GPU chips continue to be out of stock, AI and mining are the culprits
Out of memory, CPU out of stock, and out of stock of hard drives will all be heard. However, in the past six months, the global GPU has been in a state of supply shortage. The first reason is that GPUs have now become the brains of artificial intelligence. The development of the artificial intelligence industry has brought a lot of demand; the second is that a large part of mining machines that mine cryptocurrencies are video processing machines. Lots of GPUs.
GPU chip
GPU out of stock prices rose
No one would have thought that the GPU, which is the core of game and image rendering, will appear to be out of stock for a long time, and the price will rise. According to some companies that do artificial intelligence, the latest GPU price has risen by 15% compared with six months ago, and it will be delayed by 4 weeks before delivery. This has had an impact on the research and development of artificial intelligence companies. Without more GPU chips, more calculations cannot be performed and the entire company's project schedule has been postponed.
AI is enterprise-level demand Mining is consumer-grade demand
Why does the GPU appear to be out of stock? The explosive growth of the artificial intelligence industry and the cryptocurrency mining industry are the main reasons for the lack of GPUs. Now the artificial intelligence industry, on the Internet giants such as Google, Ali, Baidu, down to a large number of start-up companies are involved in this, the rapid growth of artificial intelligence industry. What follows is a large demand for GPU chips.
Another reason that caused the GPU to be out of stock was the explosion of cryptocurrencies, which allowed miners to increase their investment in mining. According to statistics, crypto-money miners purchased more than 3 million GPU chips worth $776 million last year. The wave of buying GPU chips by miners has even forced Nvidia to limit the number of chips purchased each day by each company.
Why AI and mining require a lot of GPU chips
Why artificial intelligence and mining have chosen GPU chips for computing? Because the key performance of the GPU is parallel operations, it can handle operations at the same time. And you can break complex problems into simpler problems and then perform operations at the same time. For the artificial intelligence industry, GPUs can process large amounts of data at the same time, making them suitable for deep learning use scenarios. Compared with the CPU, the GPU consumes relatively less power and requires less space on the motherboard. Through intensive deployment, it can efficiently complete a large number of mathematical operations required by the neural network.
In terms of mining, because the calculation of cryptocurrency has a certain formula, only the formula can be calculated continuously to get the answer, that is, to mine. Compared with the CPU, the GPU has many cores. CPUs generally have four cores, eight cores, and sixteen cores. The GPUs have more than one thousand cores, and the computing speed is faster. Although the CPU can give an accurate answer, but the efficiency is not as good as GPU.
Is there a new chip that can replace GPU
Is GPU chip irreplaceable? At present, Google has created its own artificial intelligence chip - TPU. Compared with the traditional GPU (graphics processor)/CPU (scalar processor) combination, the TPU (tensor processing unit) has a 15-30 times faster processing speed, and Alpha Go uses a TPU chip. Facebook is also recruiting and wants to build its own AI chip. There are also companies that are designing chips for artificial intelligence services and investing them in more fragmented markets.
In terms of mining, the use of AS's miner's efficiency has exploded the GPU chip's mining machine, and the price is still cheaper. The intensive use of ASIC mining machines will gradually make the GPU mining mode escalate, which will ease the price increase of GPUs.
Conclusion:
The lack of GPU chips and the increase in prices did not expect AI and mining to become a "curse." However, this has also prompted the emergence of new AI and mining chips. Will make the GPU's stock price increase phenomenon has been alleviated. Similarly, for semiconductor companies this is a new opportunity to cut into the AI chip. The rise of AI chips has inspired another direction of China's layout.
GPU out of stock prices rose
No one would have thought that the GPU, which is the core of game and image rendering, will appear to be out of stock for a long time, and the price will rise. According to some companies that do artificial intelligence, the latest GPU price has risen by 15% compared with six months ago, and it will be delayed by 4 weeks before delivery. This has had an impact on the research and development of artificial intelligence companies. Without more GPU chips, more calculations cannot be performed and the entire company's project schedule has been postponed.
GPU mining machine
AI is enterprise-level demand Mining is consumer-grade demand
Why does the GPU appear to be out of stock? The explosive growth of the artificial intelligence industry and the cryptocurrency mining industry are the main reasons for the lack of GPUs. Now the artificial intelligence industry, on the Internet giants such as Google, Ali, Baidu, down to a large number of start-up companies are involved in this, the rapid growth of artificial intelligence industry. What follows is a large demand for GPU chips.
Another reason that caused the GPU to be out of stock was the explosion of cryptocurrencies, which allowed miners to increase their investment in mining. According to statistics, crypto-money miners purchased more than 3 million GPU chips worth $776 million last year. The wave of buying GPU chips by miners has even forced Nvidia to limit the number of chips purchased each day by each company.
Comparison of the number of CPU and GPU cores
Why AI and mining require a lot of GPU chips
Why artificial intelligence and mining have chosen GPU chips for computing? Because the key performance of the GPU is parallel operations, it can handle operations at the same time. And you can break complex problems into simpler problems and then perform operations at the same time. For the artificial intelligence industry, GPUs can process large amounts of data at the same time, making them suitable for deep learning use scenarios. Compared with the CPU, the GPU consumes relatively less power and requires less space on the motherboard. Through intensive deployment, it can efficiently complete a large number of mathematical operations required by the neural network.
In terms of mining, because the calculation of cryptocurrency has a certain formula, only the formula can be calculated continuously to get the answer, that is, to mine. Compared with the CPU, the GPU has many cores. CPUs generally have four cores, eight cores, and sixteen cores. The GPUs have more than one thousand cores, and the computing speed is faster. Although the CPU can give an accurate answer, but the efficiency is not as good as GPU.
Google launched
Is there a new chip that can replace GPU
Is GPU chip irreplaceable? At present, Google has created its own artificial intelligence chip - TPU. Compared with the traditional GPU (graphics processor)/CPU (scalar processor) combination, the TPU (tensor processing unit) has a 15-30 times faster processing speed, and Alpha Go uses a TPU chip. Facebook is also recruiting and wants to build its own AI chip. There are also companies that are designing chips for artificial intelligence services and investing them in more fragmented markets.
In terms of mining, the use of AS's miner's efficiency has exploded the GPU chip's mining machine, and the price is still cheaper. The intensive use of ASIC mining machines will gradually make the GPU mining mode escalate, which will ease the price increase of GPUs.
Conclusion:
The lack of GPU chips and the increase in prices did not expect AI and mining to become a "curse." However, this has also prompted the emergence of new AI and mining chips. Will make the GPU's stock price increase phenomenon has been alleviated. Similarly, for semiconductor companies this is a new opportunity to cut into the AI chip. The rise of AI chips has inspired another direction of China's layout.
评论
发表评论