From don at codesourcery.com Mon Dec 1 19:25:23 2008 From: don at codesourcery.com (Don McCoy) Date: Mon, 01 Dec 2008 12:25:23 -0700 Subject: [vsipl++] [patch] Use new dispatch for reduction functions In-Reply-To: <492B6CB1.1020400@codesourcery.com> References: <492B5DF4.3010103@codesourcery.com> <492B6CB1.1020400@codesourcery.com> Message-ID: <49343A23.6060307@codesourcery.com> Stefan Seefeld wrote: > ... > > Otherwise this looks fine. May I suggest that we now convert the > signatures from "void(T &, A, ...)" to "T(A, ...)" ? This is one of > the syntactic features of the new dispatcher, and it makes the code > easier to understand. > Committed with those minor changes. I'll submit a separate patch for the above suggestion though. -- Don McCoy don (at) CodeSourcery (888) 776-0262 / (650) 331-3385, x712 From gawron at iitis.gliwice.pl Mon Dec 8 19:58:09 2008 From: gawron at iitis.gliwice.pl (Piotr Gawron) Date: Mon, 8 Dec 2008 20:58:09 +0100 Subject: vsipl++ GPL version Message-ID: <3588b19a0812081158p4a19e393xfacbf7ef41795b91@mail.gmail.com> Dear Sir or Madam I'm interesting in development of quantum computation simulation library in vsipl++. Due to academic nature of this project I'm interested in GPL version of your library. I would like to know if GPL version of your library fully supports Cell Broadband Engine architecture. I would really appreciate any information in this subject. Sincerely -- Piotr Gawron From brooks at codesourcery.com Mon Dec 8 21:46:31 2008 From: brooks at codesourcery.com (Brooks Moses) Date: Mon, 08 Dec 2008 13:46:31 -0800 Subject: [vsipl++] vsipl++ GPL version In-Reply-To: <3588b19a0812081158p4a19e393xfacbf7ef41795b91@mail.gmail.com> References: <3588b19a0812081158p4a19e393xfacbf7ef41795b91@mail.gmail.com> Message-ID: <493D95B7.7060502@codesourcery.com> Piotr, Piotr Gawron wrote, at 12/8/2008 11:58 AM: > I'm interesting in development of quantum computation simulation > library in vsipl++. Due to academic nature of this project I'm > interested in GPL version of your library. > > I would like to know if GPL version of your library fully supports > Cell Broadband Engine architecture. Thank you for your interest in Sourcery VSIPL++! At this point in time, Sourcery VSIPL++ for Cell/B.E. requires the Cell Math Library to provide the required backend for full support of the Cell/B.E. architecture. The Cell Math Library is only available in a commercial license; contact sales at codesourcery.com for details on that. However, since the VSIPL++ API (and the Sourcery VSIPL++ implementation of it) are portable across platforms, you could of course use Sourcery VSIPL++ and CML on the Cell/B.E., and still distribute your application source code for use on other platforms where we do have a pure-GPL solution. Also, I believe our sales department has been considering offering low-cost academic licensing for the full Sourcery VSIPL++ and Cell Math Library package. If you're interested in that, please send information about your project and your needs to sales at codesourcery.com. Alternately, you can compile the GPL version of Sourcery VSIPL++ to use the FFTW library as a backend for FFT computations. This will provide access to the Cell/B.E. support included in version 3.2 of FFTW, and will run other computations on the PPUs. Although this isn't full support, it may be useful for you depending on the particular needs of your computations. (I should also note the caveat that our most recent GPL release of Sourcery VSIPL++ was only tested against FFTW 3.1; although we don't expect any problems with compiling it against FFTW 3.2, we obviously can't guarantee that it works perfectly.) Thanks, - Brooks From jules at codesourcery.com Thu Dec 11 14:50:57 2008 From: jules at codesourcery.com (Jules Bergmann) Date: Thu, 11 Dec 2008 09:50:57 -0500 Subject: [patch] Misc changes Message-ID: <494128D1.2010302@codesourcery.com> A couple of minor changes/fixes. Patch applied. -- Jules -- Jules Bergmann CodeSourcery jules at codesourcery.com (650) 331-3385 x705 -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: misc.diff URL: From jules at codesourcery.com Thu Dec 11 14:56:19 2008 From: jules at codesourcery.com (Jules Bergmann) Date: Thu, 11 Dec 2008 09:56:19 -0500 Subject: [patch] Split cholesky tests Message-ID: <49412A13.3010204@codesourcery.com> This patch has been smoldering in my local checkout for some time. It splits the cholesky test up into smaller pieces (which makes compilers happy when real memory is in short supply or virtual memory nonexistant). Patch applied. -- Jules -- Jules Bergmann CodeSourcery jules at codesourcery.com (650) 331-3385 x705 -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: chol.diff URL: From gmendola at mbigroup.it Fri Dec 12 15:53:17 2008 From: gmendola at mbigroup.it (Gaetano Mendola) Date: Fri, 12 Dec 2008 16:53:17 +0100 Subject: hypot function throw specification mismatch Message-ID: <494288ED.1070702@mbigroup.it> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi all, I'm evaluating the use of vsip and until today I have used the version 1.4. Today I did an upgrade to version 2.0 but I'm obtaining the following error at compile time: g++ -DHAVE_CONFIG_H -I. -I/home/kalman/svn_store/FastConvolution/project/linux/kdevelop/Libraries -I.. -I/home/kalman/svn_store/FastConvolution/project/linux/kdevelop/Support -Weffc++ -Wall -pedantic -O2 -g0 -MT fftfactory.lo -MD -MP -MF .deps/fftfactory.Tpo -c /home/kalman/svn_store/FastConvolution/project/linux/kdevelop/Libraries/fftfactory.cpp -fPIC -DPIC -o .libs/fftfactory.o In file included from /usr/local/include/vsip/core/fns_elementwise.hpp:20, from /usr/local/include/vsip/opt/expr/lf_initfini.hpp:28, from /usr/local/include/vsip/core/block_copy.hpp:20, from /usr/local/include/vsip/core/extdata.hpp:23, from /usr/local/include/vsip/core/fast_block.hpp:25, from /usr/local/include/vsip/core/fft/util.hpp:19, from /usr/local/include/vsip/core/fft/factory.hpp:23, from /usr/local/include/vsip/core/fft.hpp:21, from /usr/local/include/vsip/signal.hpp:18, from /home/kalman/svn_store/FastConvolution/project/linux/kdevelop/Libraries/fouriertransformer.h:13, from /home/kalman/svn_store/FastConvolution/project/linux/kdevelop/Libraries/fftfactory.h:11, from /home/kalman/svn_store/FastConvolution/project/linux/kdevelop/Libraries/fftfactory.cpp:6: /usr/local/include/vsip/core/fns_scalar.hpp:32: error: declaration of 'double hypot(double, double)' throws different exceptions /usr/include/bits/mathcalls.h:163: error: from previous declaration 'double hypot(double, double) throw ()' is this a known issue? kalman at gmendola-nb:~$ g++ --version g++ (GCC) 4.2.4 (Ubuntu 4.2.4-1ubuntu3) Copyright (C) 2007 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. kalman at gmendola-nb:~$ cat /etc/issue Ubuntu 8.04.1 \n \l Regards Gaetano Mendola -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJQojt7UpzwH2SGd4RAp1LAKC2XpC8U2zYczMI24R/IXWg4Var5ACguOFW yKpkelU8XopOmU52xROGTJg= =tJRK -----END PGP SIGNATURE----- From stefan at codesourcery.com Fri Dec 12 18:05:39 2008 From: stefan at codesourcery.com (Stefan Seefeld) Date: Fri, 12 Dec 2008 13:05:39 -0500 Subject: [vsipl++] hypot function throw specification mismatch In-Reply-To: <494288ED.1070702@mbigroup.it> References: <494288ED.1070702@mbigroup.it> Message-ID: <4942A7F3.7000601@codesourcery.com> Gaetano, thank you for your interest into Sourcery VSIPL++ ! No, this isn't a known issue, but it was simple enough to identify the cause (a bug in the autoconf script). Please find attached a patch to configure.ac, which should solve the issue you are seeing. Please let me know whether that works. Thanks, Stefan -- Stefan Seefeld CodeSourcery stefan at codesourcery.com (650) 331-3385 x718 -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: configure.ac.diff URL: From gmendola at mbigroup.it Mon Dec 15 11:51:15 2008 From: gmendola at mbigroup.it (Gaetano Mendola) Date: Mon, 15 Dec 2008 12:51:15 +0100 Subject: vsip::Matrix Segmentation Fault (was Re: [vsipl++] hypot function throw specification mismatch ) In-Reply-To: <4942A7F3.7000601@codesourcery.com> References: <494288ED.1070702@mbigroup.it> <4942A7F3.7000601@codesourcery.com> Message-ID: <494644B3.9080708@mbigroup.it> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi Stefan, the patch you provided resolves the compiling problem. This morning running our regression tests using vsip 2.0 (last used was 1.4) I experienced a Segmentation fault. Digging in the problem I was able to isolate the faulty code and I was able to create a distilled application that show the problem. $ cat main.cpp #include int main() { typedef vsip::complex Complex; vsip::Matrix myMatrix(3, 65537); return 0; } $ g++ -g -O0 main.cpp -lsvpp -o test $ ./test Segmentation fault (core dumped) $ gdb ./test core GNU gdb 6.8-debian Copyright (C) 2008 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "i486-linux-gnu"... warning: Can't read pathname for load map: Input/output error. Reading symbols from /usr/lib/libstdc++.so.6...done. Loaded symbols for /usr/lib/libstdc++.so.6 Reading symbols from /lib/tls/i686/cmov/libm.so.6...done. Loaded symbols for /lib/tls/i686/cmov/libm.so.6 Reading symbols from /lib/libgcc_s.so.1...done. Loaded symbols for /lib/libgcc_s.so.1 Reading symbols from /lib/tls/i686/cmov/libc.so.6...done. Loaded symbols for /lib/tls/i686/cmov/libc.so.6 Reading symbols from /lib/ld-linux.so.2...done. Loaded symbols for /lib/ld-linux.so.2 Core was generated by `./test'. Program terminated with signal 11, Segmentation fault. [New process 18042] #0 0x080496d5 in vsip::impl::Memory_pool::allocate > (this=0x0, size=196611) at /usr/local/include/vsip/core/memory_pool.hpp:50 50 { return (T*)(impl_allocate(size * sizeof(T))); } (gdb) bt #0 0x080496d5 in vsip::impl::Memory_pool::allocate > (this=0x0, size=196611) at /usr/local/include/vsip/core/memory_pool.hpp:50 #1 0x080498be in Dense_storage (this=0x804d008, pool=0x0, size=196611, buffer=0x0) at /usr/local/include/vsip/dense.hpp:312 #2 0x08049940 in Dense_impl (this=0x804d008, dom=@0xbff45be4, map=@0xbff45c7c) at /usr/local/include/vsip/dense.hpp:1360 #3 0x080499dd in Dense (this=0x804d008, dom=@0xbff45be4, map=@0xbff45c7c) at /usr/local/include/vsip/dense.hpp:747 #4 0x08049a7d in const_Matrix (this=0xbff45c80, num_rows=3, num_cols=65537, map=@0xbff45c7c) at /usr/local/include/vsip/matrix.hpp:93 #5 0x08049afa in impl_View (this=0xbff45c80, len1=3, len2=65537, map=@0xbff45c7c) at /usr/local/include/vsip/core/view_traits.hpp:316 #6 0x08049b22 in Matrix (this=0xbff45c80, num_rows=3, num_cols=65537, map=@0xbff45c7c) at /usr/local/include/vsip/matrix.hpp:214 #7 0x0804943e in main () at main.cpp:5 if you need I can provide you further informations. Regards Gaetano Mendola Stefan Seefeld wrote: > Gaetano, > > thank you for your interest into Sourcery VSIPL++ ! > > No, this isn't a known issue, but it was simple enough to identify the > cause (a bug in the autoconf script). > > Please find attached a patch to configure.ac, which should solve the > issue you are seeing. Please let me know whether that works. > > Thanks, > Stefan > > > -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJRkSy7UpzwH2SGd4RApPtAKCqaYUTeTuLO6iGVIuuW1s7JjRMtACdEIii vmnipJr++OVGrJTpGjktXKE= =ezyS -----END PGP SIGNATURE----- From stefan at codesourcery.com Mon Dec 15 12:57:07 2008 From: stefan at codesourcery.com (Stefan Seefeld) Date: Mon, 15 Dec 2008 07:57:07 -0500 Subject: [vsipl++] vsip::Matrix Segmentation Fault (was Re: [vsipl++] hypot function throw specification mismatch ) In-Reply-To: <494644B3.9080708@mbigroup.it> References: <494288ED.1070702@mbigroup.it> <4942A7F3.7000601@codesourcery.com> <494644B3.9080708@mbigroup.it> Message-ID: <49465423.7040102@codesourcery.com> Gaetano, Gaetano Mendola wrote: > Hi Stefan, > the patch you provided resolves the compiling problem. That's good to hear. > This morning running our regression tests using vsip 2.0 (last used was > 1.4) I experienced a Segmentation fault. > Digging in the problem I was able to isolate the faulty code and I was > able to create a distilled application that show the problem. > > $ cat main.cpp > > #include > > int main() { int main(int argc, char **argv) { > typedef vsip::complex Complex; You should initialize the library here, using e.g. vsip::vsipl init(argc, argv); > vsip::Matrix myMatrix(3, 65537); Without the library initialization above this may fail because the library's memory allocator isn't set up. > return 0; > } Regards, Stefan -- Stefan Seefeld CodeSourcery stefan at codesourcery.com (650) 331-3385 x718 From gmendola at mbigroup.it Mon Dec 15 15:02:34 2008 From: gmendola at mbigroup.it (Gaetano Mendola) Date: Mon, 15 Dec 2008 16:02:34 +0100 Subject: [vsipl++] vsip::Matrix Segmentation Fault (was Re: [vsipl++] hypot function throw specification mismatch ) In-Reply-To: <49465423.7040102@codesourcery.com> References: <494288ED.1070702@mbigroup.it> <4942A7F3.7000601@codesourcery.com> <494644B3.9080708@mbigroup.it> <49465423.7040102@codesourcery.com> Message-ID: <4946718A.3000907@mbigroup.it> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Thank you, unfortunately our old code didn't initialize the library, the problem is now gone. Regards Gaetano Mendola Stefan Seefeld wrote: > Gaetano, > > Gaetano Mendola wrote: > >> Hi Stefan, >> the patch you provided resolves the compiling problem. > > That's good to hear. > >> This morning running our regression tests using vsip 2.0 (last used was >> 1.4) I experienced a Segmentation fault. >> Digging in the problem I was able to isolate the faulty code and I was >> able to create a distilled application that show the problem. >> >> $ cat main.cpp >> >> #include >> >> int main() { > > int main(int argc, char **argv) { > > >> typedef vsip::complex Complex; > > You should initialize the library here, using e.g. > > vsip::vsipl init(argc, argv); > >> vsip::Matrix myMatrix(3, 65537); > > Without the library initialization above this may fail because the > library's memory allocator isn't set up. > >> return 0; >> } > > Regards, > Stefan > > -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJRnGJ7UpzwH2SGd4RAroGAJ9x2IJzRKnWwCbUbIc2qFujElicVQCfbYl0 LJdIPdncyUyzBtMQUh3RtUM= =PWy2 -----END PGP SIGNATURE----- From don at codesourcery.com Mon Dec 15 21:52:43 2008 From: don at codesourcery.com (Don McCoy) Date: Mon, 15 Dec 2008 14:52:43 -0700 Subject: [patch, committed] Fix cases calling backend directly. Message-ID: <4946D1AB.8000504@codesourcery.com> This patch corrects two benchmarks where they call an evaluator directly rather than going through the dispatcher. Committed as obvious. -- Don McCoy don (at) CodeSourcery (888) 776-0262 / (650) 331-3385, x712 -------------- next part -------------- A non-text attachment was scrubbed... Name: eval.diff Type: text/x-diff Size: 2671 bytes Desc: not available URL: From gmendola at mbigroup.it Tue Dec 16 11:06:48 2008 From: gmendola at mbigroup.it (Gaetano Mendola) Date: Tue, 16 Dec 2008 12:06:48 +0100 Subject: Configure syntax error ( was Re: [vsipl++] hypot function throw specification mismatch ) In-Reply-To: <4942A7F3.7000601@codesourcery.com> References: <494288ED.1070702@mbigroup.it> <4942A7F3.7000601@codesourcery.com> Message-ID: <49478BC8.9060000@mbigroup.it> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 As I already told you the patch works, however applying that patch and building vsipl on another linux distribution/architecture I'm getting error around the end of compile phase (it goes all fine without patch). $ ./configure [...] Configure goes well. $ make [...] configure: fftw3 config options: --disable-dependency-tracking --silent --disable-fortran --host=i686-pc-linux-gnu --build=i686-pc-linux-gnu . =============================================================== configure: Configuring fftw3f (float). configure: extra config options: '--enable-sse'. appending configuration tag "CXX" to libtool appending configuration tag "F77" to libtool configure: Configuring fftw3 (double). configure: extra config options: '--enable-sse2'. appending configuration tag "CXX" to libtool appending configuration tag "F77" to libtool configure: Configuring fftw3l (long double). configure: extra config options: ''. appending configuration tag "CXX" to libtool appending configuration tag "F77" to libtool =============================================================== Built-in FFTW3 configures successful. configure: Will link with -lfftw3f -lfftw3 -lfftw3l. ./configure: line 11190: syntax error near unexpected token `PAS,' ./configure: line 11190: ` PKG_CHECK_MODULES(PAS, pas)' make: *** [config.status] Error 2 Regards Gaetano Mendola Stefan Seefeld wrote: > Gaetano, > > thank you for your interest into Sourcery VSIPL++ ! > > No, this isn't a known issue, but it was simple enough to identify the > cause (a bug in the autoconf script). > > Please find attached a patch to configure.ac, which should solve the > issue you are seeing. Please let me know whether that works. > > Thanks, > Stefan > > > -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJR4vI7UpzwH2SGd4RAvjxAKDDkqRAWZtphAreQ2kV1eGaLJjQ+QCcDSXQ A3miNaESgdicobWP8qkSF2k= =UfU7 -----END PGP SIGNATURE----- From stefan at codesourcery.com Tue Dec 16 13:01:31 2008 From: stefan at codesourcery.com (Stefan Seefeld) Date: Tue, 16 Dec 2008 08:01:31 -0500 Subject: [vsipl++] Configure syntax error ( was Re: [vsipl++] hypot function throw specification mismatch ) In-Reply-To: <49478BC8.9060000@mbigroup.it> References: <494288ED.1070702@mbigroup.it> <4942A7F3.7000601@codesourcery.com> <49478BC8.9060000@mbigroup.it> Message-ID: <4947A6AB.5040405@codesourcery.com> Gaetano Mendola wrote: > As I already told you the patch works, however applying that patch and building > vsipl on another linux distribution/architecture I'm getting error around the > end of compile phase (it goes all fine without patch). [...] > ./configure: line 11190: syntax error near unexpected token `PAS,' > ./configure: line 11190: ` PKG_CHECK_MODULES(PAS, pas)' > make: *** [config.status] Error 2 The error is not related to the patch itself, but to the fact that this triggered the regeneration of configure from configure.ac, which requires tools that you may not have installed. In particular, the PKG_CHECK_MODULES macro definition comes from the pkg-config tool. Regards, Stefan -- Stefan Seefeld CodeSourcery stefan at codesourcery.com (650) 331-3385 x718 From jules at codesourcery.com Tue Dec 16 15:57:53 2008 From: jules at codesourcery.com (Jules Bergmann) Date: Tue, 16 Dec 2008 10:57:53 -0500 Subject: [vsipl++] Configure syntax error ( was Re: [vsipl++] hypot function throw specification mismatch ) In-Reply-To: <4947A6AB.5040405@codesourcery.com> References: <494288ED.1070702@mbigroup.it> <4942A7F3.7000601@codesourcery.com> <49478BC8.9060000@mbigroup.it> <4947A6AB.5040405@codesourcery.com> Message-ID: <4947D001.1030802@codesourcery.com> Stefan, Can you send Gaetano a generated configure file? -- Jules Stefan Seefeld wrote: > Gaetano Mendola wrote: > >> As I already told you the patch works, however applying that patch and >> building >> vsipl on another linux distribution/architecture I'm getting error >> around the >> end of compile phase (it goes all fine without patch). > > [...] > > >> ./configure: line 11190: syntax error near unexpected token `PAS,' >> ./configure: line 11190: ` PKG_CHECK_MODULES(PAS, pas)' >> make: *** [config.status] Error 2 > > The error is not related to the patch itself, but to the fact that this > triggered the regeneration of configure from configure.ac, which > requires tools that you may not have installed. In particular, the > PKG_CHECK_MODULES macro definition comes from the pkg-config tool. > > Regards, > Stefan > > -- Jules Bergmann CodeSourcery jules at codesourcery.com (650) 331-3385 x705 From stefan at codesourcery.com Tue Dec 16 16:01:20 2008 From: stefan at codesourcery.com (Stefan Seefeld) Date: Tue, 16 Dec 2008 11:01:20 -0500 Subject: [vsipl++] Configure syntax error ( was Re: [vsipl++] hypot function throw specification mismatch ) In-Reply-To: <4947D001.1030802@codesourcery.com> References: <494288ED.1070702@mbigroup.it> <4942A7F3.7000601@codesourcery.com> <49478BC8.9060000@mbigroup.it> <4947A6AB.5040405@codesourcery.com> <4947D001.1030802@codesourcery.com> Message-ID: <4947D0D0.6010801@codesourcery.com> Jules Bergmann wrote: > Stefan, > > Can you send Gaetano a generated configure file? Sure, here it is. Regards, Stefan -- Stefan Seefeld CodeSourcery stefan at codesourcery.com (650) 331-3385 x718 -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: configure URL: From gmendola at mbigroup.it Tue Dec 16 17:46:09 2008 From: gmendola at mbigroup.it (Gaetano Mendola) Date: Tue, 16 Dec 2008 18:46:09 +0100 Subject: [vsipl++] Configure syntax error ( was Re: [vsipl++] hypot function throw specification mismatch ) In-Reply-To: <4947A6AB.5040405@codesourcery.com> References: <494288ED.1070702@mbigroup.it> <4942A7F3.7000601@codesourcery.com> <49478BC8.9060000@mbigroup.it> <4947A6AB.5040405@codesourcery.com> Message-ID: <4947E961.1040806@mbigroup.it> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Stefan Seefeld wrote: > Gaetano Mendola wrote: > >> As I already told you the patch works, however applying that patch and >> building >> vsipl on another linux distribution/architecture I'm getting error >> around the >> end of compile phase (it goes all fine without patch). > > [...] > > >> ./configure: line 11190: syntax error near unexpected token `PAS,' >> ./configure: line 11190: ` PKG_CHECK_MODULES(PAS, pas)' >> make: *** [config.status] Error 2 > > The error is not related to the patch itself, but to the fact that this > triggered the regeneration of configure from configure.ac, which > requires tools that you may not have installed. In particular, the > PKG_CHECK_MODULES macro definition comes from the pkg-config tool. Sorry to my late reply, you are correct, I installed that package and the issue is gone. Gaetano -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJR+lh7UpzwH2SGd4RAtIBAJ4wuFFMKsP1yXhn0ob3T4K3b3LSPQCg06DR miLLXx/Snn9X1JtyjkRExqA= =7uds -----END PGP SIGNATURE----- From gmendola at mbigroup.it Tue Dec 16 17:54:33 2008 From: gmendola at mbigroup.it (Gaetano Mendola) Date: Tue, 16 Dec 2008 18:54:33 +0100 Subject: [vsipl++] Configure syntax error ( was Re: [vsipl++] hypot function throw specification mismatch ) In-Reply-To: <4947D0D0.6010801@codesourcery.com> References: <494288ED.1070702@mbigroup.it> <4942A7F3.7000601@codesourcery.com> <49478BC8.9060000@mbigroup.it> <4947A6AB.5040405@codesourcery.com> <4947D001.1030802@codesourcery.com> <4947D0D0.6010801@codesourcery.com> Message-ID: <4947EB59.8010002@mbigroup.it> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Thank you Stefan, I had to reply sooner to your suggestion to install that package, in order to avoid you that work. Unfortunately I had a Xeon based computer installed with a 32bit OS and I had a degradation of performance (half than it was used to be) and I was all the day long one step away from having an heart attack and I spent almost all the day to first found the problem and then reinstall the OS. Regards Gaetano Mendola Stefan Seefeld wrote: > Jules Bergmann wrote: >> Stefan, >> >> Can you send Gaetano a generated configure file? > > Sure, here it is. > > Regards, > Stefan > -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFJR+tZ7UpzwH2SGd4RAuoEAJ9IX88bnO7/NL92N5pY6JqfTAD2nwCfUtPt P1fWXNBoyyOqGMPmYugewl0= =q8VI -----END PGP SIGNATURE----- From jules at codesourcery.com Wed Dec 24 13:39:13 2008 From: jules at codesourcery.com (Jules Bergmann) Date: Wed, 24 Dec 2008 08:39:13 -0500 Subject: [patch] Fix stride for CBE float vmul Message-ID: <49523B81.4000204@codesourcery.com> Stride is passed down in elements, not bytes! Patch applied. -- jules -- Jules Bergmann CodeSourcery jules at codesourcery.com (650) 331-3385 x705 -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: vmul.diff URL: From don at codesourcery.com Wed Dec 24 19:24:06 2008 From: don at codesourcery.com (Don McCoy) Date: Wed, 24 Dec 2008 12:24:06 -0700 Subject: [vsipl++] [patch] Fix stride for CBE float vmul In-Reply-To: <49523B81.4000204@codesourcery.com> References: <49523B81.4000204@codesourcery.com> Message-ID: <49528C56.3060609@codesourcery.com> Jules Bergmann wrote: > Stride is passed down in elements, not bytes! > > This looks good, but I had a couple of minor questions/comments. > Index: tests/regressions/large_vmul.cpp > =================================================================== > --- tests/regressions/large_vmul.cpp (revision 230289) > +++ tests/regressions/large_vmul.cpp (working copy) > @@ -46,8 +46,8 @@ > Vector Z(size); > > Rand gen(0, 0); > - A = gen.randu(size); > - B = gen.randu(size); > +// A = gen.randu(size); > +// B = gen.randu(size); > Did you mean to leave this commented out? > > @@ -61,7 +61,7 @@ > std::cout << "A(i) * B(i) = " << A(i) * B(i) << std::endl; > } > #endif > - test_assert(almost_equal(Z(i), A(i) * B(i))); > + test_assert(almost_equal(Z.get(i), A(i) * B(i))); > For consistency, should we use .get for all accesses (or drop it)? I guess for tests we ought to always use it, in my opinion. -- Don McCoy don (at) CodeSourcery (888) 776-0262 / (650) 331-3385, x712 From jules at codesourcery.com Wed Dec 24 19:28:16 2008 From: jules at codesourcery.com (Jules Bergmann) Date: Wed, 24 Dec 2008 14:28:16 -0500 Subject: [vsipl++] [patch] Fix stride for CBE float vmul In-Reply-To: <49528C56.3060609@codesourcery.com> References: <49523B81.4000204@codesourcery.com> <49528C56.3060609@codesourcery.com> Message-ID: <49528D50.2040004@codesourcery.com> Don McCoy wrote: > Jules Bergmann wrote: >> Stride is passed down in elements, not bytes! >> >> > This looks good, but I had a couple of minor questions/comments. Don, Thanks for looking over this! >> Rand gen(0, 0); >> - A = gen.randu(size); >> - B = gen.randu(size); >> +// A = gen.randu(size); >> +// B = gen.randu(size); >> > Did you mean to leave this commented out? Good catch! No, I didn't mean to leave this in, but I made a second silent checkin afterwards that fixed it. > >> >> @@ -61,7 +61,7 @@ >> std::cout << "A(i) * B(i) = " << A(i) * B(i) << std::endl; >> } >> #endif >> - test_assert(almost_equal(Z(i), A(i) * B(i))); >> + test_assert(almost_equal(Z.get(i), A(i) * B(i))); >> > For consistency, should we use .get for all accesses (or drop it)? I > guess for tests we ought to always use it, in my opinion. Using get() is probably best. Here it is a correctness issue. If A is split complex, A(i) returns an Lvalue_proxy, and no operator*() can be found to handle it. Using A.get(i) returns a complex, and all is fine. -- Jules -- Jules Bergmann CodeSourcery jules at codesourcery.com (650) 331-3385 x705 From jules at codesourcery.com Wed Dec 24 20:10:03 2008 From: jules at codesourcery.com (Jules Bergmann) Date: Wed, 24 Dec 2008 15:10:03 -0500 Subject: [vsipl++] [patch] Fix stride for CBE float vmul In-Reply-To: <49528D50.2040004@codesourcery.com> References: <49523B81.4000204@codesourcery.com> <49528C56.3060609@codesourcery.com> <49528D50.2040004@codesourcery.com> Message-ID: <4952971B.6060608@codesourcery.com> > Here it is a correctness issue. If A is split complex, A(i) returns an > Lvalue_proxy, and no operator*() can be found to handle it. Using > A.get(i) returns a complex, and all is fine. Oops! Scratch that answer, Stefan reports that Fast_blocks (which I'm using) don't have operator() enabled. -- Jules Bergmann CodeSourcery jules at codesourcery.com (650) 331-3385 x705