Port DSP code to C++ to address performance penalty #1469
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…r_filter_calculate_coefficients()
This reverts commit 1e34a43. Cost model indicates vectorization is nonbeneficial
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I found that after merging #1444 a small performance penalty was introduced. Looking at VTune's profiling report (see below) showed the outer if clause of
fluid_iir_filter_applybeing pretty busy. I assume this is because the calling code influid_rvoice_dsp.cinvokes that function twice, once with the custom filter typically being disabled, and the regular low pass filter being enabled. Before merging #1444 this was done once per voice. Now it's done for every single sample, which apparently hurts branch prediction and possibly also instruction caching.To address this, I ported the DSP code to C++98, which allows me to use templating to generate multiple versions of the DSP code. One of them e.g. never attempts to use the custom IIR filter in the DSP loop if it's unused for that voice. Also whether or not to construct a 24bit sample has been templatified.
In addition, most of the IIR filter code was moved to the header, allowing the compiler to inline it.
With these changes, our testsuite
make check_manualnow finishes up 7 seconds faster for me.Profiling before this PR (at rev e57360d):
Profiling after this PR,
fluid_iir_filter_applyis no longer that busy.