FPGAvs.CPU&GPU
Medical equipment used in CPU or GPU product program is also often found not see why FPGA has such remarkable results, even with alternative CPU and GPU "magic"? In fact, CPU and GPU are the Von Neumann architecture, FPGA can break through the restrictions on the structure and therefore has a strong energy efficiency.
Specifically, CPU and GPU requires the use of SIMD (Single Instruction stream Multiple Data stream) to perform memory, the decoder, the arithmetic unit, like the branch jump processing logic, FPGA has been determined at the time of programming the functionality of each logical unit there is no need instruction; Further, CPU and GPU is used in the shared memory, and therefore requires access arbitration, between the private cache so that the execution unit to maintain cache coherency between the members, also in the process of programming FPGA communication expressly requirements, there is no need for the shared memory communications.
Thanks to this, FPGA has a strong ability of floating-point multiplication, and comparison of the same GPU floating point lower latency. This is because, FPGA also has a pipeline parallelism and data parallelism, data parallelism and GPU only.
The force from the operator, the transition to Xilinx FPGA devices SoC will accelerate and adaptive. Xilinx achieved by accelerating the scalar engine, including ARM, real-time processor and an application processor, the core engine is adaptive the FPGA programmable logic device, it is also equipped with intelligence engine, now equipped with DSP. In particular, the platform will VersalACAP AI engine will have support to further accelerate and adaptive.
"In the medical field, such as an endoscope such applications, surgery patients have a common requirement is very low delay, or even need real-time to complete. Capture images from the camera, through pipeline processing, and then to the display may not 20 delicate time .CPU FPGA and GPU reach such a low latency, so this is the FPGA compared to CPU and GPU greatest advantage ", Subh continue to reporters, from power consumption, cost and integration on Xilinx the SoC FPGA also has a better advantage.
"A lot of areas, such as visualization, GPU to use for many years, FPGA is not impossible, but we will focus on areas of strength, that do data movement in an enclosed space, rather than memory Upload situation intermittent" Subh admitted.
FPGA applications in the medical analysis of different levels
It can be in the medical field, taking into account industry-leading AI latency and performance, extend the life cycle, only a Xilinx high quality, high reliability, high security, real-time, deterministic control and interface.
In addition to providing such Xilinx FPGA and SoC hardware devices and platforms, it also specifically to reduce FPGA development threshold to meet a wide range of market applications tailored to the needs of ViTIsAI unified software platform. I have many times before to introduce this software platform, engineers algorithms without hardware design experience, can also be directly applied algorithm implementation.
Sailive's medical solutions helped Illumina for genome analysis for severe newborns, accelerating the ICU patients and severe patients to promote Eyetech-based intraocular tracking communication tablet, and Merry to fight against new championships. FPGA adds a fear in an unwittime.
The author believes that the Saulith's FPGA device starts from high performance acceleration and adaptation, and has become the biggest competitiveness in the market. On the one hand, FPGA, ARM, application processor, real-time processor, DSP, AI engine use the film system (SOC) and software to perform high integration, which enhances the scalability of the application; on the other hand, the FPGA itself has The low delay is that it can be said to be "natural pair" for a very high medical field of delay.
From the market analysis, with the outbreak of the epidemic, the market demand in the medical equipment continued to increase, in which large data analysis and portability requirements are extremely high, this is just the characteristics of FPGASOC. On the other hand, the increase in the improvement of medical levels and the market Matthew effect, the demand for FPGA products with energy efficient and low power consumption continues to increase.
From the software analysis, the VITIS platform of Xilinx applies to different people, including hardware engineers who master HDL language, mastering software engineers with major programming languages, and is also suitable for skilled Tensorflow, Caffe, Pytorch algorithm engineers. With this flexibility, many creative startups have the possibility of displaying your fists.
Through the introduction of Xilinx, it can be said that both large equipment or portable equipment, FPGA has its own place.
How to develop in the future of Shanling Medical Innovation? SUBH said that in the medical product, Saulisi will continue to improve integration and reduce package size, and on the other hand, heterogeneous calculations increase efficiency and performance.
Thermal solutions for every industry is very important as the power is gradually higher and higher, Sinda Thermal can provide varieties heatsinks and coolers which inlcuding aluminum extruded heatsink, high performance heatsink, copper heatsink, skived fin heatsink, and heat pipe heatsink. please contact us if you have any questions about thermal solution.
website: www.sindathermal.com
contact:castio_ou@sindathermal.com
Wechat: +8618813908426
