MVerb - A Free Open Source Reverb
-
lubomir.ivanov lubomir.ivanov https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=226506
- KVRist
- 31 posts since 22 Feb, 2010
re: ccernn and caco
it all seems target specific for the performance tests looking at the results you have reported.
in the last zip i have compiled with the official mingw 3.4.2 build for the win32 dlls as it seemed the most efficient.
i have the mingw 3.4.2 builds running just a little bit slower than the original msvc build (3.4.2 = 3.6% cpu for one instance, while msvc around 3.4% cpu)
the strange part i guess comes when i compile with the more recent mingw-gcc-4.4.1-tdm and mingw-gcc-4.5.0-tdm (which are supposedly better). but these compilers generate code that is almost 2x slower.
i have tested a lot with 3.x vs 4.x for basic arithmetics (+ trigonometry, prngs etc..). the 4.x series have been dominating the performance test in every aspect.. however i have noticed that most of my test code is pretty much plain c. i have a suspicion that one of the main factors involved here is related to c++ code (as mverb is).
but also perhaps the most major factor could be that -O3 level is different between the 3.x and the 4.x series. i think this could come down to a single flag...some more test are required.
i have recently shared a report with ccernn for a possible compiler bug with mingw 4.4.1 that will generate bogus code if math.h - tanf() is used on a negative number and only in certain conditions (such as -msse -mfpmath=msse). such issues are complicated and difficult to debug, unfortunately :\
it all seems target specific for the performance tests looking at the results you have reported.
in the last zip i have compiled with the official mingw 3.4.2 build for the win32 dlls as it seemed the most efficient.
i have the mingw 3.4.2 builds running just a little bit slower than the original msvc build (3.4.2 = 3.6% cpu for one instance, while msvc around 3.4% cpu)
the strange part i guess comes when i compile with the more recent mingw-gcc-4.4.1-tdm and mingw-gcc-4.5.0-tdm (which are supposedly better). but these compilers generate code that is almost 2x slower.
i have tested a lot with 3.x vs 4.x for basic arithmetics (+ trigonometry, prngs etc..). the 4.x series have been dominating the performance test in every aspect.. however i have noticed that most of my test code is pretty much plain c. i have a suspicion that one of the main factors involved here is related to c++ code (as mverb is).
but also perhaps the most major factor could be that -O3 level is different between the 3.x and the 4.x series. i think this could come down to a single flag...some more test are required.
i have recently shared a report with ccernn for a possible compiler bug with mingw 4.4.1 that will generate bogus code if math.h - tanf() is used on a negative number and only in certain conditions (such as -msse -mfpmath=msse). such issues are complicated and difficult to debug, unfortunately :\
linux/win32 framework for plugins and exacutables:
holos.googlecode.com
holos.googlecode.com
-
- KVRian
- Topic Starter
- 995 posts since 25 Apr, 2005
Thanks ccern, I have added the Linux versions to my website so the latest versions for Windows VST, OSX VST and AU, and Linux VST and LADPSA can now all be downloaded from www.martineastwood.com 
To keep this thred on topic about MVerb, I have started a a new thread in the dsp forum about compilers, msvc vs mingw etc as I would like to continue that discussion and I am sure other people there would be interested in your findings
To keep this thred on topic about MVerb, I have started a a new thread in the dsp forum about compilers, msvc vs mingw etc as I would like to continue that discussion and I am sure other people there would be interested in your findings
- KVRian
- 1297 posts since 23 Jun, 2007 from Findlay OH USA
Just a quick note to say "Thanks!" to Martin, ccernn, and lubomir. I'm always pleased to see new stuff for Linux, and this is a sweet reverb.
Best regards,
dp
Best regards,
dp
- KVRAF
- 7091 posts since 19 Apr, 2002 from Utah
Me too. Thank you!!!!!!
--Sean
--Sean
-
- KVRist
- 154 posts since 20 Jan, 2006
Hi. Thanks for this great plugin.
Is it possible to get a version without GUI?
Thanks.
Is it possible to get a version without GUI?
Thanks.
-
- KVRian
- Topic Starter
- 995 posts since 25 Apr, 2005
Hi gemada,gemada wrote:Hi. Thanks for this great plugin.
Is it possible to get a version without GUI?
Thanks.
the original Windows version I released without the GUI is still avilable Here
-
- KVRist
- 154 posts since 20 Jan, 2006
Thanks a lot for this.Caco wrote:Hi gemada,gemada wrote:Hi. Thanks for this great plugin.
Is it possible to get a version without GUI?
Thanks.
the original Windows version I released without the GUI is still avilable Here
It's a great plugin indeed
All the best
-
- KVRAF
- 1666 posts since 28 Jun, 2007 from Amazon rain forest
Just tried it and I must say it's the best free reverb I've tested so far. Really smooth, natural and low CPU.
Thank you very much for this one!
Thank you very much for this one!
-
- KVRian
- 1439 posts since 25 Nov, 2008 from Seattle, WA
-
- KVRian
- Topic Starter
- 995 posts since 25 Apr, 2005
I have been given an x64 LADSPA version (Linux only), which I will upload when I get chanceaircargo wrote:Has anyone attempted to build an x64 native version of this?
-
- KVRian
- 1439 posts since 25 Nov, 2008 from Seattle, WA
-
- KVRian
- 676 posts since 22 Sep, 2007 from Caracas, Venezuela
Yes, a x64 version for Windows would be great 
-
lubomir.ivanov lubomir.ivanov https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=226506
- KVRist
- 31 posts since 22 Feb, 2010
here is a debug, vst build of mverb for 64bit windows:
http://dl.dropbox.com/u/1627980/axon/mv ... ug-x64.zip
(1.6mb zip)
if the plugin crashes on your machine please report:
- cpu model
- windows version
- crash log
- does it crash only in specific hosts
- etc
this will be very helpful to us,
thanks.
ps: note that this debug build is around 28 times larger and 2-3 times slower.
http://dl.dropbox.com/u/1627980/axon/mv ... ug-x64.zip
(1.6mb zip)
if the plugin crashes on your machine please report:
- cpu model
- windows version
- crash log
- does it crash only in specific hosts
- etc
this will be very helpful to us,
thanks.
ps: note that this debug build is around 28 times larger and 2-3 times slower.
linux/win32 framework for plugins and exacutables:
holos.googlecode.com
holos.googlecode.com
-
- KVRian
- 676 posts since 22 Sep, 2007 from Caracas, Venezuela
Thank you very much lubomir!lubomir.ivanov wrote:here is a debug, vst build of mverb for 64bit windows:
http://dl.dropbox.com/u/1627980/axon/mv ... ug-x64.zip
(1.6mb zip)
if the plugin crashes on your machine please report:
- cpu model
- windows version
- crash log
- does it crash only in specific hosts
- etc
this will be very helpful to us,
thanks.
ps: note that this debug build is around 28 times larger and 2-3 times slower.
Are you LiteOn from Reaper Forums and the creator of the amaizing JS pack?
