (
@start /B /WAIT /LOW flac.exe -d -s "%%~dpnS.flac" -o - | wavpack.exe -mihhb5x3 - "%%~dpnS.wv"
@tag.exe --fromfile "%%~dpnS.flac" --ape2 "%%~dpnS.wv"
@del "%%~dpnS.flac"
@echo ********************************************************************************
)
@for /F "delims=" %%S in ('dir /S /A-D /B *.ape') do @ (
@start /B /WAIT /LOW mac.exe "%%~dpnS.ape" - -d | wavpack.exe -mihhb5x3 - "%%~dpnS.wv"
@tag.exe --fromfile "%%~dpnS.ape" --ape2 "%%~dpnS.wv"
@del "%%~dpnS.ape"
@echo ********************************************************************************
)My script))
This script launch only one wavpack.exe
I downloaded wavpack sources and compiled it with intel compiler 11.1.067. It is as fast as original .exe or (using different compiler options) crashes.
i thought intel's compiler produces faster binaries due to auto-vectorization of the exe.
It may be that because the WavPack code is so optimized for the MSVC compiler, there is no advantage to going to Intel’s, at least not without making other changes. I know that in the past this was the case.
I've done some assembly optimization for PAQ8 for x86_64, but only for time critical functions, the rest was compiled with a standard gcc.
If you really want to optimize, re-implement SSE compatible parts in assembly and assemble them with YASM. non-computing heavy code may be open for compiler optimization.
There is no reason to use assembly when intrinsics are available.
Using intrinsics lets you write assembler like writing C, so there's no need for writing several assembler routines for the different compilers.