[Rtk-users] Type of CudaDataManager::m_BufferSize

[louie L]

Exactly. I changed this when I was his message a few weeks back in my own copy.
Thank you.
Maybe I will spend some time in the next few month to do some optimizations of the cuda part.

Best regards,
Louie

Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows 10

From: Simon Rit<mailto:simon.rit at creatis.insa-lyon.fr>
Sent: Thursday, January 5, 2017 17:58
To: Chao Wu<mailto:wuchao04 at gmail.com>
Cc: rtk-users at public.kitware.com<mailto:Rtk-users at public.kitware.com>
Subject: Re: [Rtk-users] Type of CudaDataManager::m_BufferSize

Hi,
I have modified it, check that this is enough:
https://github.com/SimonRit/RTK/commit/ef23a0668a70c8a874eb2e2718e5f30b41f27f14
Thinking about this, I think it fixes one issue encountered by the user louie <ghostcz at hotmail.com<mailto:ghostcz at hotmail.com>> with large images (that I discussed with him during the training last June). Maybe he could confirm if he still follows the project...
Thanks again,
Simon

On Thu, Dec 15, 2016 at 12:15 PM, Chao Wu <wuchao04 at gmail.com<mailto:wuchao04 at gmail.com>> wrote:
Hi all,

In utilities\ITKCudaCommon\include\itkCudaDataManager.h the type of CudaDataManager::m_BufferSize is unsigned int, which prevents buffering >4GB image data in CUDA memory. In the same file, the type of GPUMemPointer::m_BufferSize is size_t which looks more proper.

Is there any special reason for this unsigned int type? I am going to change the type from unsigned int to size_t and test what will happen, but if anyone already have the experience please let me know. Many thanks.

Best regards,
Chao

_______________________________________________
Rtk-users mailing list
Rtk-users at public.kitware.com<mailto:Rtk-users at public.kitware.com>
http://public.kitware.com/mailman/listinfo/rtk-users


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/rtk-users/attachments/20170105/08ec9c2d/attachment.html>