Image processing and threading

I think you people are little bit aware of this issue which I am going to post. I am preparing to create 2D picture processing DLL which use typical Erosion/Dilation algorithms but I have to make this on the several million pictures in kind of a batch mode. I am talking about the compatibility of the 64 bit with the 32 bit. The reason in this is that I am new to the Intel Toolset so please first you advice me that which really I need. I have ordered the compiler and vTunes. I think I need the MKL and the thread checker. Please leave your thoughts.

I think this might be better deal with the visual computing forum. MKL TBB and IPP are built-in in the Intel proficient c++ compiler. Vtune is the outstanding tool for performance tuning for the batch mode runs captivating upwards of quite a lot of seconds. The openmp profiling library is important for the OpenMP, /Qparallel, and MKL threaded region scrutiny and it is with no the VTune. There is a parallel studio inspector has been advocate pending a latest product capable of running on Windows 7 as a heir to Thread Checker.

I have read one article from one of the web site I think it was drdobbs.com. In that title “Two Variations on Parallel Pipelines”. In this article it demonstrates the profit of using a fine crafted parallel pipeline which is designed for image processing. There is an option of the email on this site you can use that and even there is a doubt after reading this article.

When I came on this I have read the content which is posted by most of the user. Then I came to know that little bit of information regarding this. I am also facing this issue from last six days. But here I just want to say thanks to you. Because after trying this I mean your solution then I have tried this then it solves my issue. One more time thanks for sharing such an important thing.

I was just going to create another thread but when I have read this then it sounded similar to what I wanted to write. I have a disk on that there is a data of around 1.9 MB of jpeg images each. I have read this in a sequence one by one that doing some image processing from image to image. So can anybody please advise a competent way to read those images through disk? I came to know that it is very slowest part in the application. I have an idea which I want to share with you it is that using TBB to thread the read but a few group declared that might not help. So can anyone tell me that Anything in IPP that would be useful? Or I have to do image slicing.

Some month around two to three months back I have also faced this issue. Regarding this I was very worry and always wanted to know that how to solve this issue. But here you would likely require a RAID array or SSD before you could evaluate where additional presentation might be increased by software parallelism.

Erosion/Dilation on million of pictures why not generate a big picture and do a collage through giant pictures including batches of these pictures particularly if they are of the similar dimension, if not, it is simple to create them by adding pixels. The trouble then fittingness is for the GPU. Some troubles are more appropriate for the CPU but this difficulty I consider can be deal more proficiently in the GPU.

May be you are referring to a movie most likely it is a movie that makes a lot of pictures, then you have to be sure the same dimension. I believe creation of a collage of the frames in a giant frame in the CPU in Parallel feeding it toward the GPU and after that returning back the output and allow the CPU dismantle the giant frame in Parallel once more you can accomplish real time results through some overhead of making the collage and dismantling it. This works for the reason that the SE operates in a local area and the operation wants the original pixel elements so this creates the entire process totally independent. So I think threads in the GPU can supremely work on every pixel with no caring what the other threads are doing.