U++ framework: U++ Library : Other (not classified elsewhere) » Eigen updated

Home » U++ Library support » U++ Library : Other (not classified elsewhere) » Eigen updated

Show: Today's Messages :: Polls :: Message Navigator
E-mail to friend

Eigen updated [message #53851]

Sat, 09 May 2020 11:03

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

From 2018 Eigen algebra library had no new version, and since interesting improvements have been accumulated, we have updated to the last master commit.
This commit has not given any compilation or functioning problems. However, please check your applications just in case.

Best regards
Iñaki

Report message to a moderator

Re: Eigen updated [message #53854 is a reply to message #53851]

Sat, 09 May 2020 15:17

Novo
Messages: 1432
Registered: December 2006

Ultimate Contributor

koldo wrote on Sat, 09 May 2020 05:03

From 2018 Eigen algebra library had no new version, and since interesting improvements have been accumulated, we have updated to the last master commit.
This commit has not given any compilation or functioning problems. However, please check your applications just in case.

And reference/Eigen_demo is broken with CLANG on Linux.

Regards,
Novo

Report message to a moderator

Re: Eigen updated [message #53859 is a reply to message #53854]

Sat, 09 May 2020 19:26

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Quote:

And reference/Eigen_demo is broken with CLANG on Linux.

Bad luck. I do not have CLANG in Linux. I will try to install it. With GCC in Linux and with CLANG in Windows there is no problem. Thank you for reporting.

Best regards
Iñaki

[Updated on: Sat, 09 May 2020 19:26]

Report message to a moderator

Re: Eigen updated [message #53860 is a reply to message #53859]

Sat, 09 May 2020 19:35

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

I have just compiled and run Eigen_demo in Ubuntu 18.04 with CLANG and GCC and works perfectly.

Best regards
Iñaki

[Updated on: Sat, 09 May 2020 19:35]

Report message to a moderator

Re: Eigen updated [message #53861 is a reply to message #53860]

Sat, 09 May 2020 20:08

Novo
Messages: 1432
Registered: December 2006

Ultimate Contributor

koldo wrote on Sat, 09 May 2020 13:35

I have just compiled and run Eigen_demo in Ubuntu 18.04 with CLANG and GCC and works perfectly.

Ubuntu 20.04
./umk reference Eigen_demo CLANG -bus
Running full rebuild now ...
I was getting this:

BLITZ: eigen_demo.cpp non-linear.cpp fft.cpp
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/.cache/upp.out/Eigen_demo/CLANG.Debug.Debug_Full.Main.Shared/Eigen_demo$blitz.cpp:7:
/home/ssg/.local/soft/bb-worker/worker/upp/build/reference/Eigen_demo/non-linear.cpp:26:9: error: reference to 'Vector' is ambiguous
        static Vector<double> _x, _y;
               ^
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/Core/Vcont.h:104:7: note: candidate found by name lookup is 'Upp::Vector'
class Vector : public MoveableAndDeepCopyOption< Vector<T> > {
      ^
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/src/Core/Matrix.h:549:1: note: candidate found by name lookup is 'Eigen::Vector'
using Vector = Matrix<Type, Size, 1>;
^
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/.cache/upp.out/Eigen_demo/CLANG.Debug.Debug_Full.Main.Shared/Eigen_demo$blitz.cpp:7:
/home/ssg/.local/soft/bb-worker/worker/upp/build/reference/Eigen_demo/non-linear.cpp:46:1: error: reference to 'Vector' is ambiguous
Vector<double> Thurber_functor::_x, Thurber_functor::_y;
^
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/Core/Vcont.h:104:7: note: candidate found by name lookup is 'Upp::Vector'
class Vector : public MoveableAndDeepCopyOption< Vector<T> > {
      ^
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/src/Core/Matrix.h:549:1: note: candidate found by name lookup is 'Eigen::Vector'
using Vector = Matrix<Type, Size, 1>;
^
2 errors generated.
Eigen_demo: 3 file(s) built in (0:12.87), 4291 msecs / file
There were errors. (0:13.04)

Regards,
Novo

Report message to a moderator

Re: Eigen updated [message #53862 is a reply to message #53861]

Sat, 09 May 2020 20:15

Novo
Messages: 1432
Registered: December 2006

Ultimate Contributor

After full rebuild:
reference/ScatterDraw_Demo
./umk reference ScatterDraw_Demo CLANG -bus
ScatterDraw_Demo.cpp

/home/ssg/.local/soft/bb-worker/worker/upp/build/reference/ScatterDraw_Demo/ScatterDraw_Demo.cpp:16:2: error: reference to 'Vector' is ambiguous
        Vector<Pointf> s1;
        ^
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/Core/Vcont.h:104:7: note: candidate found by name lookup is 'Upp::Vector'
class Vector : public MoveableAndDeepCopyOption< Vector<T> > {
      ^
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/src/Core/Matrix.h:549:1: note: candidate found by name lookup is 'Eigen::Vector'
using Vector = Matrix<Type, Size, 1>;
^
1 error generated.
ScatterDraw_Demo: 1 file(s) built in (0:01.08), 1088 msecs / file
ScatterDraw: 8 file(s) built in (0:08.75), 1094 msecs / file
There were errors. (0:09.92)
program finished with exit code 1

It is still broken ...

Regards,
Novo

Report message to a moderator

Re: Eigen updated [message #53867 is a reply to message #53862]

Sat, 09 May 2020 22:59

Novo
Messages: 1432
Registered: December 2006

Ultimate Contributor

I double-checked that with TheIDE against GIT rev. 1189b35
The problem is reproducible.

Regards,
Novo

Report message to a moderator

Re: Eigen updated [message #53869 is a reply to message #53867]

Sat, 09 May 2020 23:56

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Dear Novo

Yes, there was other problem. I forgot that I have also added VectorXd data access to ScatterDraw, so it should have to be also fully checked. Now it is done and, I think it is finally solved. Thank you!

Best regards
Iñaki

Report message to a moderator

Re: Eigen updated [message #53870 is a reply to message #53869]

Sun, 10 May 2020 01:37

Novo
Messages: 1432
Registered: December 2006

Ultimate Contributor

No problem.
Do not thank me, thank a build automation system Rolling Eyes

Rolling Eyes

Regards,
Novo

Report message to a moderator

Re: Eigen updated [message #53871 is a reply to message #53870]

Sun, 10 May 2020 08:56

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Laughing

From now on I will use it not partially, but massively.

Best regards
Iñaki

Report message to a moderator

Re: Eigen updated [message #53924 is a reply to message #53862]

Fri, 15 May 2020 13:44

coolman
Messages: 119
Registered: April 2006
Location: Czech Republic

Experienced Member

Novo wrote on Sat, 09 May 2020 20:15

After full rebuild:
...

/home/ssg/.local/soft/bb-worker/worker/upp/build/reference/ScatterDraw_Demo/ScatterDraw_Demo.cpp:16:2: error: reference to 'Vector' is ambiguous
        Vector<Pointf> s1;
        ^
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/Core/Vcont.h:104:7: note: candidate found by name lookup is 'Upp::Vector'
class Vector : public MoveableAndDeepCopyOption< Vector<T> > {
      ^
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/src/Core/Matrix.h:549:1: note: candidate found by name lookup is 'Eigen::Vector'
using Vector = Matrix<Type, Size, 1>;
^
1 error generated.
...

It is still broken ...

Hi,

There is similar problem with Array in my personal application. And when you are using SQL the Select is ambiguous too. Build on Ubuntu 18.04 CLANG.

BR, Radek

Report message to a moderator

Re: Eigen updated [message #53930 is a reply to message #53924]

Fri, 15 May 2020 18:25

Didier
Messages: 740
Registered: November 2008
Location: France

Contributor

Hello Radek,

I had the same problem last week : it was my application that had problems.
The problem was the use of 'using namespace Upp' in some header files ... which leads to these ambiguity errors

Maybe it's the same for you

[Updated on: Fri, 15 May 2020 18:26]

Report message to a moderator

Re: Eigen updated [message #53931 is a reply to message #53930]

Fri, 15 May 2020 19:39

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Didier wrote on Fri, 15 May 2020 18:25

Hello Radek,

I had the same problem last week : it was my application that had problems.
The problem was the use of 'using namespace Upp' in some header files ... which leads to these ambiguity errors

Maybe it's the same for you

I agree with Didier. Please try to put #include <ScatterDraw/ScatterDraw.h> before using namespace Upp;, and tell us if this solves the problem.

Best regards
Iñaki

Report message to a moderator

Re: Eigen updated [message #53932 is a reply to message #53931]

Fri, 15 May 2020 20:04

coolman
Messages: 119
Registered: April 2006
Location: Czech Republic

Experienced Member

koldo wrote on Fri, 15 May 2020 19:39

Didier wrote on Fri, 15 May 2020 18:25
Hello Radek,

I had the same problem last week : it was my application that had problems.
The problem was the use of 'using namespace Upp' in some header files ... which leads to these ambiguity errors

Maybe it's the same for you
I agree with Didier. Please try to put #include <ScatterDraw/ScatterDraw.h> before using namespace Upp;, and tell us if this solves the problem.

Hi,

I don't use 'using namespace Upp' in header files. Only in code files after all includes.

BR, Radek

Report message to a moderator

Re: Eigen updated [message #53933 is a reply to message #53932]

Fri, 15 May 2020 22:28

Didier
Messages: 740
Registered: November 2008
Location: France

Contributor

I think it still is the source of you're problem.
In my case I removed ALL the 'using namespace Upp' (in header and in .cpp files) and replaced them with 'namespace Upp { }'

[Updated on: Fri, 15 May 2020 22:29]

Report message to a moderator

Re: Eigen updated [message #53937 is a reply to message #53933]

Sat, 16 May 2020 08:20

coolman
Messages: 119
Registered: April 2006
Location: Czech Republic

Experienced Member

Hi,

Problem was with the include of <ScatterCtrl/ScatterCtrl.h> and <ScatterCtrl/PieCtrl.h>. They were in a wrong place. The 'using namespace Upp;' was correct.

Thanks, Radek

Report message to a moderator

Re: Eigen updated [message #53988 is a reply to message #53851]

Wed, 20 May 2020 00:19

Klugier
Messages: 1118
Registered: September 2012
Location: Poland, Kraków

Senior Contributor

Hello Koldo,

If I may suggest something, I would never use unstable code in the production environment. By introducing nightly build you risk all the problems resulting from this fact. One of that problem is that library can let you down at the most inopportune moment. I think Eigen is one of that libraries with the well establish release cycle and we should respect it. The same thing is true for Upp. You can hit the moment when some important thing doesn't work like we want to. The best example for that is initial GTK 3 port.

Currently, we have 3.3.7 and they are working hard to bring 3.4 or 3.3.8. In my opinion you should be patient and wait for that release. It will happen sooner or later. I highly recommend to revert current master version and back to stable 3.3.7.

Sincerely,
Klugier

U++ - one framework to rule them all.

[Updated on: Wed, 20 May 2020 02:01]

Report message to a moderator

Re: Eigen updated [message #54001 is a reply to message #53988]

Wed, 20 May 2020 12:56

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Dear Klugier

I think you're totally disoriented as to the cause of this problem.

On the one hand, although the problem appeared when updating Eigen, it was actually because of adding a new option, dependent on Eigen, to ScatterDraw.
In other words, this problem would have appeared even if Eigen had not been updated.

Moreover, the update was made immediately after Mirek announced the official release of version 2020.1.

Finally, the problem was minimal and quickly resolved.

Best regards
Iñaki

Report message to a moderator

Re: Eigen updated [message #54009 is a reply to message #54001]

Wed, 20 May 2020 18:42

Novo
Messages: 1432
Registered: December 2006

Ultimate Contributor

I'm sorry to say this, but Eigen_demo is broken again. This time after Mirek's commit.

template<> EIGEN_STRONG_INLINE Packet16b pset1<Packet16b>(const bool&    from) { return _mm_set1_epi8(static_cast<char>(from)); }
                                                                                        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/src/Core/arch/SSE/PacketMath.h:227:93: error: use of undeclared identifier '_mm_castsi128_ps'; did you mean 'Upp::_mm_castsi128_ps'?
template<> EIGEN_STRONG_INLINE Packet4f pset1frombits<Packet4f>(unsigned int from) { return _mm_castsi128_ps(pset1<Packet4i>(from)); }
                                                                                            ^~~~~~~~~~~~~~~~
                                                                                            Upp::_mm_castsi128_ps
/usr/lib/llvm-10/lib/clang/10.0.0/include/emmintrin.h:4930:1: note: 'Upp::_mm_castsi128_ps' declared here
_mm_castsi128_ps(__m128i __a)
^
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen.cpp:2:
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen.h:19:
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/Eigen:1:
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/Dense:1:
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/Core:193:
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/src/Core/arch/SSE/PacketMath.h:227:110: error: cannot initialize a parameter of type 'Upp::__m128i' (vector of 2 'long long' values) with an rvalue of type 'Eigen::internal::Packet4i' (aka 'int')
template<> EIGEN_STRONG_INLINE Packet4f pset1frombits<Packet4f>(unsigned int from) { return _mm_castsi128_ps(pset1<Packet4i>(from)); }
                                                                                                             ^~~~~~~~~~~~~~~~~~~~~
/usr/lib/llvm-10/lib/clang/10.0.0/include/emmintrin.h:4930:26: note: passing argument to parameter '__a' here
_mm_castsi128_ps(__m128i __a)
                         ^
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen.cpp:2:
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen.h:19:
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/Eigen:1:
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/Dense:1:
In file included from /home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/Core:193:
/home/ssg/.local/soft/bb-worker/worker/upp/build/uppsrc/plugin/Eigen/Eigen/src/Core/arch/SSE/PacketMath.h:229:79: error: use of undeclared identifier '_mm_setzero_ps'; did you mean 'Upp::_mm_setzero_ps'?
template<> EIGEN_STRONG_INLINE Packet4f pzero(const Packet4f& /*a*/) { return _mm_setzero_ps(); }
                                                                              ^~~~~~~~~~~~~~
                                                                              Upp::_mm_setzero_ps
/usr/lib/llvm-10/lib/clang/10.0.0/include/xmmintrin.h:1903:1: note: 'Upp::_mm_setzero_ps' declared here
_mm_setzero_ps(void)
^
fatal error: too many errors emitted, stopping now [-ferror-limit=]
20 errors generated.
plugin/Eigen: 1 file(s) built in (0:00.88), 886 msecs / file
Eigen_demo: 3 file(s) built in (0:00.91), 305 msecs / file

Regards,
Novo

Report message to a moderator

Re: Eigen updated [message #54016 is a reply to message #54009]

Thu, 21 May 2020 15:12

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Thank you Novo

Yes, the new version of Core/Ops.h adds an #include <smmintrin.h> that declares __m128i, but in some way smashes the rest of references to __m128i in other include files. And Eigen uses vectorisation.

I am working for hours on it... I will try to solve the problem.
(And this is not because of Eigen's new version Smile

Smile

)

Best regards
Iñaki

Report message to a moderator

Re: Eigen updated [message #54019 is a reply to message #54016]

Thu, 21 May 2020 16:50

Sender Ghost
Messages: 301
Registered: November 2008

Senior Member

Hello.

Please check second variant of the patch in Redmine #2035, which may fix reference/Eigen_demo build.

Report message to a moderator

Re: Eigen updated [message #54020 is a reply to message #54019]

Thu, 21 May 2020 17:48

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Great Sender

With your first patch, now MSVC works in W10 (MinGW and CLANG worked before).
However the problem remains in Ubuntu.
I have added information in your Redmine bug as in Core/Oops.h, in line 395, RGBA should have to be dword.

Best regards
Iñaki

Report message to a moderator

Re: Eigen updated [message #54025 is a reply to message #54020]

Fri, 22 May 2020 09:25

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Dear Sender

Some changes have been included in last version, but not all.
I am overwhelmed, vectorization and intrinsics are out of my knowledge.

It is like if including #include <smmintrin.h> in an include (Core/Blit.h), hides __m128* unions. However I do not know why. Maybe caused by name mangling?
Replacing #ifdef CPU_X86 with #if defined(CPU_X86) && defined(__SSE2__) just avoids the problem by avoiding #include <smmintrin.h>.

Maybe these new options in Core/Blit.h require adding new compiler flags?

Best regards
Iñaki

[Updated on: Fri, 22 May 2020 09:26]

Report message to a moderator

Re: Eigen updated [message #54029 is a reply to message #54025]

Fri, 22 May 2020 10:08

Sender Ghost
Messages: 301
Registered: November 2008

Senior Member

koldo wrote on Fri, 22 May 2020 07:25

Replacing #ifdef CPU_X86 with #if defined(CPU_X86) && defined(__SSE2__) just avoids the problem by avoiding #include <smmintrin.h>

The __SSE2__ define was used for <emmintrin.h> include, because of _mm_set1_epi32 usage. There is a possibility to use <immintrin.h> include for available intrinsics:

#ifdef __SSE2__
#include <emmintrin.h>
#endif

or <x86intrin.h>:

#include <immintrin.h>

But may need to check for related define where related intrinsic(s) used, I guess.

Some examples of how this was done for Eigen:
uppsrc/plugin/Eigen/Eigen/src/Core/util/ConfigureVectorizati on.h:236-238:

#ifdef __SSE4_1__
	#define EIGEN_VECTORIZE_SSE4_1
#endif

uppsrc/plugin/Eigen/Eigen/src/Core/util/ConfigureVectorizati on.h:350-352:

#ifdef EIGEN_VECTORIZE_SSE4_1
	#include <smmintrin.h>
#endif

for OpenGL Mathematics (GLM):
uppsrc/plugin/glm/simd/platform.h:242:

#define GLM_ARCH_SSE2	(GLM_ARCH_SSE2_BIT | GLM_ARCH_SSE)

uppsrc/plugin/glm/simd/platform.h:296-297:

#	elif defined(__SSE2__) || defined(__x86_64__) || defined(_M_X64) || defined(_M_IX86_FP)
# define GLM_ARCH (GLM_ARCH_SSE2)

uppsrc/plugin/glm/simd/platform.h:340-342:

#elif GLM_ARCH & GLM_ARCH_SSE2_BIT
#	include <emmintrin.h>
#endif//GLM_ARCH

Possible to fix Eigen build issue, if place usage of intrinsic(s) (e.g. in uppsrc/Core) inside of C/C++ (*.c/*.cpp) instead of header (*.h) files. Maybe there are other methods.

[Updated on: Fri, 22 May 2020 10:25]

Report message to a moderator

Re: Eigen updated [message #54030 is a reply to message #54029]

Fri, 22 May 2020 10:23

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Quote:

Possible to fix Eigen build issue, if place usage of intrinsic(s) (e.g. in uppsrc/Core) inside of C++ (*.cpp) instead of header (*.h) files. Maybe there are other methods.

Yes, that would solve it all.
The problem of this is that inline functions have to be in .h, and Core/Blit.h includes inline void memsetd().

Best regards
Iñaki

Report message to a moderator

Re: Eigen updated [message #54037 is a reply to message #54030]

Fri, 22 May 2020 11:42

Sender Ghost
Messages: 301
Registered: November 2008

Senior Member

koldo wrote on Fri, 22 May 2020 08:23

Quote:
Possible to fix Eigen build issue, if place usage of intrinsic(s) (e.g. in uppsrc/Core) inside of C++ (*.cpp) instead of header (*.h) files. Maybe there are other methods.
Yes, that would solve it all.
The problem of this is that inline functions have to be in .h, and Core/Blit.h includes inline void memsetd().

Looks like, the Eigen build issue with <smmintrin.h> include was because of its include inside of (Upp) namespace.
I attached newer patch in Redmine #2035. But if there is a need to support CPUs without SSE2 instructions, there is a need to provide other implementations, based on __SSE2__ (or other) check, I guess.

[Updated on: Fri, 22 May 2020 11:54]

Report message to a moderator

Re: Eigen updated [message #54041 is a reply to message #54037]

Fri, 22 May 2020 12:49

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Awesome Sender, you got it!
I have asked Mirek and Tom here to consider your patch.
I owe you one.

Best regards
Iñaki

Report message to a moderator

Re: Eigen updated [message #54071 is a reply to message #54041]

Sat, 30 May 2020 00:10

Novo
Messages: 1432
Registered: December 2006

Ultimate Contributor

This is me again with bad news.
I tried to compile Eigen_demo for Windows using CLANG (actually, I compiled on Linux using wine), and CLANG never fished compilation. I guess this is a bug with CLANG shipped with Upp.
Or it is a bug with Eigen_demo itself ...
Command used for compilation:

wine umk reference Eigen_demo CLANG -bus

Regards,
Novo

Report message to a moderator

Re: Eigen updated [message #54076 is a reply to message #54071]

Sat, 30 May 2020 11:32

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Dear Novo

I cannot reproduce your problem in my W10, using CLANG from U++ and MSVC, both 32 and 64 bits.
Anyway, now it is just updated the very last version, with tiny changes. Please check it just in case, and send the error messages.
Thank you for your support.

Best regards
Iñaki

[Updated on: Sat, 30 May 2020 11:33]

Report message to a moderator

Re: Eigen updated [message #54089 is a reply to message #54076]

Sun, 31 May 2020 07:05

Novo
Messages: 1432
Registered: December 2006

Ultimate Contributor

I was able to compile it in Release configuration.

$ wine umk reference Eigen_demo CLANGx64 -brus
----- Core ( CLANG SHARED BLITZ WIN32 ) (1 / 4)
----- plugin/Eigen ( CLANG SHARED BLITZ WIN32 ) (2 / 4)
----- plugin/z ( CLANG SHARED BLITZ WIN32 ) (3 / 4)
----- Eigen_demo ( MAIN CLANG SHARED BLITZ WIN32 ) (4 / 4)
BLITZ: eigen_demo.cpp non-linear.cpp fft.cpp
Eigen_demo: 3 file(s) built in (1:37.71), 32570 msecs / file
Linking...
Z:\home\ssg\.local\soft\bb-worker\worker\wine-upp\build\.cache\upp.out\CLANGx64.Blitz.Shared\Eigen_demo.exe (2488320 B) linked in (0:00.69)

It takes a minute and a half.
I never saw it finishing compilation in Debug.
TheIde and Clang are from March 3-rd 2020. I guess this is a release.
Tested on Linux with wine.

Regards,
Novo

[Updated on: Sun, 31 May 2020 07:05]

Report message to a moderator

Re: Eigen updated [message #54093 is a reply to message #54089]

Sun, 31 May 2020 20:44

koldo
Messages: 3460
Registered: August 2008

Senior Veteran

Dear Novo

I have never used wine with U++. However, I will try to run it all to try to reproduce your problem.
On the other side, I can promise that it works everyday in tenths of computers with W10.

Best regards
Iñaki

Report message to a moderator

Previous Topic:	[Proposal] Blur algorithm (fast box blur with gaussian approximation)
Next Topic:	Set mouse position

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

PDF

]

Current Time: Sun Jun 14 09:11:38 GMT+2 2026

Total time taken to generate the page: 0.00866 seconds