A Study Of PCI Card Based On DM642

The subject of this design is come from a real item of Shanghai Isvision Science and Technology Company. The purpose of the item is to development a video collection card based on PCI interface. And this card is a main module in video surveillance system. Recently, the growing of applications with multimedia based on video signal is patency.For the purpose of base on present and face tomorrow, we choice the DM642 and H.264 to construct the card system. H.264/AVC is newest video coding standard of the ITU-T Video Coding Experts Group and the ISO/IEC Moving Picture Experts Group. The main goals of the H.264/AVC standardization effort have been enhanced compression performance and provision of a "network-friendly" video representation. H.264/AVC has achieved a significant improvement in rate-distortion efficiency relative to existing standards.TMS320DM642 is the highest-performance fixed-point DSP generation in the TMS320C6000 DSP platform. It based on the second-generation high-performance, advanced VelociTI very-long-instruction-word (VLIW) architecture developed by Texas Instalments, making these DSPs an encellent choice for digital media applications. With performance of up to 4800 million instructions per second (MIPS) at a clock rate of 600 MHz,The purpose of this design is the transplantation and optimizing of the coding with H.264 based on the DM642 platform, and makes it meet the requirement ofvideo application. These works relate to all kinds aspects of the platform based on DM642, and you must have a profundity understanding with the standardization of the H.264. So the design can obtain an expecting result only when you have a good understanding of both the standardization of H.264 and platform based on DM642.Transplant the H.264 code to the platform based on DM642 is the basic of this design. For we can optimize the code only we transplant it. The purpose of transplant is not only let the coding running in the platform, but also the collection of the video stream.The optimizing for H.264 code in the platform is very important for this design. There are all kinds of methods and utilities we used in design in this part. TI offers us a recommended code development flow. This flow involves utilitiing the C6000 code generation tools to aid in optimization tather than forcing the programmer to code by hand in assembly. These advantages allow the compiler to do all the laborious work of instruction selection, parallelizing, pipelining, and register allocation.In the first phase, we need only compile and profile native C code, and are concerned about nothing with the platform based on DM642. Moreover, we need make use of the profile utilities to find out range of code that consume CPU. At the end of this phase, if we can't achieve the desired performance in out code, we must move to the nest step. In the phase two, we concentrate on the optimizing of the C code. Now, we optimize our code with add restrict qualifier, loop iteration count, memory bank, data alignment, and using other C6000 instrinsics and other methods. Over this phase, we profile again and need to move to next phase if the code can't achieve the desired performance again. In the phase three, the majority of energy is write linear assembly. Development time of write linear assembly is greatly reduced from that of hand-coding, but the performance approaches the best that can be achieved by hand. All these we talked above are the process of the optimizing H.264 code in platform based on DM642.It's a great deal of hard work for me, much time and energy comsumed, and inthe end, I achieved the desired performance with the H.264 code that running in the platform based on the DM642.