GPU computing power test

1. I don't know if you can open these two foreign web pages. These are the two pictures I got from Google http://www.crunchgear.com/2008/02/25/gpu-programming-now-on-osx/
http://www.tacc.utexas.e/research/users/features/dragon.php
GPU has strong computing power mainly because most of its circuits are arithmetic units, In fact, adders and multipliers are relatively small circuits, even if they do many such operation units, they will not occupy too much chip area. And because other parts of GPU occupy a small area, it can also have more registers and caches to store data. On the one hand, CPU is so slow because it has a large number of units for processing other programs, such as branch loops, and because CPU processing requires a certain degree of flexibility, the structure of arithmetic logic unit of CPU is also much more complex. In short, in order to improve the processing speed of branch instructions, many components of CPU are used to do branch prediction, and correct and recover the results of Alu when the branch prediction error occurs. These greatly increase the complexity of the device
in addition, the current CPU design is also learning from GPU, that is, adding floating-point operation units with parallel computing and not so many control structures. For example, Intel's SSE Instruction set can perform four floating-point operations at the same time, and many registers are added.

2. The 2400g Vega has no video memory, but many mining tools can't run directly with such integrated graphics cards, and even if they can, it's useless. People can connect at least six pieces with a rx560 machine, but 2400g can't be used together at all. A 2400g must correspond to a motherboard, which is much higher than the cost of graphics card.

3. Computing power is a concept put forward by NVIDIA when it released CUDA (Compute Unified Device Architecture, a programming language for GPU, similar to C programming for CPU). Because the graphics card itself is a floating-point computing chip, it can be used as a computing card, so the graphics card has computing power. Different graphics cards have different computing power. In order to show the difference, NVIDIA put forward the corresponding version of computing power x.x on the procts of different periods. Computing power 1.0 appeared on early graphics cards, such as the original 8800 ultra and many 8000 Series cards, as well as Tesla C / D / s870s cards. Cuda1.0 was released corresponding to these graphics cards. Today, computing power 1.0 has been eliminated from the market. Then there was computing power 1.1, which appeared on many 9000 Series graphics cards. Computing power 1.2 appears together with GT200 Series graphics card, while computing power 1.3 is proposed when upgrading from GT200 to GT200 A / b revision. In the future, there will be computing power 2.0, 2.1, 3.0 and other versions. The latest released version is computing power 6.1, which is supported by the latest Pascal architecture graphics card. At the same time, CUDA version is also updated to cuda8.0

ordinary users do not need to care about the computing power of the graphics card, only GPU programmers care about this problem when they write CUDA programs to develop GPU computing. As long as you know the model of your computer's graphics card, you can find the corresponding computing power https://developer.nvidia.com/cuda-gpus .

4.

It includes CUDA instruction set architecture (ISA) and parallel computing engine in GPU. Developers can now use C language to support CUDA; Architecture programming, C language is the most widely used high-level programming language. The program can then support CUDA & 8482; Runs at ultra-high performance on the processor. Other languages, including FORTRAN and C + +, will be supported in the future

with the development of graphics card, GPU becomes more and more powerful, and GPU optimizes the display image. It has surpassed the general CPU in computing. If such a powerful chip is only used as a graphics card, it would be too wasteful. Therefore, NVIDIA launched CUDA, which enables the graphics card to be used for purposes other than image computing

At present, only NVIDIA graphics cards on g80, G92, G94 and GT200 platforms can use CUDA, and the core of the toolkit is a C language compiler. G80 has 128 separate ALUs, so it is very suitable for parallel computing, and the speed of numerical calculation is much faster than CPU

The compiler and development platform in CUDA SDK support windows and Linux systems, and can be integrated with Visual Studio 2005

at present, this technology is in its infancy, which only supports 32-bit system, and the compiler does not support double precision data, which will be solved later. Geforce8cuda (Compute Unified Device Architecture) is a new infrastructure, which can use GPU to solve complex computing problems in business, instry and science. It is a complete GPGPU solution, which provides direct access interface to hardware instead of relying on graphical API interface to achieve GPU access

in the architecture, a new computing architecture is adopted to use the hardware resources provided by GPU, which provides a more powerful computing power than CPU for large-scale data computing applications. CUDA uses C language as programming language to provide a large number of high-performance computing instruction development capabilities, which enables developers to build a more efficient data intensive computing solution based on the powerful computing power of GPU< br />

5.

Yes, when NVIDIA designs and selects models, Ti has better performance than no ti. It can also be said that GPU has strong processing power. Sometimes in detail analysis, sometimes without ti is better. For example, in the figure below, the acceleration frequency and basic speed of Ti are better, but the overall performance of Ti is much better

READ MORE