http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05 Support for the k-Wave MATLAB toolbox en-US Tue, 12 May 2026 23:42:53 +0000 http://bbpress.org/?v=1.0.2 q http://www.k-wave.org/forum/search.php http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05/page/2#post-594 Mon, 09 Jul 2012 03:40:09 +0000 yinhaoscu 594@http://www.k-wave.org/forum/ <p>Hi Brad,</p> <p>It works. Thanks a lot.</p> <p>Hao </p> http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05/page/2#post-591 Fri, 06 Jul 2012 02:06:01 +0000 Bradley Treeby 591@http://www.k-wave.org/forum/ <p>Hi Hao,</p> <p>As you say, the problem is with <code>reshape</code>. The GPUmat implementation doesn't allow the use of <code>[]</code> to allow the dimensions to be calculated automatically. This error should only occur when using an object of the kWaveTransducer class as a sensor with GPUmat.</p> <p>In this case, you can replace the line where you are receiving an error with:</p> <p>sensor_data(:, file_index - sensor_data_buffer_size + 1) = sum(reshape(sum(reshape(current_vals, sensor.number_active_elements*sensor.element_width, sensor.element_length), 2), sensor.element_width, sensor.number_active_elements).', 2);</p> <p>And similarly for the case without elevation focusing:</p> <p>sensor_data(:, file_index) = sum(reshape(sum(reshape(p(sensor_mask_ind), sensor.number_active_elements*sensor.element_width, sensor.element_length), 2), sensor.element_width, sensor.number_active_elements).', 2);</p> <p>Let me know if you still receive errors after this.</p> <p>Brad. </p> http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05#post-588 Thu, 05 Jul 2012 02:51:04 +0000 yinhaoscu 588@http://www.k-wave.org/forum/ <p>the same problems </p> http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05#post-572 Thu, 28 Jun 2012 02:55:47 +0000 Bradley Treeby 572@http://www.k-wave.org/forum/ <p>What about for the "Defining an Ultrasound Transducer" and "Simulating Ultrasound Beam Patterns" examples? </p> http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05#post-571 Thu, 28 Jun 2012 02:15:25 +0000 yinhaoscu 571@http://www.k-wave.org/forum/ <p>Hi Brad. The same problem with example_us_transducer_as_sensor.m<br /> ??? Error using ==&gt; reshape<br /> Dimension cannot be zero.</p> <p>Error in ==&gt; kspaceFirstOrder3D at 940<br /> sensor_data(:, file_index - sensor_data_buffer_size + 1) = ...</p> <p>Error in ==&gt; example_us_transducer_as_sensor at 137<br /> [sensor_data] = kspaceFirstOrder3D(kgrid, medium, source, transducer, input_args{:});</p> <p>Hao </p> http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05#post-569 Wed, 27 Jun 2012 23:49:32 +0000 Bradley Treeby 569@http://www.k-wave.org/forum/ <p>Hi Hao,</p> <p>Thanks for your follow-up. Do you also get an error when you use GPUmat with other k-Wave examples, or is just when you use a <code>kWaveTransducer</code> object as a sensor (i.e., the final two US examples)?</p> <p>Brad. </p> http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05#post-564 Mon, 25 Jun 2012 04:22:22 +0000 yinhaoscu 564@http://www.k-wave.org/forum/ <p>Brad, thanks for your reply. I followed your guideline, it worked further step. The problem changed to:<br /> Error using ==&gt; reshape<br /> Dimension cannot be zero.</p> <p>Error in ==&gt; kspaceFirstOrder3D at 940<br /> sensor_data(:, file_index - sensor_data_buffer_size + 1) = ...</p> <p>Error in ==&gt; example_us_bmode_linear_transducer at 239<br /> sensor_data = kspaceFirstOrder3D(kgrid, medium, transducer, transducer, input_args{:});</p> <p>***** It seems that sensor_data is a GPUtype, however reshape need a CPU type, so the dimension is zero. </p> http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05#post-562 Thu, 21 Jun 2012 05:52:11 +0000 Bradley Treeby 562@http://www.k-wave.org/forum/ <p>Hi Hao,</p> <p>Do you mean you downloaded the bsxfun from the k-Wave forum? Also, you will need to fix a data cast bug listed in the B.05 <a href="http://www.k-wave.org/forum/topic/b05-known-issues">known issues</a> on the 10th of April.</p> <p>The @ symbol in the @GPUtype folder name tells MATLAB to only use the functions within the folder when the inputs are of GPUtype. This is correct.</p> <p>Brad </p> http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05#post-561 Thu, 21 Jun 2012 03:44:13 +0000 yinhaoscu 561@http://www.k-wave.org/forum/ <p>Hi Brad.</p> <p>Thanks for your help. </p> <p>I did download bsxfun shipped with GPUmat and put it in @GPUtype directory. It seems that k-wave used the bsxfun provided by GPUmat, however it tried to use the times funtion from Matlab instead of the corresponding times located in the same directory of bsxfun (@GPUtype). Why there is a symbol @ before the name of directory @GPUtype? What can I do?</p> <p>Hao </p> http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05#post-556 Wed, 20 Jun 2012 08:40:13 +0000 Bradley Treeby 556@http://www.k-wave.org/forum/ <p>Hi Hao,</p> <p>Are you using the bsxfun shipped with GPUmat, or the one from the k-Wave forum? You will need to use the version from <a href="http://www.k-wave.org/forum/topic/getting-started-with-gpumat">here</a>.</p> <p>Brad. </p> http://www.k-wave.org/forum/topic/problem-of-gpumat-with-b05#post-555 Wed, 20 Jun 2012 07:37:52 +0000 yinhaoscu 555@http://www.k-wave.org/forum/ <p>??? Error using ==&gt; times<br /> Operations between Matlab arrays and GPU arrays are not supported. (ERROR code 5)</p> <p>Error in ==&gt; kspaceFirstOrder3D at 741<br /> ux_sgx = bsxfun(@times, pml_x_sgx, ...</p> <p>I checked that times.m is in @GPUType directory, what can I do? Thanks </p>