Asiccn miner
What about ASIC? It is a more optimized circuit than the CPU mentioned above. There is no specific definition, but there is a rule in the field of chip design: the more general computing platform, the less efficient it is to complete specific computing. ASIC is the most dedicated computing platform. If we understand this, we can come to such a basic conclusion: in the mining field, if we use the computing power proof mechanism, as long as the CPU can dig, ASIC can dig. No matter what algorithm is used, ASIC can dig. There is no reason that CPU can dig, but ASIC can't, So the biggest difference between GPU and ASIC is their specific computing power.
The main differences between 7Nm and 10nm are as follows:
1, gate length is different. The width of the gate formed on the CPU is also called the gate length. The 7Nm process can integrate more transistors into CPU and GPU, and make the processor have more functions and higher performance
2. Different power consumption. For 7Nm technology and 10nm technology, the volume of 7Nm will be smaller when the same number of transistors are plugged. The large volume of 10nm, because of the process problems, will lead to the larger capacitance of the original, the required voltage is higher than 7Nm, resulting in higher overall power consumption
performance:
the chip is composed of transistors. The smaller the process, the more transistors in the same area of the chip, and the stronger the natural performance. The performance of 7Nm is better than that of 10nm
take Huawei Qilin 980 as Qilin 970, where Qilin 980 is a chip with 7Nm process and Qilin 970 is a chip with 10nm process
first look at the number of transistors, Kirin 980 has 6.9 billion transistors, Kirin 970 has 5.5 billion transistors, an increase of about 25.5%. In terms of performance, it's not as simple as 25.5%, because it involves not only the number of transistors, but also the upgrade of CPU, GPU, NPU and other IP cores
in terms of specific values, for example, the running points of CPU, Qilin 980, GPU and NPU are about 50% higher, twice as high
extended data:
integrated circuits have two main advantages for discrete transistors: cost and performance. The low cost is e to the fact that the chip prints all the components as a unit through photolithography, rather than making only one transistor at a time. The high performance is e to the fast switching of components, which consumes less energy, because the components are small and close to each other. In 2006, the chip area ranged from a few square millimeters to 350 mm, with a million transistors per mm
registered capital: RMB 40.45 million