But this was just for HEAD. Compiling 2.27.2 does not work if you want
to compile the interpreter without llvm, as I reported originally:
https://github.com/grame-cncm/faust/blob/2.27.2/compiler/generator/interpreter/fbc_compiler.hh#L31
On Sun, Dec 27, 2020 at 5:48 PM Kjetil Matheussen
wrote
; is ignored if
"STATIC DYNAMIC" is included as well.
Sorry for the noise.
On Sun, Dec 27, 2020 at 5:12 PM Kjetil Matheussen
wrote:
>
> No, disabling CPP_BACKEND made no difference. It's not working. This
> bug has been introduced before the release of 2.27.2.
>
&g
No, disabling CPP_BACKEND made no difference. It's not working. This
bug has been introduced before the release of 2.27.2.
On Sun, Dec 27, 2020 at 5:04 PM Kjetil Matheussen
wrote:
>
> As far as I can see, that Makefile does exactly the same as me, EXCEPT
> that I also compile
hout LLVM;
>
> https://github.com/VCVRack/VCV-Prototype/blob/faust/Makefile
>
> Stéphane
>
> > Le 27 déc. 2020 à 10:51, Kjetil Matheussen a
> > écrit :
> >
> > Thank you. I haven't tried HEAD yet, but in 2.27.2 the LLVM
> > interpreter was compiled in when
(Sorry about the lineshifts. Last time I checked it was impossible to
prevent gmail from automatically adding lineshifts)
On Sun, Dec 27, 2020 at 2:58 PM Kjetil Matheussen
wrote:
>
> I'm not sure which interpreter is actually used, but I'm certain that
> libfaust requires LL
home/kjetil/radium/bin/packages/faust/compiler/generator/interpreter/interpreter_code_container.cpp:28,
from
/home/kjetil/radium/bin/packages/faust/compiler/libcode.cpp:81:
/home/kjetil/radium/bin/packages/faust/compiler/generator/interpreter/fbc_compiler.hh:31:2:
error: #error "
ne Letz wrote:
>
> The LLVM code in the Interp backend is not activated by default, so you
> should be able to compile a version of libfaust using Interp backend with no
> LLVM dependency.
>
> Stéphane
>
> > Le 26 déc. 2020 à 15:56, Kjetil Matheussen a
> > écr
Hi,
Seems like the interpreter in libfaust 2.27.2 requires LLVM. Is this
how it's going to be in the future, or would it be safe to enable the
old non-LLVM interpreter? (I saw there was a hardcoded c-macro that
perhaps can be used to get the old version back.)
The main reason I don't want LLVM is
On Fri, Apr 19, 2019 at 11:23 AM Kjetil Matheussen
wrote:
>
> The reason I noticed this was because in Debug mode I always run the
> -fsanitize=address, which sometimes increases CPU usage a lot, and the
> zita reverb often caused overruns. This stopped when not using -vec.
The d
On Fri, Apr 19, 2019 at 10:20 AM Hermann Meyer wrote:
>
>
> As well, no proper benchmark here, but, I just give zita_rev1 a shot
> here, and didn't see any spices in guitarix. zita rev in vector mode
> (32) use here ~0.8% DSP when switched on in guitarix.
>
guitarix probably don't use enough memo
On Fri, Apr 19, 2019 at 9:41 AM Hermann Meyer wrote:
>
>
> Am 29.11.17 um 12:18 schrieb Stéphane Letz:
> >> I'm still on faust(0.9.90), but this reads very interesting. I'll check
> >> that out as soon I've switched to the latest faust-git.
> > OK
> >> Surly we can't use that in our dsp2cc script
ome kind of global lock provided by faust in the case of
radium.)
On Tue, Apr 9, 2019 at 10:19 AM Stéphane Letz wrote:
>
>
> > Le 9 avr. 2019 à 10:10, Kjetil Matheussen a
> > écrit :
> >
> > On Tue, Apr 9, 2019 at 10:07 AM Stéphane Letz wrote:
> >>
>
On Tue, Apr 9, 2019 at 10:07 AM Stéphane Letz wrote:
>
>
> > Le 9 avr. 2019 à 09:17, Kjetil Matheussen a
> > écrit :
> >
> > Hi,
> >
> > I'm experimenting with using the interpreter factory instead of the
> > llvm factory to compile fa
Hi,
I'm experimenting with using the interpreter factory instead of the
llvm factory to compile faust programs in Radium. It's working really
great! (haven't benchmarked properly, but at least when radium is
compiled in DEBUG mode, the CPU difference doesn't seem very big). I'm
going to skip the i
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 Mat
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
On Fri, Jul 20, 2018 at 11:07 AM, Kjetil Matheussen <
k.s.matheus...@gmail.com> wrote:
>
> 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 prog
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:
vi
On Thu, Jun 8, 2017 at 1:14 PM, Stéphane Letz wrote:
> The reason is that we want to share a unique table between all instances
> of the class. But… it appears that:
>
> - even if the table is shared, the classInit code is actually called each
> time in the instance::init(SR) method, which is rat
On Thu, Jun 8, 2017 at 12:31 PM, Martin Klang wrote:
> I've got another question regarding tables, hopefully not as silly as
> the last one ->
>
> When creating a table in Faust it produces C++ code with a static global
> variable to hold the data.
>
> `static float ftbl0mydspSIG0[65537];`
>
> (f
Hi,
While attempting to lower gc pause time in Radium I've tried to minimize
the root set
used by the bdw-gc garbage collector.
I discovered that Radium had a 25Mb .bss section, i.e. 25Mb of
uninitialized static data.
(25Mb could cause maybe around 5-20 ms of pause time.)
After digging, it turns
Hi,
Sometimes when compiling faust code for windows, it fails because
the word "interface" is a defined a "struct" in some of the system headers.
This is a known problem:
http://stackoverflow.com/questions/299565/how-to-disable-interface-keyword-on-visual-c-express-2008
Here's a fix for mingw:
Very interesting that you get it down to only 4-5 times slower.
I guess this could also be useful to avoid linking in LLVM, which takes up
20-40 megabytes
or thereabout. And, to avoid LLVM linking problems. I've given up making
faust
with LLVM work on windows 32, at least for now.
On Fri, Jul 1,
#define LLVM_MALLOC "malloc"
#define LLVM_FREE "free"
#endif
On Wed, Jun 22, 2016 at 2:51 PM, Kjetil Matheussen wrote:
> No, that test could fail. The important thing is whether llvm has fabsf,
> etc.
>
> Here is another attempt:
>
> http://users.no
gmail trying to be smart when copying and pasting urls, so the link in the
previous mail pointed to the wrong url. This one should be correct:
http://users.notam02.no/~kjetism/diff.diff
(written manually)
On Wed, Jun 22, 2016 at 2:42 PM, Kjetil Matheussen wrote:
> Okay, here is another pa
if
#define LLVM_MALLOC "malloc"
#define LLVM_FREE "free"
#endif
)
On Tue, Jun 21, 2016 at 2:38 PM, Kjetil Matheussen wrote:
> I guess the cause for this problem is that fabsf/etc. are defined as
> macros (?).
>
> In case, how about this:
>
t 2:20 PM, Stéphane Letz wrote:
>
> > Le 21 juin 2016 à 00:19, Kjetil Matheussen a
> écrit :
> >
> >
> >
> > 1. To avoid "losing precision" error:
> >
> > -? (long)array_typed1 > (long)array_typed1
> >
"free"
#endif
On Tue, Jun 21, 2016 at 2:22 PM, Stéphane Letz wrote:
> This was indeed required with (at least..) LLVM 3.1 and possibly some
> versions later on.
>
> If you can prepare a patch to deal with that this would be great.
>
> Thanks.
>
> Stép
I'm compiling libfaust for windows, and have been strugging
a lot with the llvm_math stuff.
Because of this code in llvm_dsp_aux.cpp:
#ifdef _WIN32
#define LLVM_MALLOC "llvm_malloc"
#define LLVM_FREE "llvm_free"
#else
#define LLVM_MALLOC "malloc"
#define LLVM_FREE "free"
#endi
1. To avoid "losing precision" error:
-? (long)array_typed1 > (long)array_typed1
+? (intptr_t)array_typed1 >
(intptr_t)array_typed1
2. Windows snprintf doesn't understand "g", but the special
__mingw_snprintf function does:
+#if defined(
Hi, here's a suggestion for a fix. You might want to do it
a different way:
http://users.notam02.no/~kjetism/path_to_content.patch
--
What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic
patterns at
ackend with its own logic. So adding this «
>>> static » if not the right way to do that, and this patch cannot be applied
>>> in this form.
>>>
>>> Sorry for that ((-;
>>>
>>> Stéphane
>>>
>>>
>>>
>>> > Le 14 j
logic. So adding this «
> static » if not the right way to do that, and this patch cannot be applied
> in this form.
>
> Sorry for that ((-;
>
> Stéphane
>
>
>
> > Le 14 juin 2016 à 16:19, Kjetil Matheussen a
> écrit :
> >
> > It works for my use. U
ons of linker errors » is a problem we are
> aware of, but that is a bit more complex to solve in a clean way that this
> simple patch….
>
> Stéphane
>
>
> > Le 14 juin 2016 à 16:08, Kjetil Matheussen a
> écrit :
> >
> > The autogenerated "faustpowerN_f&
The autogenerated "faustpowerN_f" functions in faust2 must
be declared static to avoid "multilple definitions of"
linker errors when linking several faust files.
This solution seem to work:
--- compiler/generator/code_container.cpp~2016-06-08 13:36:37.0
+0200
+++ compiler/generator/co
t;>
>> Yann
>>
>> Yann Orlarey
>>
>> Directeur scientifique
>> www.grame.fr
>>
>>
>>
>> 2016-06-13 23:38 GMT+02:00 Kjetil Matheussen :
>>
>>>
>>>
>>> On Mon, Jun 13, 2016 at 7:25 PM, Albert Graef wr
Thank you, that was faster than the speed of light.
On Tue, Jun 14, 2016 at 12:08 PM, Stéphane Letz wrote:
> It has already be done recently.
>
> Stéphane
>
>
>
> > Le 14 juin 2016 à 12:04, Kjetil Matheussen a
> écrit :
> >
> >
> > to the dsp clas
Hi,
I wonder if you could add
virtual void instanceInit(int samplingFreq) = 0;
to the dsp class in architecture/faust/audio/dsp.h in faust1?
It seems like it's been forgotten.
--
What NetFlow Analyzer can do for
On Mon, Jun 13, 2016 at 7:25 PM, Albert Graef wrote:
> On Mon, Jun 13, 2016 at 6:54 PM, hermann meyer wrote:
>
>> Did that mean that I could have now multiple process calls in one faust
>> source?
>>
>
> Nope, that program is in error. :) But at least you wouldn't expect the
> compiler to crash
On Mon, Jun 13, 2016 at 7:24 PM, Albert Graef wrote:
> On Mon, Jun 13, 2016 at 5:15 PM, Kjetil Matheussen <
> k.s.matheus...@gmail.com> wrote:
>
>> I get a crash when compiling this program:
>>
>> "
>> process = *(v), *(v) with { v = hsli
Hi,
I get a crash when compiling this program:
"
process = *(v), *(v) with { v = hslider("volume[style:knob]", 0.5, 0, 5,
0.05);};
process = *(v), *(v) with { v = hslider("volume[style:knob]", 0.5, 0, 5,
0.05);};
"
It's reproducable using faust2. You might need to compile faust with
-fsanitize=a
e it they work as
> they should.
>
> Stéphane
>
>
> > Le 13 juin 2016 à 10:45, Kjetil Matheussen a
> écrit :
> >
> > Hi,
> >
> > I'm calling functions to create and destroy instances of
> > llvm_dsp_factory and llvm_dsp from multiple thread
Hi,
I'm calling functions to create and destroy instances of
llvm_dsp_factory and llvm_dsp from multiple threads simultaneously,
and I have used a lock to protect access to these calls.
But when I look at the source code in faust (llvm_dsp_aux.cpp),
these functions are already protected by a lock
or::cleanup();
On Fri, Jun 10, 2016 at 9:49 AM, Kjetil Matheussen wrote:
> Oops, sorry, that would cause gCurrentLocal not to be set back to its
> original value.
> I guess this is good enough:
>
> [kjetil@ttlush faust2]$ diff -u compiler/global.cpp~ compiler/global.cpp
> --- co
s/2012-May/023530.html)
-setlocale(LC_ALL, "C");
+gCurrentLocal = setlocale(LC_ALL, "C");
gAllocationCount = 0;
On Fri, Jun 10, 2016 at 9:47 AM, Kjetil Matheussen wrote:
> Hi,
>
> When compiling faust with -fsanitizer=address (asan), I get this crash:
&g
Hi,
When compiling faust with -fsanitizer=address (asan), I get this crash:
==6397==ERROR: AddressSanitizer: heap-use-after-free on address
0x612000434a40 at pc 0x770e074d bp 0x7fff9ed0 sp 0x7fff9680
READ of size 3 at 0x612000434a40 thread T0
#0 0x770e074c in __interceptor_set
-#if defined(LLVM_37) || defined(LLVM_38)
-type_def = array_type;
-#endif
return gvar_array_string0;
} else {
return fGlobalStringTable[str];
On Wed, Jun 8, 2016 at 6:00 PM, Kjetil Matheussen
wrote:
> Very strange. It seem
arre, it does not crash here..
>
> Stéphane
>
> > Le 8 juin 2016 à 17:50, Kjetil Matheussen a
> écrit :
> >
> >
> >
> > On Wed, Jun 8, 2016 at 5:27 PM, Kjetil Matheussen <
> k.s.matheus...@gmail.com> wrote:
> >
> > Could it be that
On Wed, Jun 8, 2016 at 5:27 PM, Kjetil Matheussen
wrote:
>
> Could it be that this assert inside LLVM is not very important, and I just
> need to recompile it with NDEBUG defined?
>
No, commenting out the assert line in LLVM
phane Letz wrote:
> Applied and tested (a bit…) on OSX.
>
> Thanks a lot!
>
> Stéphane
>
> > Le 7 juin 2016 à 14:04, Kjetil Matheussen a
> écrit :
> >
> > Hi,
> >
> > Here is a patch to make faust2 compile with llvm 3.8.0:
> > http://users.n
Hi,
Here is a patch to make faust2 compile with llvm 3.8.0:
http://users.notam02.no/~kjetism/faust2_llvm38.patch
Although I mostly didn't know what I was doing, it seems
to work.
I had to recompile llvm though to support rtti (run time type information),
since llvm 3.8 is configured with -no-rtt
pp:8385:38: error: expected ',' or '...' before 'struct'
> virtual void buildUserInterface(UI* interface) {
>
> which I don't know how to fix since the "UI" class is defined in the top
> of test.cpp.
> Anyone seen this error before and know how
ce) {
which I don't know how to fix since the "UI" class is defined in the top of
test.cpp.
Anyone seen this error before and know how to fix it?
On Thu, Jun 2, 2016 at 12:19 AM, Kjetil Matheussen wrote:
> I just looked at the faust2jaqt script. I didn't know that it
ere can I find these intermediate files that generated by
> > "faust2jaqt/faust2jack"?
> >
> > Thank you again for your kindness!
> >
> >
> > Best regards,
> > Jingjie Zhang
> >
> > -原始邮件-
> > 发件人:"Kje
t; I tried it on an old version of Faust.(There is no faust2jack command in
> this version)
> It seems that the Faust command does not have "-I" option.
>
> Tomorrow I will try it on a newer version of Faust.
>
>
> Best regards,
> Jingjie Zhang
>
> -原
it seems
> that MXE needs the whole qt/gtk program files to do cross-compiling.
> Where can I find these intermediate files that generated by
> "faust2jaqt/faust2jack"?
>
> Thank you again for your kindness!
>
>
> Best regards,
> Jingjie Zhang
>
>
On Mon, May 30, 2016 at 5:42 PM, Jingjie Zhang
wrote:
>
> Dear Sirs,
>
>
> This is my first post, and call me JJ if you like.
> I am having trouble compiling a Faust program with a C++ foreign function
> into a standalone application so that I can hear and debug.
>
> I have FaustLive and JACK on
Downloaded and compiled successfully.
Thanks Yann!
On Wed, Jan 6, 2016 at 10:15 AM, Yann Orlarey wrote:
> Hi Kjetil,
>
> Thanks for mentioning the problem. I don't understand the cause, but I
> have uploaded the file again.
> Cheers
>
> Yann
>
> 2016-01-05 2
Hi, it seems like latest faust tar ball file is incomplete:
[kjetil@localhost Downloads]$ gzip -d faust-0.9.73.tgz
gzip: faust-0.9.73.tgz: unexpected end of file
[kjetil@localhost Downloads]$ tar xvzf faust-0.9.73\(1\).tgz
...
faust-0.9.73/documentation/faust-quick-reference.pdf
faust-0.9.73/doc
On Sun, Dec 7, 2014 at 11:56 AM, Oliver Larkin
wrote:
> My version of Cockos' IPlug is another option, which is liberally
> licensed. It does win/mac VST2, VST3, AU, RTAS, AAX and standalone app (no
> linux though).
>
> A while ago I made a (very basic) faust2iplug template project and python
> s
> Or someone needs to sit down and port faust-vst to AU. Shouldn't be that
> hard, so I might do that some time, but don't hold your breath for it. ;-)
>
Another alternative could be make a JUCE architecture file, then you'll get
support for vst/au/rtas/npapi/activex in one architecture file. My
e
61 matches
Mail list logo
