<?xml version="1.0" encoding="utf-8"?>
<feed xml:lang="en" xmlns="http://www.w3.org/2005/Atom"><title>Activity for ViennaCL</title><link href="https://sourceforge.net/p/viennacl/activity/" rel="alternate"/><id>https://sourceforge.net/p/viennacl/activity/</id><updated>2024-06-17T07:10:50.857000Z</updated><subtitle>Recent activity for ViennaCL</subtitle><entry><title>Marcin Błażejowski posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/9a966b9881/?limit=25#3e0d" rel="alternate"/><published>2024-06-17T07:10:50.857000Z</published><updated>2024-06-17T07:10:50.857000Z</updated><author><name>Marcin Błażejowski</name><uri>https://sourceforge.net/u/marcingretl/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/9a966b9881/?limit=25#3e0d</id><summary type="html">Hi, I'd like to ask about the current projects' status: is it still developed or already abandoned? Cheers, Marcin</summary></entry><entry><title>Yucheng posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/3c35cc783c/?limit=25#2da3" rel="alternate"/><published>2020-10-12T19:22:54.479000Z</published><updated>2020-10-12T19:22:54.479000Z</updated><author><name>Yucheng</name><uri>https://sourceforge.net/u/yuchengs/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/3c35cc783c/?limit=25#2da3</id><summary type="html">I have the same program running on both Tesla V100 and GTX1080. However, it seems Tesla V100 is significantly slower than GTX1080 on solving sparse linear systems (bicgstab with viennacl::linalg::chow_patel_ilu_precond) I think the problem is the library is not tuned for Tesla V100 according to its number of SM. Can anyone confirm this?</summary></entry><entry><title>p ding posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/617fcaaaa6/?limit=25#5e63" rel="alternate"/><published>2020-08-22T23:41:06.156000Z</published><updated>2020-08-22T23:41:06.156000Z</updated><author><name>p ding</name><uri>https://sourceforge.net/u/ztdep/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/617fcaaaa6/?limit=25#5e63</id><summary type="html">Thanks. but which command should i use to select corresponding GPU devecies?</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/617fcaaaa6/?limit=25#c3d5" rel="alternate"/><published>2020-08-22T14:28:33.098000Z</published><updated>2020-08-22T14:28:33.098000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/617fcaaaa6/?limit=25#c3d5</id><summary type="html">You'll have to do the data copying manually through the host via viennacl::copy (for fast_copy) and switch between the decives accordingly.</summary></entry><entry><title>p ding posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/617fcaaaa6/?limit=25#8871" rel="alternate"/><published>2020-08-22T01:19:52.640000Z</published><updated>2020-08-22T01:19:52.640000Z</updated><author><name>p ding</name><uri>https://sourceforge.net/u/ztdep/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/617fcaaaa6/?limit=25#8871</id><summary type="html">If i have 2 GPU cards. how to exchange data with ViennaCL. Does ViennalCL support 2 GPU cards default.</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#2afe" rel="alternate"/><published>2020-03-28T10:02:04.474000Z</published><updated>2020-03-28T10:02:04.474000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#2afe</id><summary type="html">Hi carl. It is a great news. Let me sat thanks for your work. It really great. Actually I replaced unsigned int occurrences in compressed_matrix.hpp, sparse_matrix_hpp and in many places for couple of iterative solvers and now it seems to work fine eve for system of: rows = 210 mlns, columns = 210mls and with about 50 nonzeros in each row. I tested it in my fem geomechanics and hydrodynaics code, seems to work fine. But of course, I have not capabilities to do it for each preconditioner and each...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#02d5" rel="alternate"/><published>2020-03-28T03:58:42.284000Z</published><updated>2020-03-28T03:58:42.284000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#02d5</id><summary type="html">Hi Dmitriy, I'll give the conversion to 64-bit integers a shot in a separate branch in the developer repository. It will take a day or two until I get to it, though. Best regards, Karli</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#4ed6" rel="alternate"/><published>2020-03-25T05:35:53.598000Z</published><updated>2020-03-25T05:35:53.598000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#4ed6</id><summary type="html">Ah, so the suspected 32-bit overflow. :-/ If you are on Windows, you should attempt a 64-bit build in order to get 64-bit long. On Linux and MacOS 'long' is usually 64 bits anyway. It won't help with the unsigned int overflow, though. There is no easy way to get around the unsigned int issue without making a significant impact on the overall code. It's certainly doable, but will take some time. ViennaCL had primarily GPUs in mind, for which memory is limited to a few GB, thus the GPU runs out of...</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#cc9b" rel="alternate"/><published>2020-03-25T03:59:28.962000Z</published><updated>2020-03-25T03:59:28.962000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#cc9b</id><summary type="html">So, as I see this problems really connected with very large indexes. In file: cpu_ram.hpp (line 94) My size_in_bytes was about 2.3e9. After classical cast: long(size_in_bytes) it becomes negative and data don't copy. So I replaced (long) to (long long) and everything is ok. But are there any good way to work with indexes larger that (unsigned int)? I mean if I have number of nonzeros &gt; 4.6e9? Of course, I can rewrite function copy and a lot of othwer functions in compressed_matrix.hpp to replace...</summary></entry><entry><title>Dmitriy modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#9684" rel="alternate"/><published>2020-03-24T23:17:37.576000Z</published><updated>2020-03-24T23:17:37.576000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#9684</id><summary type="html">I had long time of debugging, and saw unsigned_int in many functions of viennacl::compressed_matrix. For example in function copy type of row_jumper and col_buffer must be unsigned int as I see. Does it mean that I have a limitation on number of nonzeros connected with unsigned int maximum? The error with copy ocure much before such sizes, but I can't understand if I have such a limit.</summary></entry><entry><title>Dmitriy modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#9684" rel="alternate"/><published>2020-03-24T23:16:17.976000Z</published><updated>2020-03-24T23:16:17.976000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#9684</id><summary type="html">I had long time of debugging, and saw unsigned_int in many functions of viennacl::compressed_matrix. For example in function copy type of row_jumper and col_buffer must be unsigned int as I see. Does it mean that I have a limitation on number of nonzeros connected with unsigned int maximum?</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#9684" rel="alternate"/><published>2020-03-24T23:15:29.092000Z</published><updated>2020-03-24T23:15:29.092000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#9684</id><summary type="html">I had long time of debugging, and saw unsigned_int in many functions of viennacl::compressed_matrix. For example in function copy type of row_jumper and col_buffer must be unsigned int as I see. Does it mean that I hacve a limitation on number of nonzeros connected with unsigned int maximum?</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#ae0a" rel="alternate"/><published>2020-03-24T12:09:28.432000Z</published><updated>2020-03-24T12:09:28.432000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#ae0a</id><summary type="html">Hi Karl, I now have tested it for compressed matrix with 6 millions rows and colums. If it has about 50 nonzeros in each row, viennacl::copy doesn't work. But if it has only 30 or lesss nonzeros in each row, everything is good. No, there are no problems with lack of memory. I am testing it on 64 gb machine and also on NUMA workstation with 1TB. I found such a problem during calculation my FEM problems. If I doing: solve(A, B, tag, precond) with A, B, precond - ublas objects, everything is good. But...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#e9f2" rel="alternate"/><published>2020-03-24T11:27:58.294000Z</published><updated>2020-03-24T11:27:58.294000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#e9f2</id><summary type="html">Hi Dmitriy, are you referring to 10 million nonzeros, or 10 million rows and columns? Can you provide the matrix in matrix market format in order to reproduce the problem? Maybe you run out of memory? Best regards, Karli</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#e5b6" rel="alternate"/><published>2020-03-24T11:09:16.663000Z</published><updated>2020-03-24T11:09:16.663000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/130f02a39c/?limit=25#e5b6</id><summary type="html">I have an error during copy from ublas::compressed_matrix to viennaCL::compressed_matrix It work good for matrices with nonzeros about 500000, but if my ublas matrix is large, for example 10 millions, the result viennaCL::compressed matrix is empty, has all zeros as non-zero values. I use OpenMP on CPU. If anybody faced with such a problem?</summary></entry><entry><title>Wen-Yin Wei posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/37b0de47/?limit=25#c0ac" rel="alternate"/><published>2019-11-18T00:58:19.274000Z</published><updated>2019-11-18T00:58:19.274000Z</updated><author><name>Wen-Yin Wei</name><uri>https://sourceforge.net/u/wenyin/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/37b0de47/?limit=25#c0ac</id><summary type="html">Hi Karl, I am planning to implement a realtime image-processing system which would handle video streams from several cameras. And I want to make it cross-vendor (Nvidia or AMD) without loss of efficiency. ViennaCL is very elegant and self-contained, so that I can not help making it a CV package from it. Now I am utilizing ImageMagcik++ as an external image package to help me with handling reading and writing images. On the base of that , I select different channels to be put in different matrix (R/G/B)....</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/37b0de47/?limit=25#716b" rel="alternate"/><published>2019-11-17T16:54:26.539000Z</published><updated>2019-11-17T16:54:26.539000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/37b0de47/?limit=25#716b</id><summary type="html">Hi Wenyin, no, image types are still not supported and are currently not on the road map. What's your use case? Best regards, Karli</summary></entry><entry><title>Wen-Yin Wei posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/37b0de47/?limit=25#c394" rel="alternate"/><published>2019-11-16T09:14:39.595000Z</published><updated>2019-11-16T09:14:39.595000Z</updated><author><name>Wen-Yin Wei</name><uri>https://sourceforge.net/u/wenyin/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/37b0de47/?limit=25#c394</id><summary type="html">Hello Karl, Did Vienna team make any progress on image processing these years? It worth a new package. However, as long as the basic input and output image functionality is equipped, users could utilize it to establish what they want. Best regards, Wenyin</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/a556e1bd58/?limit=25#6fdc" rel="alternate"/><published>2019-11-15T09:02:09.017000Z</published><updated>2019-11-15T09:02:09.017000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/a556e1bd58/?limit=25#6fdc</id><summary type="html">Hi, if you want to set up a CSR matrix on the GPU, then you need to populate the CSR arrays yourself (this depends a lot on the type of application you have; in some cases this is fairly straight-forward, in other cases it is more involved). Once you have the three arrays, you can pass them to a viennacl::compressed_matrix&lt;t&gt; by using the .set() member function.&lt;/t&gt; Best regards, Karli</summary></entry><entry><title>p ding posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/a556e1bd58/?limit=25#d63c" rel="alternate"/><published>2019-11-15T06:15:04.845000Z</published><updated>2019-11-15T06:15:04.845000Z</updated><author><name>p ding</name><uri>https://sourceforge.net/u/ztdep/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/a556e1bd58/?limit=25#d63c</id><summary type="html">Dear sir: Ii want to setup the csr matrix on GPU, which API can i use. Regards</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/f636b5a1df/?limit=25#ac69" rel="alternate"/><published>2019-10-31T06:44:35.388000Z</published><updated>2019-10-31T06:44:35.388000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/f636b5a1df/?limit=25#ac69</id><summary type="html">Hi Dmitriy, you can use the monitor functions for this. There are dedicated solver classes with set_monitor-routines available, e.g. https://github.com/viennacl/viennacl-dev/blob/master/viennacl/linalg/bicgstab.hpp#L537 See also iterative-custom.cpp as an example: https://github.com/viennacl/viennacl-dev/blob/master/examples/tutorial/iterative-custom.cpp Best regards, Karli</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/f636b5a1df/?limit=25#e642" rel="alternate"/><published>2019-10-31T05:07:43.242000Z</published><updated>2019-10-31T05:07:43.242000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/f636b5a1df/?limit=25#e642</id><summary type="html">Hello. Are there any way to get current iteration number or current iteration residual for parallel iterative Solvers during calculation. I have large linear system for FEM and sometimes I need to get this information to show progress to user.</summary></entry><entry><title>p ding posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/e2ef742511/?limit=25#41f8" rel="alternate"/><published>2019-04-12T10:09:13.179000Z</published><updated>2019-04-12T10:09:13.179000Z</updated><author><name>p ding</name><uri>https://sourceforge.net/u/ztdep/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/e2ef742511/?limit=25#41f8</id><summary type="html">Thanks , it works for my problems.</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/e2ef742511/?limit=25#752a" rel="alternate"/><published>2019-04-11T14:05:16.677000Z</published><updated>2019-04-11T14:05:16.677000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/e2ef742511/?limit=25#752a</id><summary type="html">Hi p ding, Which backend are you targeting? In either case, you can wrap your existing buffers by a viennacl::vector&lt;&gt;, which would be a zero-copy operation :-) Here is an example on how to wrap a user-provided CUDA buffer: https://github.com/viennacl/viennacl-dev/blob/master/examples/tutorial/custom-cuda.cu (this can be done similarly for OpenCL and OpenMP)</summary></entry><entry><title>p ding posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/e2ef742511/?limit=25#7381" rel="alternate"/><published>2019-04-11T12:06:31.661000Z</published><updated>2019-04-11T12:06:31.661000Z</updated><author><name>p ding</name><uri>https://sourceforge.net/u/ztdep/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/e2ef742511/?limit=25#7381</id><summary type="html">We can initilize the device array with many threads, but it seems this is not supproted by ViennaCL. So i want to copy the device array to a viennacl vector. how to resovle this.</summary></entry><entry><title>Doug Schouten posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/bebebdfc/?limit=25#18fd" rel="alternate"/><published>2019-03-21T18:32:36.450000Z</published><updated>2019-03-21T18:32:36.450000Z</updated><author><name>Doug Schouten</name><uri>https://sourceforge.net/u/dougschouten/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/bebebdfc/?limit=25#18fd</id><summary type="html">Hi ... sorry for bumping a very old post. I can't find an update on the status of SVD in ViennaCL except a brief mention in the change log. Can someone comment on the current (circa 2018) useability of the SVD code in ViennaCL? Thanks!</summary></entry><entry><title>venkatesh karthick modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#88ed" rel="alternate"/><published>2019-02-26T07:25:14.429000Z</published><updated>2019-02-26T07:25:14.429000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#88ed</id><summary type="html">Hi Sir, @ Karl Rupp Thanks a lot I finally got the output correctly after applying all your instructions. Changed my code like this code int main() { typedef float ScalarType; viennacl::tools::timer timer; double exec_timecpu; double exec_timegpu; viennacl::tools::uniform_random_numbers&lt;ScalarType&gt; randomNumber; viennacl::ocl::set_context_platform_index(1, 1); viennacl::ocl::set_context_platform_index(0, 0); viennacl::ocl::switch_context(1); viennacl::matrix&lt;ScalarType&gt; vcl_A(BLAS3_MATRIX_SIZE, BLAS3_MATRIX_SIZE);...</summary></entry><entry><title>venkatesh karthick modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#88ed" rel="alternate"/><published>2019-02-26T05:04:42.223000Z</published><updated>2019-02-26T05:04:42.223000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#88ed</id><summary type="html">Hi Sir, @ Karl Rupp Thanks a lot I finally got the output correctly after applying all your instructions. Changed my code like this code int main() { typedef float ScalarType; viennacl::tools::timer timer; double exec_timecpu; double exec_timegpu; viennacl::tools::uniform_random_numbers&lt;ScalarType&gt; randomNumber; viennacl::ocl::set_context_platform_index(1, 1); viennacl::ocl::set_context_platform_index(0, 0); viennacl::ocl::switch_context(1); viennacl::matrix&lt;ScalarType&gt; vcl_A(BLAS3_MATRIX_SIZE, BLAS3_MATRIX_SIZE);...</summary></entry><entry><title>venkatesh karthick modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#88ed" rel="alternate"/><published>2019-02-26T05:03:43.096000Z</published><updated>2019-02-26T05:03:43.096000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#88ed</id><summary type="html">Hi Sir, @ Karl Rupp Thanks a lot I finally got the output correctly after applying all your instructions. Changed my code like this code ~~~ int main() { typedef float ScalarType; viennacl::tools::timer timer; double exec_timecpu; double exec_timegpu; viennacl::tools::uniform_random_numbers&lt;ScalarType&gt; randomNumber; viennacl::ocl::set_context_platform_index(1, 1); viennacl::ocl::set_context_platform_index(0, 0); viennacl::ocl::switch_context(1); viennacl::matrix&lt;ScalarType&gt; vcl_A(BLAS3_MATRIX_SIZE,...</summary></entry><entry><title>venkatesh karthick modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#88ed" rel="alternate"/><published>2019-02-26T05:02:55.549000Z</published><updated>2019-02-26T05:02:55.549000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#88ed</id><summary type="html">Hi Sir, @ Karl Rupp Thanks a lot I finally got the output correctly after applying all your instructions. I changed my code like this : code ~~~ int main() { typedef float ScalarType; viennacl::tools::timer timer; double exec_timecpu; double exec_timegpu; viennacl::tools::uniform_random_numbers&lt;ScalarType&gt; randomNumber; viennacl::ocl::set_context_platform_index(1, 1); viennacl::ocl::set_context_platform_index(0, 0); viennacl::ocl::switch_context(1); viennacl::matrix&lt;ScalarType&gt; vcl_A(BLAS3_MATRIX_SIZE,...</summary></entry><entry><title>venkatesh karthick posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#88ed" rel="alternate"/><published>2019-02-26T05:01:41.714000Z</published><updated>2019-02-26T05:01:41.714000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#88ed</id><summary type="html">Hi Sir, @ Karl Rupp Thanks a lot I finally got the output correctly after applying all your instructions. I changed my code like this : **code** ~~~ int main() { typedef float ScalarType; viennacl::tools::timer timer; double exec_timecpu; double exec_timegpu; viennacl::tools::uniform_random_numbers&lt;ScalarType&gt; randomNumber; viennacl::ocl::set_context_platform_index(1, 1); viennacl::ocl::set_context_platform_index(0, 0); viennacl::ocl::switch_context(1); viennacl::matrix&lt;ScalarType&gt; vcl_A(BLAS3_MATRIX_SIZE,...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#c15e" rel="alternate"/><published>2019-02-25T16:21:57.288000Z</published><updated>2019-02-25T16:21:57.288000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#c15e</id><summary type="html">Hi, several things to note: directly assigning values via operator= is VERY slow. Use viennacl::copy() instead. make sure VIENNACL_WITH_OPENCL is defined. This is most likely the case, since you obtain the GPU name as output. the first execution of a particular kernel may involve compilation of the OpenCL kernels. Thus, you should do something like the following: vcl_GC = viennacl::linalg::prod(vcl_GA, vcl_GB); timer.start(); vcl_GC = viennacl::linalg::prod(vcl_GA, vcl_GB); viennacl::backend::finish();...</summary></entry><entry><title>venkatesh karthick posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#18c9" rel="alternate"/><published>2019-02-25T08:18:19.794000Z</published><updated>2019-02-25T08:18:19.794000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#18c9</id><summary type="html">Hi sir @ Karl Rupp I created the new buffers in the new context but even after changing the thing still getting the same execution time . Code int main() { typedef float ScalarType; viennacl::tools::timer timer; double exec_timecpu; double exec_timegpu; viennacl::tools::uniform_random_numbers&lt;ScalarType&gt; randomNumber; viennacl::ocl::set_context_platform_index(1, 1); viennacl::ocl::set_context_platform_index(0, 0); viennacl::ocl::switch_context(1); viennacl::matrix&lt;ScalarType&gt; vcl_A(BLAS3_MATRIX_SIZE,...</summary></entry><entry><title>venkatesh karthick modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#2262" rel="alternate"/><published>2019-02-25T05:04:09.875000Z</published><updated>2019-02-25T05:04:09.875000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#2262</id><summary type="html">Hi sir @ Karl Rupp I created the new buffers in the new context but even after changing the thing still getting the same execution time . Code int main() { typedef float ScalarType; viennacl::tools::timer timer; double exec_timecpu; double exec_timegpu; viennacl::tools::uniform_random_numbers&lt;ScalarType&gt; randomNumber; viennacl::ocl::set_context_platform_index(1, 1); viennacl::ocl::set_context_platform_index(0, 0); viennacl::ocl::switch_context(1); viennacl::matrix&lt;ScalarType&gt; vcl_A(BLAS3_MATRIX_SIZE,...</summary></entry><entry><title>venkatesh karthick modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#2262" rel="alternate"/><published>2019-02-25T05:02:30.022000Z</published><updated>2019-02-25T05:02:30.022000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#2262</id><summary type="html">Hi sir @ Karl Rupp I created the new buffers in the new context but even after changing the thing still getting the same execution time . Code int main() { typedef float ScalarType; viennacl::tools::timer timer; double exec_timecpu; double exec_timegpu; viennacl::tools::uniform_random_numbers&lt;ScalarType&gt; randomNumber; viennacl::ocl::set_context_platform_index(1, 1); viennacl::ocl::set_context_platform_index(0, 0); viennacl::ocl::switch_context(1); viennacl::matrix&lt;ScalarType&gt; vcl_A(BLAS3_MATRIX_SIZE,...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#d107" rel="alternate"/><published>2019-02-25T04:56:37.994000Z</published><updated>2019-02-25T04:56:37.994000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#d107</id><summary type="html">No, you can't use a buffer in multiple contexts. The reason is that memory allocations in OpenCL are tied to a particular context.</summary></entry><entry><title>venkatesh karthick modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#2262" rel="alternate"/><published>2019-02-25T04:47:34.763000Z</published><updated>2019-02-25T04:47:34.763000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#2262</id><summary type="html"/></entry><entry><title>venkatesh karthick posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#2262" rel="alternate"/><published>2019-02-25T04:45:42.302000Z</published><updated>2019-02-25T04:45:42.302000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#2262</id><summary type="html">So I can't use same buffers for different context ??</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6170b890ac/?limit=25#c1bb" rel="alternate"/><published>2019-02-24T00:52:17.311000Z</published><updated>2019-02-24T00:52:17.311000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6170b890ac/?limit=25#c1bb</id><summary type="html">See here; https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#1f03" rel="alternate"/><published>2019-02-24T00:51:53.956000Z</published><updated>2019-02-24T00:51:53.956000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#1f03</id><summary type="html">After you switch_memory_context(), you need to create some buffer in the new context. The reason why you obtained the same results is because the operands still "live" in the old context, so the operations on the 'old' objects still happen in the "null space" ;-)</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#3301" rel="alternate"/><published>2019-02-24T00:47:53.679000Z</published><updated>2019-02-24T00:47:53.679000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#3301</id><summary type="html">Yes, cou can provide the raw-buffers from the GPU to avoid the host&lt;-&gt;device copy. Have a look at viennacl::backend::memory_write(): http://viennacl.sourceforge.net/doc/namespaceviennacl_1_1backend.html</summary></entry><entry><title>venkatesh karthick modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6170b890ac/?limit=25#c3f0" rel="alternate"/><published>2019-02-22T09:31:07.125000Z</published><updated>2019-02-22T09:31:07.125000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6170b890ac/?limit=25#c3f0</id><summary type="html">I am getting the same result even if i select GPU like this viennacl::ocl::set_context_device_type(0, viennacl::ocl::gpu_tag()); and in CPU like this viennacl::ocl::set_context_device_type(1, viennacl::ocl::cpu_tag()); my guess this both is running on GPU only i need help in running the code in CPU and need to note the time taken by the both operation. i tried matrix multiplication. any answers will be appreciated Thanks in advance.</summary></entry><entry><title>venkatesh karthick posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#5c05" rel="alternate"/><published>2019-02-22T05:40:50.738000Z</published><updated>2019-02-22T05:40:50.738000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/7b9bdc3bbe/?limit=25#5c05</id><summary type="html">here is my code: viennacl::ocl::set_context_platform_index(1, 1); viennacl::ocl::set_context_platform_index(0, 0); viennacl::ocl::switch_context(0); std::cout &lt;&lt; "--- Computing matrix-matrix product using viennacl in GPU ---" &lt;&lt; std::endl; timer.start(); vcl_C = viennacl::linalg::prod(vcl_A, vcl_B); exec_time = timer.get(); std::cout &lt;&lt; " - Execution time: " &lt;&lt; exec_time &lt;&lt; std::endl; std::cout &lt;&lt; "result on GPU: "&lt;&lt;viennacl::ocl::current_device().name() &lt;&lt; std::endl; //same operation on CPU std::cout...</summary></entry><entry><title>venkatesh karthick posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6170b890ac/?limit=25#c3f0" rel="alternate"/><published>2019-02-22T04:27:22.483000Z</published><updated>2019-02-22T04:27:22.483000Z</updated><author><name>venkatesh karthick</name><uri>https://sourceforge.net/u/venkateshkarthi/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6170b890ac/?limit=25#c3f0</id><summary type="html">I am getting the same result even if i select GPU like this viennacl::ocl::set_context_device_type(n, viennacl::ocl::gpu_tag()); and viennacl::ocl::set_context_device_type(n, viennacl::ocl::cpu_tag()); my guess this both is running on GPU only i need help in running the code in CPU and need to note the time taken by the both operation. i tried matrix multiplication. any answers will be appreciated Thanks in advance.</summary></entry><entry><title>Robert Bischof posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#72a1" rel="alternate"/><published>2019-02-05T13:41:52.446000Z</published><updated>2019-02-05T13:41:52.446000Z</updated><author><name>Robert Bischof</name><uri>https://sourceforge.net/u/rbischof/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#72a1</id><summary type="html">Thanks for you fast reply. You are right: pardiso uses 1-base indexed arrays. After converting to 0-base indexed arrays ILU and gmres will work without problems. If have another question. Is it possible to use a parallel preconditioner and initialize the compressed_matrix with arrays from GPU memory, so I could save the copy time from GPU-CPU-GPU?</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#76b8" rel="alternate"/><published>2019-02-05T11:30:21.017000Z</published><updated>2019-02-05T11:30:21.017000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#76b8</id><summary type="html">Hi Robert, please verify the following: ia and ij are 32-bit integers ia[0] is zero length of ia is ld+1 do you have zeros on the main diagonal? If so, ILU will not work. * does the solver converge for a simple Poisson problem without preconditioner Try to print the matrix after calling .set(); it should reproduce the input matrix. Best regards, Karli</summary></entry><entry><title>Robert Bischof modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#452c" rel="alternate"/><published>2019-02-05T10:59:59.437000Z</published><updated>2019-02-05T10:59:59.437000Z</updated><author><name>Robert Bischof</name><uri>https://sourceforge.net/u/rbischof/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#452c</id><summary type="html">Hello, I will use the gmres solver for solving my linear equation system on GPU (CUDA). I have the 3 Arrays of the CSR Sparse format as pure C-Arrays (int, int, double) and try to initialize the vcl_compressed_matrix with the set function. typedef double ScalarType; viennacl::compressed_matrix&lt;ScalarType&gt; vcl_matrix; vcl_matrix.set(ia,ja,val,ld,ld,nnz); But after solving with gmres (using ILU preconditioner) I get a vector with only nan entries. Is there something wrong with my way of initializing?...</summary></entry><entry><title>Robert Bischof modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#452c" rel="alternate"/><published>2019-02-05T10:59:12.174000Z</published><updated>2019-02-05T10:59:12.174000Z</updated><author><name>Robert Bischof</name><uri>https://sourceforge.net/u/rbischof/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#452c</id><summary type="html">Hello, I will use the gmres solver for solving my linear equation system on GPU (CUDA). I have the 3 Arrays of the CSR Sparse format as pure C-Arrays (int, int, double) and try to initialize the vcl_compressed_matrix with the set function. typedef double ScalarType; viennacl::compressed_matrix&lt;ScalarType&gt; vcl_matrix; vcl_matrix.set(ia,ja,val,ld,ld,nnz); But after solving with gmres (using ILU preconditioner) I get a vector with only nan entries. Is there something wrong with my way of initializing?...</summary></entry><entry><title>Robert Bischof modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#452c" rel="alternate"/><published>2019-02-05T10:58:36.694000Z</published><updated>2019-02-05T10:58:36.694000Z</updated><author><name>Robert Bischof</name><uri>https://sourceforge.net/u/rbischof/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#452c</id><summary type="html">Hello, I will use the gmres solver for solving my linear equation system on GPU (CUDA). I have the 3 Arrays of the CSR Sparse format as pure C-Arrays (int, int, double) and try to initialize the vcl_compressed_matrix with the set function. viennacl::compressed_matrix&lt;ScalarType&gt; vcl_matrix; vcl_matrix.set(ia,ja,val,ld,ld,nnz); But after solving with gmres (using ILU preconditioner) I get a vector with only nan entries. Is there something wrong with my way of initializing? A few more information:...</summary></entry><entry><title>Robert Bischof posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#452c" rel="alternate"/><published>2019-02-05T10:58:16.685000Z</published><updated>2019-02-05T10:58:16.685000Z</updated><author><name>Robert Bischof</name><uri>https://sourceforge.net/u/rbischof/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/50c1de3aa7/?limit=25#452c</id><summary type="html">Hello, I will use the gmres solver for solving my linear equation system on GPU (CUDA). I have the 3 Arrays of the CSR Sparse format as pure C-Arrays (int, int, double) and try to initialize the vcl_compressed_matrix with the set function. iennacl::compressed_matrix&lt;ScalarType&gt; vcl_matrix; vcl_matrix.set(ia,ja,val,ld,ld,nnz); But after solving with gmres (using ILU preconditioner) I get a vector with only nan entries. Is there something wrong with my way of initializing? A few more information: I...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/02f2473f90/?limit=25#bf80" rel="alternate"/><published>2018-12-06T09:37:01.162000Z</published><updated>2018-12-06T09:37:01.162000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/02f2473f90/?limit=25#bf80</id><summary type="html">Hi Marek, the main developer of pyViennaCL has unfortunately moved on to other projects. Thus, the project is pretty much abandoned right now. I'm happy to integrate bug-fixes, though. Best regards, Karli</summary></entry><entry><title>Marek posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/02f2473f90/?limit=25#5a58" rel="alternate"/><published>2018-12-04T13:11:55.904000Z</published><updated>2018-12-04T13:11:55.904000Z</updated><author><name>Marek</name><uri>https://sourceforge.net/u/mwojc/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/02f2473f90/?limit=25#5a58</id><summary type="html">Hi! I'd lie to ask what is a status of pyviennacl wrapper? It seems that it is currently couple of years behind ViennaCL. I checked that it still compiles and works (at least on my linux and python 2.7 machines), but... are there are there any plans for updated release? Regards, Marek</summary></entry><entry><title>p ding posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/cbd74d21/?limit=25#2f85" rel="alternate"/><published>2018-09-16T08:52:10.172000Z</published><updated>2018-09-16T08:52:10.172000Z</updated><author><name>p ding</name><uri>https://sourceforge.net/u/ztdep/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/cbd74d21/?limit=25#2f85</id><summary type="html">Dear sir: It seems that there is no need to set the dimensions of block and grid. Could you please tell me how to dermine it in the ViennalCL. Regards</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c973679c/?limit=25#5ab4" rel="alternate"/><published>2018-07-19T17:15:57.819000Z</published><updated>2018-07-19T17:15:57.819000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c973679c/?limit=25#5ab4</id><summary type="html">a temporary array is the easiest, yes. yes, it is also possible to copy one row at a time. You create a vector_range obtained via row() and copy your data to that. No, the padding only rounds up to the next multiple of 128. For a 10000x10000 matrix, this would be 10112x10112.</summary></entry><entry><title>André posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c973679c/?limit=25#8236" rel="alternate"/><published>2018-07-19T17:07:51.816000Z</published><updated>2018-07-19T17:07:51.816000Z</updated><author><name>André</name><uri>https://sourceforge.net/u/techwinder/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c973679c/?limit=25#8236</id><summary type="html">Thanks, however this raises a few more questions. I'm dealing with large dense matrices, up to 10,000x10,000. 1. Since my app stores the matrix continuously in memory, do I need to build a new temporary stl array including the padding, then use the fast_copy() method? This will cause some RAM issues. 2. Or is it possible to copy one row at a time in the ViennaCl matrix? If so I could'nt find a way to do this, because the copy() method seems to accept only a reference to the matrix and not a pointer...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c973679c/?limit=25#74f5" rel="alternate"/><published>2018-07-19T16:02:26.156000Z</published><updated>2018-07-19T16:02:26.156000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c973679c/?limit=25#74f5</id><summary type="html">Hi Andre, you need to take padding into account: http://viennacl.sourceforge.net/doc/manual-types.html#manual-types-matrix Best regards, Karli</summary></entry><entry><title>André posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c973679c/?limit=25#f3ea" rel="alternate"/><published>2018-07-19T15:46:25.919000Z</published><updated>2018-07-19T15:46:25.919000Z</updated><author><name>André</name><uri>https://sourceforge.net/u/techwinder/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c973679c/?limit=25#f3ea</id><summary type="html">Hi Karl, Thanks for making this great app. I'm considering using it for my own project xflr5 and it will really speed up things. I store the square dense matrix as a continuous 1 dimensional array, so it seems fast_copy is required to convert it to ViennaCL format. However the following test code seems to copy only the first row. Could you explain what I am missing? viennacl::matrix&lt;double&gt; vcl_matrix(N, N); std::vector&lt;double&gt; aij(N*N); //fill a random matrix for(int i=0; i&lt;N; i++) for(int j=0;...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/3b37dbdb/?limit=25#1723" rel="alternate"/><published>2018-07-17T11:36:10.069000Z</published><updated>2018-07-17T11:36:10.069000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/3b37dbdb/?limit=25#1723</id><summary type="html">Full copies are used. You may be observing virtual memory consumption instead of "real" memory consumption. Virtual memory consumption is often not a good representation of the actual memory consumed. In most application the factor of 2 for the duplicate system is not a big concern, because you will have auxiliary datastructures that are of similar size. Also, preconditioners may consume more memory than the system itself. If you need to save every byte, you can still populate the data arrays of...</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/3b37dbdb/?limit=25#c12a" rel="alternate"/><published>2018-07-17T06:16:53.064000Z</published><updated>2018-07-17T06:16:53.064000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/3b37dbdb/?limit=25#c12a</id><summary type="html">Ok. Thanks for answers. Now I firstly set data in ublas::compressd_matrix and when do a copy to viennaCl::compressed_matrix. I tested and it is parallel. As i see memory not grow in 2. Does it mean that viennacl::copy not doing a full copy and just do bew pointers to allocated memory?</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/3b37dbdb/?limit=25#d256" rel="alternate"/><published>2018-07-16T14:11:46.193000Z</published><updated>2018-07-16T14:11:46.193000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/3b37dbdb/?limit=25#d256</id><summary type="html">Addendum: To benefit from OpenMP, one way is to pass ViennaCL objects (viennacl::compressed_matrix and viennacl::vector) and make sure VIENNACL_WITH_OPENMP is defined.</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/3b37dbdb/?limit=25#7de7" rel="alternate"/><published>2018-07-16T14:09:49.278000Z</published><updated>2018-07-16T14:09:49.278000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/3b37dbdb/?limit=25#7de7</id><summary type="html">Hi Dmitriy, as far as I know, the vector operations in uBLAS (which get used if you pass uBLAS types to the solvers) do not benefit from OpenMP. Best regards, Karli</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/3b37dbdb/?limit=25#fac6" rel="alternate"/><published>2018-07-16T13:30:30.681000Z</published><updated>2018-07-16T13:30:30.681000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/3b37dbdb/?limit=25#fac6</id><summary type="html">Hello. Wanted to ask if OpenMp parallization works, if i set ublas::compressed_matrix and ublas::vector to solve function?</summary></entry><entry><title>Kaushik posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/81ef1cff/?limit=25#6e4c" rel="alternate"/><published>2018-06-26T01:14:02.797000Z</published><updated>2018-06-26T01:14:02.797000Z</updated><author><name>Kaushik</name><uri>https://sourceforge.net/u/kaushikggg/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/81ef1cff/?limit=25#6e4c</id><summary type="html">Hello Karl, Yes that works. Thanks a lot for looking into it. (Quite an interesting bug! :) )</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/81ef1cff/?limit=25#cc14" rel="alternate"/><published>2018-06-26T00:28:51.737000Z</published><updated>2018-06-26T00:28:51.737000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/81ef1cff/?limit=25#cc14</id><summary type="html">Hi Kaushik, this was very interesting to debug. Short answer: The assert() is correct and incorrect at the same time; you have to comment line 231 in arch-linux2-c-debug/include/viennacl/ocl/kernel.hpp. Longer answer: The issue is caused by the header-only nature of ViennaCL and how mapped memory spaces work with shared objects. When x_viennacl is created inside PETSc, the ViennaCL routines compiled into the PETSc shared library allocate the vector. At that point, x_viennacl keeps track of the context...</summary></entry><entry><title>Kaushik posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/81ef1cff/?limit=25#1622" rel="alternate"/><published>2018-06-25T20:50:50.607000Z</published><updated>2018-06-25T20:50:50.607000Z</updated><author><name>Kaushik</name><uri>https://sourceforge.net/u/kaushikggg/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/81ef1cff/?limit=25#1622</id><summary type="html">Hello all, I have been trying to make use of custom kernels to operate on a precomputed vector obtained from PETSc, but I am getting an error which reads: /home/user/Installation_dir/petsc/arch-linux2-c-debug/include/viennacl/ocl/kernel.hpp:231: void viennacl::ocl::kernel::arg(unsigned int, const VCL_TYPE&amp;) [with VCL_TYPE = viennacl::vector&lt;double&gt;]: Assertion `&amp;val.handle().opencl_handle().context() == &amp;handle_.context() &amp;&amp; bool("Kernel and memory object not in the same context!")' failed. I am...</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#b45f" rel="alternate"/><published>2018-06-06T05:51:11.851000Z</published><updated>2018-06-06T05:51:11.851000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#b45f</id><summary type="html">Now I don't call static void delete_context() { initialized_.clear(); contexts_.clear(); current_context_id_ = 0; } in my class destructor, but in an outer class. The workflow runs like this: 1. construct b 2. run b (use viennacl::linalg::solve()) 3. destruct b 4. construct c 5. run c (use viennacl::linalg::solve()) 6. destruct c ... 7. delete_context() This temporarily solves my crash issue, though I have not figured out the crash reason.</summary></entry><entry><title>Siqi Zhuang modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#8eda" rel="alternate"/><published>2018-06-06T03:00:40.888000Z</published><updated>2018-06-06T03:00:40.888000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#8eda</id><summary type="html">I found that the error message in crash point is ViennaCL: Could not find program 'float_vector' So after I delete the contexts in b, some programs are deleted? But I set the context in c and switched to that context. Theoretically, all programs should be recompiled.</summary></entry><entry><title>Siqi Zhuang modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#8eda" rel="alternate"/><published>2018-06-06T02:13:07.300000Z</published><updated>2018-06-06T02:13:07.300000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#8eda</id><summary type="html">I found that the error message in crash point is ViennaCL: Could not find program 'float_vector' So after I delete the contexts in b, some programs are deleted? But I set the context in c and switched to that context.</summary></entry><entry><title>Siqi Zhuang modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#8eda" rel="alternate"/><published>2018-06-06T02:12:34.475000Z</published><updated>2018-06-06T02:12:34.475000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#8eda</id><summary type="html">I found that the error message in crash point is ViennaCL: Could not find program 'float_vector' So after I delete the contexts in a, some programs are deleted? But I set the context in b and switched to that context.</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#8eda" rel="alternate"/><published>2018-06-06T01:51:40.651000Z</published><updated>2018-06-06T01:51:40.651000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#8eda</id><summary type="html">I found that the error message in crash point is ViennaCL: Could not find program 'float_vector'</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#ae5a" rel="alternate"/><published>2018-06-05T08:57:18.535000Z</published><updated>2018-06-05T08:57:18.535000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#ae5a</id><summary type="html">I create cl_context and pass it to viennacl. Then I ran kernels both in my program and math function in viennacl. When I release this context, how could I perfectly relase all resources of viennacl related to this context?</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#f76b" rel="alternate"/><published>2018-06-05T08:47:19.130000Z</published><updated>2018-06-05T08:47:19.130000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#f76b</id><summary type="html">This issue is very strange. It only occurs when I embed the codes to software. Strangely when I run my console program, it won't crash. The software don't use ViennaCL except for this piece of code. Theoretically speaking, it should be same when I run console or the software run the code. And I can't debug the software since I don't have source code. That's the reason why I can't really locate the issue... One thing reproduces the issue is clearing contexts_ in backend.hpp But if I don't clear the...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#326e" rel="alternate"/><published>2018-06-05T06:22:41.090000Z</published><updated>2018-06-05T06:22:41.090000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#326e</id><summary type="html">Is it possible for you to provide a minmal example that reproduces the issue? This would make it much easier for me to say what is going wrong.</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#f128" rel="alternate"/><published>2018-06-05T06:18:33.787000Z</published><updated>2018-06-05T06:18:33.787000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#f128</id><summary type="html">Hi. After adding static void delete_context() { initialized_.clear(); contexts_.clear(); current_context_id_ = 0; } in backend.hpp, and the program run this way: 1. construct b 2. run b (use viennacl::linalg::solve()) 3. destruct b 4. construct c 5. run c (use viennacl::linalg::solve()) 6. destruct c Sadly I found that program crashes when running to the viennacl::linalg::solve() function in Step 5. But if I don't use delete_context() in the destructor of b (step 3), it won't crash. It's interesting...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#d3ab" rel="alternate"/><published>2018-06-01T13:50:26.863000Z</published><updated>2018-06-01T13:50:26.863000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#d3ab</id><summary type="html">Hi Siqi, as you've observed, ViennaCL isn't good at freeing contexts. The context is modeled as a singleton that hovers in the background. Your workaround is fine and reasonable. Best regards, Karli</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#fa79" rel="alternate"/><published>2018-06-01T01:10:17.420000Z</published><updated>2018-06-01T01:10:17.420000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#fa79</id><summary type="html">Currently I add following codes in backend.hpp static void delete_context() { initialized_.clear(); contexts_.clear(); current_context_id_ = 0; } I call this function in my class destructor so as to release the static contex member. After doing this, it won't cause crash after my software exits. I'll keep watching.</summary></entry><entry><title>Siqi Zhuang modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#508d" rel="alternate"/><published>2018-05-31T06:44:09.559000Z</published><updated>2018-05-31T06:44:09.559000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#508d</id><summary type="html">I figure out a way to solve this problem. I check the current context id in the initialization of ViennaCL. Then I add a new context and switch to that context. long current_context_id = viennacl::ocl::backend&lt;false&gt;::current_context_id(); viennacl::ocl::setup_context(current_context_id + 1, m_context(), m_devices[0](), m_queue()); viennacl::ocl::switch_context(current_context_id + 1); But I don't really like this way. Can I delete the context in my class destructor by myself, so that each time when...</summary></entry><entry><title>Siqi Zhuang modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#508d" rel="alternate"/><published>2018-05-31T05:06:39.813000Z</published><updated>2018-05-31T05:06:39.813000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#508d</id><summary type="html">I figure out a way to solve this problem. I check the current context id in the initialization of ViennaCL. Then I add a new context and switch to that context. long current_context_id = viennacl::ocl::backend&lt;false&gt;::current_context_id(); viennacl::ocl::setup_context(current_context_id + 1, m_context(), m_devices[0](), m_queue()); viennacl::ocl::switch_context(current_context_id + 1); But I don't really like this way. Can I delete the context in my class destructor by myself, so that each time when...</summary></entry><entry><title>Siqi Zhuang modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#508d" rel="alternate"/><published>2018-05-31T05:02:25.840000Z</published><updated>2018-05-31T05:02:25.840000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#508d</id><summary type="html">I figure out a way to solve this problem. I check the current context id in the initialization of ViennaCL. Then I add a new context and switch to that context. long current_context_id = viennacl::ocl::backend&lt;false&gt;::current_context_id(); viennacl::ocl::setup_context(current_context_id + 1, m_context(), m_devices[0](), m_queue()); viennacl::ocl::switch_context(current_context_id + 1); But I don't really like this way. Can I delete the context in my class destructor by myself? And one more thing....</summary></entry><entry><title>Siqi Zhuang modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#b79c" rel="alternate"/><published>2018-05-31T04:53:50.017000Z</published><updated>2018-05-31T04:53:50.017000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#b79c</id><summary type="html">By the way, I have to pass cl_mem objects to viennacl::compressed_matrix&lt;cl_float&gt;. In object c, program crashes when running to this line because of the context problem: viennacl::compressed_matrix&lt;cl_float&gt; *vcl_sparse_matrix = new viennacl::compressed_matrix&lt;cl_float&gt;((*buf_node2NBRAddress)(), (*buf_node2NBRCol)(), (*buf_matValueL)(), nbBlockNodes, nbBlockNodes, tolMatElementNum);</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#508d" rel="alternate"/><published>2018-05-31T04:44:29.392000Z</published><updated>2018-05-31T04:44:29.392000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#508d</id><summary type="html">I figure out a way to solve this problem. I check the current context id in the initialization of ViennaCL. Then I add a new context and switch to that context. long current_context_id = viennacl::ocl::backend&lt;false&gt;::current_context_id(); viennacl::ocl::setup_context(current_context_id + 1, m_context(), m_devices[0](), m_queue()); viennacl::ocl::switch_context(current_context_id + 1); But I don't really like this way. Can I delete the context in my class destructor by myself?</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#b79c" rel="alternate"/><published>2018-05-31T04:04:38.194000Z</published><updated>2018-05-31T04:04:38.194000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#b79c</id><summary type="html">By the way, I have to pass cl_mem objects to viennacl::compressed_matrix&lt;cl_float&gt;. In obect c, program crashes when running to this line because of the context problem: viennacl::compressed_matrix&lt;cl_float&gt; *vcl_sparse_matrix = new viennacl::compressed_matrix&lt;cl_float&gt;((*buf_node2NBRAddress)(), (*buf_node2NBRCol)(), (*buf_matValueL)(), nbBlockNodes, nbBlockNodes, tolMatElementNum);</summary></entry><entry><title>Siqi Zhuang modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#20b2" rel="alternate"/><published>2018-05-31T04:00:37.675000Z</published><updated>2018-05-31T04:00:37.675000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#20b2</id><summary type="html">Hi. I have a Class A where an existed OpenCL context is provided by using viennacl::ocl::setup_context(). Now I have two Class A objests, says b and c. The program runs like this: 1. construct b 2. run b 3. destruct b 4. construct c 5. run c 6. destruct c But I found that setting context in c always fails. I had a warning in debug mode: ViennaCL: Warning in init_context(): Providing a list of devices has no effect, because context for ViennaCL is already created! I wonder how to elegantly solve this...</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#20b2" rel="alternate"/><published>2018-05-31T03:58:13.197000Z</published><updated>2018-05-31T03:58:13.197000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/6288ede5/?limit=25#20b2</id><summary type="html">Hi. I have a Class A where an existed OpenCL context is provided by using viennacl::ocl::setup_context(). Now I have two Class A objests, says b and c. The program runs like this: 1. construct b 2. run b 3. destruct b 4. construct c 5. run c 6. destruct c But I found that setting context in c always fails. I had a warning in debug mode: ViennaCL: Warning in init_context(): Providing a list of devices has no effect, because context for ViennaCL i s already created! I wonder how to elegantly solve...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#487e/dea8" rel="alternate"/><published>2018-04-23T10:34:19.493000Z</published><updated>2018-04-23T10:34:19.493000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#487e/dea8</id><summary type="html">Which hardware do you run on? On desktop CPUs with just 2-4 cores it's often hard to see notable performance gains, because memory channels may be saturated with just a single core.</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#487e" rel="alternate"/><published>2018-04-23T10:24:20.633000Z</published><updated>2018-04-23T10:24:20.633000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#487e</id><summary type="html">Hello. I tried different iterative solvers and a clear cpp code, not launched from another thread (without GUI). I compiled it on MinGW, g++, but I have not scale in no case. Maybe you have an example with OpenMp Acceleration? To test it.</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac/8ea4/e955/e588/c3e5" rel="alternate"/><published>2018-04-23T09:24:03.537000Z</published><updated>2018-04-23T09:24:03.537000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac/8ea4/e955/e588/c3e5</id><summary type="html">Great, thanks for the feedback. Yes, a new release is over-due already :-)</summary></entry><entry><title>Karl Rupp modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/0a21f2be/?limit=25#df52" rel="alternate"/><published>2018-04-23T09:22:59.928000Z</published><updated>2018-04-23T09:22:59.928000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/0a21f2be/?limit=25#df52</id><summary type="html">No, there is no functionality in ViennaCL (natively) that allows you to use multiple GPUs for jointly solving a single large system. This is only possible through PETSc by using multiple CPU processes (via MPI). Addendum: If you want to use ViennaCL through PETSc, you provide the command line options -mat_type aijviennacl -vec_type viennacl and specify a proper preconditioner. Typical choices (as of 04/2018) with a single MPI rank are: -pc_type none to run without preconditioner -pc_type saviennacl...</summary></entry><entry><title>Karl Rupp modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/0a21f2be/?limit=25#df52" rel="alternate"/><published>2018-04-23T09:22:19.765000Z</published><updated>2018-04-23T09:22:19.765000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/0a21f2be/?limit=25#df52</id><summary type="html">No, there is no functionality in ViennaCL (natively) that allows you to use multiple GPUs for jointly solving a single large system. This is only possible through PETSc by using multiple CPU processes (via MPI). Addendum: If you want to use ViennaCL through PETSc, you provide the command line options -mat_type aijviennacl -vec_type viennacl and specify a proper preconditioner. Typical choices (as of 04/2018) with a single MPI rank are: -pc_type none to run without preconditioner -pc_type saviennacl...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/0a21f2be/?limit=25#df52" rel="alternate"/><published>2018-04-23T09:15:26.072000Z</published><updated>2018-04-23T09:15:26.072000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/0a21f2be/?limit=25#df52</id><summary type="html">No, there is no functionality in ViennaCL (natively) that allows you to use multiple GPUs for jointly solving a single large system. This is only possible through PETSc by using multiple CPU processes (via MPI).</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac/8ea4/e955/e588" rel="alternate"/><published>2018-04-23T02:36:40.769000Z</published><updated>2018-04-23T02:36:40.769000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac/8ea4/e955/e588</id><summary type="html">Hi, Karl. That finally worked. Thanks. It's interesting that it's been a long time since the latest release 1.7.1 released.</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac/8ea4/e955" rel="alternate"/><published>2018-04-20T09:15:33.496000Z</published><updated>2018-04-20T09:15:33.496000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac/8ea4/e955</id><summary type="html">Thank you, Siqi, this indeed exposed a bug in ViennaCL's compressed_matrix. I pushed a fix here: https://github.com/viennacl/viennacl-dev/commit/c580f4cb4286d4d2043e6847fdeb19b4351fea7c Please update your local copy of ViennaCL accordingly, or directly use the developer Version of ViennaCL from GitHub. Thanks again!</summary></entry><entry><title>Siqi Zhuang modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac/8ea4" rel="alternate"/><published>2018-04-20T08:57:33.277000Z</published><updated>2018-04-20T08:57:33.277000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac/8ea4</id><summary type="html">Hi. I've set the contex using viennacl::ocl::setup_context I refer to the code you mentioned above and set the context. But program still failed. I found the problem may really be the context. Here is the debug info my program prints: Platform info: NVIDIA Corporation: OpenCL 1.2 CUDA 9.1.84 Quadro M3000M Number of devices for custom context: 1 Creating context... Existing context: 00000000027B5630 ViennaCL uses context: 00000000027B5630 Assertion failed: p_context_ != NULL &amp;&amp; bool("Logic error:...</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#89da/a9b6" rel="alternate"/><published>2018-04-20T08:57:22.083000Z</published><updated>2018-04-20T08:57:22.083000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#89da/a9b6</id><summary type="html">Yes, please try dev-version from GitHub. I don't remember the exact timeline; it is possible that OpenMP-support has only been added after the last release.</summary></entry><entry><title>Siqi Zhuang posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac/8ea4" rel="alternate"/><published>2018-04-20T08:56:19.032000Z</published><updated>2018-04-20T08:56:19.032000Z</updated><author><name>Siqi Zhuang</name><uri>https://sourceforge.net/u/antibes0415/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac/8ea4</id><summary type="html">Hi. I've set the contex using viennacl::ocl::setup_context I refer to the code you mentioned above and set the context. But program still failed. I found the problem may really be the context. Here is the debug info my program prints: Platform info: NVIDIA Corporation: OpenCL 1.2 CUDA 9.1.84 Quadro M3000M Number of devices for custom context: 1 Creating context... Existing context: 00000000027B5630 ViennaCL uses context: 00000000027B5630 Assertion failed: p_context_ != NULL &amp;&amp; bool("Logic error:...</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#89da" rel="alternate"/><published>2018-04-20T08:51:25.830000Z</published><updated>2018-04-20T08:51:25.830000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#89da</id><summary type="html">So in my iterative_operations.hpp no such lines: #pragma omp parallel for reduction(+: inner_prod_ApAp, inner_prod_pAp, inner_prod_Ap_r0star) I downloaded it from http://viennacl.sourceforge.net/. Do I need to take a dev version from github?</summary></entry><entry><title>Dmitriy posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#86a4" rel="alternate"/><published>2018-04-20T08:35:03.015000Z</published><updated>2018-04-20T08:35:03.015000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#86a4</id><summary type="html">about 3millions of unknowns. And in each row of a matrix about 60 non-zero elements</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#8121/4ff9" rel="alternate"/><published>2018-04-20T08:27:22.912000Z</published><updated>2018-04-20T08:27:22.912000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#8121/4ff9</id><summary type="html">Yes, the BiCGStab routines without preconditioner use OpenMP: https://github.com/viennacl/viennacl-dev/blob/master/viennacl/linalg/host_based/iterative_operations.hpp How large is your test problem? For less than ~100.000 unknowns you will not see any gains from threading, as synchronization points dominate.</summary></entry><entry><title>Karl Rupp posted a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac" rel="alternate"/><published>2018-04-20T08:21:39.931000Z</published><updated>2018-04-20T08:21:39.931000Z</updated><author><name>Karl Rupp</name><uri>https://sourceforge.net/u/koarl0815/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/4b2f6d7b/?limit=25#37ac</id><summary type="html">Do you set up ViennaCL to use your custom OpenCL context? Have a look here for an example: https://github.com/viennacl/viennacl-dev/blob/master/examples/tutorial/custom-context.cpp In particular, viennacl::ocl::setup_context() is important: https://github.com/viennacl/viennacl-dev/blob/master/examples/tutorial/custom-context.cpp#L210 See also http://viennacl.sourceforge.net/doc/manual-custom-contexts.html If you don't make your own OpenCL context (I assume you use your own context) known to ViennaCL,...</summary></entry><entry><title>Dmitriy modified a comment on discussion General Discussion</title><link href="https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#8121" rel="alternate"/><published>2018-04-20T08:20:22.829000Z</published><updated>2018-04-20T08:20:22.829000Z</updated><author><name>Dmitriy</name><uri>https://sourceforge.net/u/dmitriyp/profile/</uri></author><id>https://sourceforge.net/p/viennacl/discussion/1143678/thread/c6267c63/?limit=25#8121</id><summary type="html">So I tried without preconditioner at all, and proc load remain 25%. Do you know if bicgstab solver parallel using openMP?</summary></entry></feed>