<?xml version="1.0" encoding="UTF-8"?>
<!-- generator="bbPress/1.0.2" -->
<rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom">
	<channel>
		<title>k-Wave User Forum &#187; Topic: GPU Memory</title>
		<link>http://www.k-wave.org/forum/topic/gpu-memory</link>
		<description>Support for the k-Wave MATLAB toolbox</description>
		<language>en-US</language>
		<pubDate>Wed, 13 May 2026 15:10:23 +0000</pubDate>
		<generator>http://bbpress.org/?v=1.0.2</generator>
		<textInput>
			<title><![CDATA[Search]]></title>
			<description><![CDATA[Search all topics from these forums.]]></description>
			<name>q</name>
			<link>http://www.k-wave.org/forum/search.php</link>
		</textInput>
		<atom:link href="http://www.k-wave.org/forum/rss/topic/gpu-memory" rel="self" type="application/rss+xml" />

		<item>
			<title>Bradley Treeby on "GPU Memory"</title>
			<link>http://www.k-wave.org/forum/topic/gpu-memory#post-5996</link>
			<pubDate>Mon, 19 Jun 2017 20:48:44 +0000</pubDate>
			<dc:creator>Bradley Treeby</dc:creator>
			<guid isPermaLink="false">5996@http://www.k-wave.org/forum/</guid>
			<description>&#60;p&#62;Hi Javi,&#60;/p&#62;
&#60;p&#62;Nothing I can think of immediately. I guess try searching for how to alter MATLAB's maximum array size preference.&#60;/p&#62;
&#60;p&#62;Brad.
&#60;/p&#62;</description>
		</item>
		<item>
			<title>jcudeiro on "GPU Memory"</title>
			<link>http://www.k-wave.org/forum/topic/gpu-memory#post-5979</link>
			<pubDate>Thu, 08 Jun 2017 17:23:02 +0000</pubDate>
			<dc:creator>jcudeiro</dc:creator>
			<guid isPermaLink="false">5979@http://www.k-wave.org/forum/</guid>
			<description>&#60;p&#62;When using the GPU, I create a temp file of 36GB. The computation finishes, but then I get this:&#60;/p&#62;
&#60;p&#62;Error using h5readc&#60;br /&#62;
Requested 211x211x145x1439 (34.6GB) array exceeds maximum array size preference. Creation of arrays greater than this limit may take a long time and cause MATLAB to become unresponsive. See array size limit or preference panel for more information.&#60;/p&#62;
&#60;p&#62;Is there any way around this error?&#60;/p&#62;
&#60;p&#62;Many thanks in advance!&#60;/p&#62;
&#60;p&#62;Best,&#60;br /&#62;
Javi
&#60;/p&#62;</description>
		</item>
		<item>
			<title>jcudeiro on "GPU Memory"</title>
			<link>http://www.k-wave.org/forum/topic/gpu-memory#post-5733</link>
			<pubDate>Thu, 03 Nov 2016 12:10:29 +0000</pubDate>
			<dc:creator>jcudeiro</dc:creator>
			<guid isPermaLink="false">5733@http://www.k-wave.org/forum/</guid>
			<description>&#60;p&#62;Thank you very much Brad!
&#60;/p&#62;</description>
		</item>
		<item>
			<title>Bradley Treeby on "GPU Memory"</title>
			<link>http://www.k-wave.org/forum/topic/gpu-memory#post-5728</link>
			<pubDate>Tue, 01 Nov 2016 22:38:56 +0000</pubDate>
			<dc:creator>Bradley Treeby</dc:creator>
			<guid isPermaLink="false">5728@http://www.k-wave.org/forum/</guid>
			<description>&#60;p&#62;Hi jcudeiro,&#60;/p&#62;
&#60;p&#62;The two function calls you mention run exclusively on the GPU. Unfortunately, the simulations are too big to fit in GPU memory as the error messages suggest. There is currently no way to split the simulation between CPU and GPU. We did experiment with just offloading the FFTs to the GPU, but the communication overhead meant that there wasn't any real performance benefit. &#60;/p&#62;
&#60;p&#62;There is no way to save a movie using the C++ or CUDA codes, as they don't have a graphical output.&#60;/p&#62;
&#60;p&#62;Hope that helps,&#60;/p&#62;
&#60;p&#62;Brad.
&#60;/p&#62;</description>
		</item>
		<item>
			<title>jcudeiro on "GPU Memory"</title>
			<link>http://www.k-wave.org/forum/topic/gpu-memory#post-5727</link>
			<pubDate>Mon, 31 Oct 2016 10:25:48 +0000</pubDate>
			<dc:creator>jcudeiro</dc:creator>
			<guid isPermaLink="false">5727@http://www.k-wave.org/forum/</guid>
			<description>&#60;p&#62;Hi there!&#60;/p&#62;
&#60;p&#62;First of all, amazing software. Really nice job!&#60;/p&#62;
&#60;p&#62;Second of all, I am trying to use a GPU (NVIDIA GeForce 1080, with an Available Memory of: 7.0829e+09) to run a simulation in 3D (heterogeneous medium). My computational grid size is: 633 by 633 by 446 grid points and 23,745 time steps.&#60;/p&#62;
&#60;p&#62;When using kspaceFirstOrder3DG (datacast set to single or gpuArray-single) I get this error:&#60;br /&#62;
&#60;code&#62;Error: Not enough CPU or GPU memory to run this simulation.&#60;/code&#62;&#60;br /&#62;
&#60;pre&#62;&#60;code&#62;Error using h5readc
Unable to open file. Filename may be corrupt or have unsupported characters.
Error in h5read (line 58)
[data,var_class] = h5readc(Filename,Dataset,start,count,stride)&#60;/code&#62;&#60;/pre&#62;
&#60;p&#62;And with kspaceFirstOrder3D (datacast set to gpuArray-single) this one:&#60;br /&#62;
&#60;pre&#62;&#60;code&#62;Error using gpuArray
Out of memory on device. To view more detail about available memory on the GPU, use &#38;#39;gpuDevice()&#38;#39;. If the problem persists, reset the GPU
by calling &#38;#39;gpuDevice(1)&#38;#39;.

Error in kspaceFirstOrder_dataCast (line 206)
eval([cast_variables{cast_index} &#38;#39; = &#38;#39; data_cast &#38;#39;(&#38;#39; data_cast_prepend &#38;#39;(&#38;#39; cast_variables{cast_index} &#38;#39;));&#38;#39;]);&#60;/code&#62;&#60;/pre&#62;
&#60;p&#62; - Is there any way I can split the simulation to run it just with the GPU? (For smaller grids it works fine). Or between the GPU and CPU?&#60;/p&#62;
&#60;p&#62; - Is it possible to record the 'movie' when using kspaceFirstOrder3DG?&#60;/p&#62;
&#60;p&#62;Many thanks in advance!
&#60;/p&#62;</description>
		</item>

	</channel>
</rss>
