Re: [Faudiostream-users] [Faudiostream-devel] "ZDF" SVF in C++, could this be converted to FAUST?
This even simpler code :
foo(x) = 5;
process = (foo ~ _) ;
Produces suboptimal code:
virtual void compute(int count, FAUSTFLOAT** inputs, FAUSTFLOAT** outputs)
{
FAUSTFLOAT* output0 = outputs[0];
for (int i = 0; (i < count); i = (i + 1)) {
int iRec0 = 5;
output0[i] = FAUSTFLOAT(iRec0);
}
}
Stéphane
> Le 20 juil. 2018 à 13:08, Kjetil Matheussen a
> écrit :
>
>
> On Fri, Jul 20, 2018 at 11:35 AM, Kjetil Matheussen
> wrote:
>
> Actually, Faust should have been able to convert
> m_a1/m_a2/m_a3/m_m0/m_m1/m_m2 into
> slow variables so that we would have avoided all the conditional code in the
> inner loop.
> I'll see if I can come up with a smaller example.
>
>
> I simplified the problem down to this faust program:
>
> """
> modeSlider = hslider ( "Mode" , 0 , 0 , 1 , 1 ) ;
>
> returnfunc(a) = select2(a==0, 2, 3);
>
> process(x) = (function__1936 ~ (_)) : returnfunc with{
> function__1936(a) = modeSlider;
> };
> """
>
> which compiles down to this C++ code:
>
> """
> virtual void compute (int count, FAUSTFLOAT** input, FAUSTFLOAT**
> output) {
>FAUSTFLOAT* input0 = input[0];
>FAUSTFLOAT* output0 = output[0];
>for (int i=0; ifloat fRec0 = float(fslider0);
>output0[i] = (FAUSTFLOAT)((int((fRec0 == 0)))?3:2);
>}
>}
> }
> """
>
> which would have been faster if "fRec0" and "(FAUSTFLOAT)((int((fRec0 ==
> 0)))?3:2);"
> had been compiled down to slow variables.
>
>
> --
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org!
> http://sdm.link/slashdot___
> Faudiostream-devel mailing list
> faudiostream-de...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/faudiostream-devel
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Faudiostream-users mailing list
Faudiostream-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/faudiostream-users
Re: [Faudiostream-users] [Faudiostream-devel] "ZDF" SVF in C++, could this be converted to FAUST?
On Fri, Jul 20, 2018 at 11:35 AM, Kjetil Matheussen <
k.s.matheus...@gmail.com> wrote:
>
> Actually, Faust should have been able to convert
> m_a1/m_a2/m_a3/m_m0/m_m1/m_m2 into
> slow variables so that we would have avoided all the conditional code in
> the inner loop.
> I'll see if I can come up with a smaller example.
>
>
I simplified the problem down to this faust program:
"""
modeSlider = hslider ( "Mode" , 0 , 0 , 1 , 1 ) ;
returnfunc(a) = select2(a==0, 2, 3);
process(x) = (function__1936 ~ (_)) : returnfunc with{
function__1936(a) = modeSlider;
};
"""
which compiles down to this C++ code:
"""
virtual void compute (int count, FAUSTFLOAT** input, FAUSTFLOAT**
output) {
FAUSTFLOAT* input0 = input[0];
FAUSTFLOAT* output0 = output[0];
for (int i=0; i--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Faudiostream-users mailing list
Faudiostream-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/faudiostream-users
Re: [Faudiostream-users] [Faudiostream-devel] "ZDF" SVF in C++, could this be converted to FAUST?
Poing Imperatif was made for making it easier to convert C++ and similar
languages into Faust.
I made this Poing Imperatif program from your C++ code:
https://pastebin.com/DuVWvu7w
Poing Imperatif created this Faust code: https://pastebin.com/fX0nkaGJ
And finally Faust created this C++ code:
virtual void compute(int count, FAUSTFLOAT** inputs, FAUSTFLOAT** outputs) {
FAUSTFLOAT* input0 = inputs[0];
FAUSTFLOAT* output0 = outputs[0];
int iSlow0 = int(float(fHslider0));
int iSlow1 = (iSlow0 == 0);
int iSlow2 = (iSlow0 == 1);
int iSlow3 = (iSlow0 == 2);
int iSlow4 = (iSlow0 == 3);
int iSlow5 = (iSlow0 == 4);
int iSlow6 = (iSlow0 == 5);
int iSlow7 = (iSlow0 == 6);
int iSlow8 = (iSlow0 == 7);
float fSlow9 = float(fHslider1);
float fSlow10 = std::tan((6.54498435e-05f * fSlow9));
float fSlow11 = float(fHslider2);
float fSlow12 = std::pow(10.0f, (0.025004f * fSlow11));
float fSlow13 = std::sqrt(fSlow12);
float fSlow14 = (fSlow10 / fSlow13);
float fSlow15 = float(fHslider3);
float fSlow16 = (1.0f / fSlow15);
float fSlow17 = (((fSlow10 * (fSlow14 + fSlow16)) / fSlow13) + 1.0f);
float fSlow18 = (fSlow10 / (fSlow13 * fSlow17));
float fSlow19 = ((fSlow10 * (fSlow10 + fSlow16)) + 1.0f);
float fSlow20 = (fSlow10 / fSlow19);
float fSlow21 = (1.0f / fSlow17);
float fSlow22 = (1.0f / fSlow19);
float fSlow23 = mydsp_faustpower2_f(fSlow10);
float fSlow24 = (fSlow23 / (mydsp_faustpower2_f(fSlow13) * fSlow17));
float fSlow25 = (fSlow23 / fSlow19);
float fSlow26 = (2.0f * fSlow18);
float fSlow27 = (2.0f * fSlow20);
float fSlow28 = (2.0f / fSlow17);
float fSlow29 = (2.0f / fSlow19);
float fSlow30 = (2.0f * fSlow24);
float fSlow31 = (2.0f * fSlow25);
float fSlow32 = mydsp_faustpower2_f(fSlow12);
float fSlow33 =
(iSlow1?0.0f:(iSlow2?1.0f:(iSlow3?0.0f:(iSlow4?1.0f:(iSlow5?1.0f:(iSlow6?1.0f:(iSlow7?1.0f:(iSlow8?fSlow32:0.0f;
float fSlow34 = (fSlow32 + -1.0f);
float fSlow35 = (0.0f - fSlow16);
float fSlow36 =
(iSlow1?0.0f:(iSlow2?fSlow35:(iSlow3?1.0f:(iSlow4?fSlow35:(iSlow5?fSlow35:(iSlow6?(fSlow34
/ fSlow15):(iSlow7?((fSlow12 + -1.0f) / fSlow15):(iSlow8?((fSlow12 * (1.0f
- fSlow12)) / fSlow15):0.0f;
float fSlow37 =
(iSlow1?1.0f:(iSlow2?-1.0f:(iSlow3?0.0f:(iSlow4?0.0f:(iSlow5?-2.0f:(iSlow6?0.0f:(iSlow7?fSlow34:(iSlow8?(1.0f
- fSlow32):0.0f;
for (int i = 0; (i < count); i = (i + 1)) {
int iTemp0 = iRec8[1];
float fTemp1 = float(input0[i]);
float fTemp2 = (fSlow18 * fTemp1);
float fTemp3 = (fSlow20 * fTemp1);
float fTemp4 = (fTemp1 - fRec3[1]);
float fTemp5 = (fRec2[1] + (fSlow14 * fTemp4));
float fTemp6 = (fSlow21 * fTemp5);
float fTemp7 = (fRec2[1] + (fSlow10 * fTemp4));
float fTemp8 = (fSlow22 * fTemp7);
float fRec0 =
(iTemp0?(iSlow1?fTemp8:(iSlow2?fTemp8:(iSlow3?fTemp8:(iSlow4?fTemp8:(iSlow5?fTemp8:(iSlow6?fTemp8:(iSlow7?fTemp6:(iSlow8?fTemp6:0.0f:(iSlow1?fTemp3:(iSlow2?fTemp3:(iSlow3?fTemp3:(iSlow4?fTemp3:(iSlow5?fTemp3:(iSlow6?fTemp3:(iSlow7?fTemp2:(iSlow8?fTemp2:0.0f);
float fTemp9 = (fSlow24 * fTemp1);
float fTemp10 = (fSlow25 * fTemp1);
float fTemp11 = (fRec3[1] + (fSlow18 * fTemp5));
float fTemp12 = (fRec3[1] + (fSlow20 * fTemp7));
float fRec1 =
(iTemp0?(iSlow1?fTemp12:(iSlow2?fTemp12:(iSlow3?fTemp12:(iSlow4?fTemp12:(iSlow5?fTemp12:(iSlow6?fTemp12:(iSlow7?fTemp11:(iSlow8?fTemp11:fRec3[1]:(iSlow1?fTemp10:(iSlow2?fTemp10:(iSlow3?fTemp10:(iSlow4?fTemp10:(iSlow5?fTemp10:(iSlow6?fTemp10:(iSlow7?fTemp9:(iSlow8?fTemp9:0.0f);
float fTemp13 = (fSlow26 * fTemp1);
float fTemp14 = (fSlow27 * fTemp1);
float fTemp15 = ((fSlow28 * fTemp5) - fRec2[1]);
float fTemp16 = ((fSlow29 * fTemp7) - fRec2[1]);
fRec2[0] =
(iTemp0?(iSlow1?fTemp16:(iSlow2?fTemp16:(iSlow3?fTemp16:(iSlow4?fTemp16:(iSlow5?fTemp16:(iSlow6?fTemp16:(iSlow7?fTemp15:(iSlow8?fTemp15:(0.0f
-
fRec2[1]):(iSlow1?fTemp14:(iSlow2?fTemp14:(iSlow3?fTemp14:(iSlow4?fTemp14:(iSlow5?fTemp14:(iSlow6?fTemp14:(iSlow7?fTemp13:(iSlow8?fTemp13:0.0f);
float fTemp17 = (fSlow30 * fTemp1);
float fTemp18 = (fSlow31 * fTemp1);
float fTemp19 = ((2.0f * fTemp11) - fRec3[1]);
float fTemp20 = ((2.0f * fTemp12) - fRec3[1]);
fRec3[0] =
(iTemp0?(iSlow1?fTemp20:(iSlow2?fTemp20:(iSlow3?fTemp20:(iSlow4?fTemp20:(iSlow5?fTemp20:(iSlow6?fTemp20:(iSlow7?fTemp19:(iSlow8?fTemp19:fRec3[1]:(iSlow1?fTemp18:(iSlow2?fTemp18:(iSlow3?fTemp18:(iSlow4?fTemp18:(iSlow5?fTemp18:(iSlow6?fTemp18:(iSlow7?fTemp17:(iSlow8?fTemp17:0.0f);
float fRec4 = fSlow33;
float fRec5 = fSlow36;
float fRec6 = fSlow37;
float fRec7 = fTemp1;
iRec8[0] = 1;
output0[i] = FAUSTFLOATfRec4 * fRec7) + (fRec5 * fRec0)) + (fRec6 *
fRec1)));
fRec2[1] = fRec2[0];
fRec3[1] = fRec3[0];
iRec8[1] = iRec8[0];
}
}
If faust had known that "mode" doesn't change very often, it could probably
have optimized a bit here.
Or perhaps it would be more naturally to make one faust program for each
mode.
On Thu, Jul 19, 2018 at 12:55 PM, Oliver Larkin via Faudiostream-devel <
faudiostream-de...@lists.sourceforge.net> wrote:
> Well done to the IFC team for a great conference!
