1 / 12

HETEROGENEOUS ARCHITECTURES

HETEROGENEOUS ARCHITECTURES. Heterogeneous Architecture. A modern platform has: Multi-Core CPU (s) GPU DSP processors etc. What is behind such an Evolution?. GPU is specialized for compute-intensive, highly parallel computation – exactly what graphics rendering is about -

shad-hansen
Download Presentation

HETEROGENEOUS ARCHITECTURES

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. HETEROGENEOUS ARCHITECTURES

  2. Heterogeneous Architecture A modern platform has: • Multi-Core CPU (s) • GPU • DSP processors • etc.

  3. What is behind such an Evolution? GPU is specialized for compute-intensive, highly parallel computation – exactly what graphics rendering is about - and therefore designed such that more transistors are devoted to data processing rather than data caching and flow control.

  4. CUDA Parallel Computing Architecture

  5. GeForce 8800 GTX

  6. Multithreaded program on CUDA C

  7. Matrix multiplication example

  8. Running on Intel Core2 Duo Processor P8600 and GPU NVIDIA GeForce 9600M GT

  9. Smooth Particle Hydrodinamics method The interpolated value of a function A at position r using the SPH methodology is:

  10. OpenCL Platform Model Host device –usually х86 processor One or moreOpenCL Devices: GPU, multi-core CPU, etc. Each device contain N compute units Each Compute Unit contain one or more processing elements • Processing elements execute instruction: • SIMD (Single Instruction, Multiple Data) – GPU , vector units inCPU • SPMD (Single Program, Multiple Data) - CPU

  11. Table with results for N-body example

More Related