Witam wszystkich.
Jutro muszę wygłosić krótką prezentację. Napisałem tekst i prosiłbym o sprawdzenie. Bardzo mi na tym zależy, bo w mojej grupie językowej jest bardzo wysoki poziom, a dla mnie każdy punkt jest na wagę złota. Dziękuję.
Hello everyone. My name is and I study computer science. In additional I work on a research project about a virtual bronchoscopy. Recently I have been working at a topic about computing optimization based on GPU and I am going to present you how fast and imporant gpu accelerated computing is.
GPU-accelerated computing is the use of a graphics processing unit (GPU) together with a CPU to accelerate scientific, analytics, engineering, consumer, and enterprise applications.GPUs are accelerating applications in platforms ranging from cars, to mobile phones and tablets, to drones and robots. Other known concept which can we meet in the Internet is GPGPU general-purpose computing on graphics processing units.
A simple way to understand the difference between a CPU and GPU is to compare how they process tasks. A CPU consists of a few cores optimized for sequential serial processing while a GPU has a massively parallel architecture consisting of thousands of smaller, more efficient cores designed for handling multiple tasks simultaneously.
A simple GPU application that is working looks like any other C code. BAcctually, you can type any C application using CUDA. The real magic happens when we start deciding what functions we want to execute on the CPU and what we want to execute on the GPU. Lastly we have to run proper compiler for instance nvcc provided by Nvidia. On the other hand writing a programs which compute a medical image could consume a lot of time, because for example processing of computed tomography have to take into consideration three-dimensional data model. In fact, parallelization of data from computed tomography is very desirable. Recently, I have convinced myself how fast gpu accelerated computing is.Thus I have wrote a program comparing processing data from computed tomography. The program convolved data with a kernel 3x3x3. The result have suprised whole team, because CPU have processed the data in 150 seconds but GPU have processed the data in 2 seconds.
What is more, Eric Smistad proposed that we can use GPU for processing airway segmentation or vessel segmentation. The implementation is tested on several datasets from three different modalities: airways from Computer Tomopgraphy, blood vessels from magnetic resonanse and 3D Doppler Ultrasound. The results show that the method is able to exstract airways and vessels in 3-4 seconds on a modern GPU.
In conclusion, in nowadays serious big data algorithms are written mostly on the GPU, because this is an incredible performence boost in parallelization. The last experience was eye-opening for me what opportuinies hides GPU-accelerated computing. As the utilization of medical imaging studies increases and the volume of scan data produced per study continue to increase, the development of new methods for distributing, viewing, and manipulating data is essential.