Blockchain GPU
Hardware
algorithm is difficult to solve in memory. In order to make DAG suitable for memory, each GPU needs 1-2gb of memory. If you get an error prompt: error GPU mining. GPU memory fragmentation? It means you don't have enough memory. GPU mining software is based on OpenCL, amd GPU will be faster than NVIDIA GPU of the same level. ASIC and FPGA are relatively inefficient, so they are blocked. To get OpenCL for chip integration platform, try:
amd SDK OpenCL
NVIDIA CUDA OpenCL
Ubuntu Linux settings
for this quick guide, you will need Ubuntu 14.04 or 15.04 and fglrx image driver. You can also use NVIDIA drives and other platforms, but you have to find your own way to get an effective OpenCL installation, such as genoil's ethminer fork
if you are using 15.04, go to & quot; Software and updates 〉 extra drives & quot; Set to & quot; Use the video driver & quot; from fglrx for AMD graphics accelerator
if you are using 14.04, go to & quot; Software and updates 〉 extra drives & quot; Set to & quot; Use the video driver & quot; from fglrx for AMD graphics accelerator;. Unfortunately, for some people, this method may not work, because there is a known program error in Ubuntu 14.04.02 that will prevent you from switching to the exclusive graphics drive necessary for GPU mining
therefore, if you encounter this program error, go to & quot; "Software and update" update & quot; Select & quot; Pre release reliable update proposal & quot;. Then, back to & quot; Software and updates 〉 extra drives & quot; Set to & quot; Use the video driver & quot; from fglrx for AMD graphics accelerator;. After restarting, it's worth checking that the drive is now properly installed (for example, by going to & quot; Additional drives & quot;)
no matter what you do, if you are using 14.04.02, once installed, do not change the drive or drive configuration. For example, the use of aticonfig – initial (especially the - F, - force option) will & quot; Destruction & quot; Your settings. If you accidentally change the configuration, you will need to uninstall the drive, restart, install the drive again and restart.
zilliqa project's token name is ZIL, which has been circulated and traded on 23 exchanges. Mainstream exchanges such as fire coin and coin an support it, so it is very convenient to participate. ZIL's maximum supply is 21 billion, of which 12.6 billion will be released directly through contracts, and the other 8.4 billion will be released in the form of mining awards within 10 years
the main network of zilliqa will be launched at the end of 2018 or the beginning of 2019, and the mining reward is rich, especially in the early stage. The electric power cost of zilliqa mining is only 1 / 9 of that of Ethereum, so it is more feasible with less energy consumption. In addition, mining can be carried out through GPU graphics card miner, and can simultaneously double dig ZIL and eth, bringing higher benefits for miners
pay attention to "zilliqacn" to learn more about the progress of the project.
The blockchain project is definitely the best one in terms of mining. This project is also the most popular and risk-free one in blockchain projects. Moreover, mining is also the cornerstone of blockchain and plays an important role in the blockchain instry
hayu miner is a good miner in the mining instry. You can experience hayu miner , and you must make a good choice of blockchain projects, otherwise you will lose all your money
Zilliqa's positioning is the bottom public chain , and its core competitiveness is to greatly improve the throughput of TPS (i.e. transaction processing speed) through fragmentation technology. In the latest experiment, the processing capacity of zilliqa test network reaches 2488 TPS, which is more than 100 times of Ethereum's maximum speed of 20 TPS and 355 times of bitcoin! Through fragmentation technology, when there are more nodes in the network, the transaction processing speed will be higher. When the number of nodes reaches tens of thousands, the transaction speed is very likely to reach the speed of visa, a centralized organization
zilliqa project's token name is ZIL, which has been circulated and traded on 23 exchanges. Mainstream exchanges such as fire coin and coin security support it, so it is very convenient to participate. ZIL's maximum supply is 21 billion, of which 12.6 billion will be released directly through contracts, and the other 8.4 billion will be released in the form of mining awards within 10 years
The main network of zilliqa will be launched at the end of 2018 or the beginning of 2019, and the mining reward is rich, especially in the early stage zilliqa mining power cost is only 1 / 9 of Ethereum, energy consumption is small, more practical. In addition, mining can be carried out by GPU graphics card miner, and it can double dig ZIL and eth at the same time, bringing higher benefits to minerspay attention to "zilliqacn" to learn more about the progress of the project
GPU is a chip with thousands of cores, while CPU usually has only 4-8 cores< You can understand CPU as 4-8 doctoral degree cores, while GPU is a core of 1000 or 2000 primary school students.
CPU can process instruction set and assign work to other cores in the computer. GPU can only deal with some of the simplest data, such as floating-point operations. If we use the examples in our life, it is to calculate 1 + 1, 1 + 2, and do the fast multiplication formula. If it's just this level of calculation, thousands of primary school students are certainly much faster than eight doctors
a GTX 1660ti can process 5.4 trillion floating-point numbers in one second. The top consumer CPU, i9900k, can only process 236 billion floating-point data in one second. We can see that the floating-point processing power of a middle end graphics card is even 23 times higher than that of the most high-end CPU
the screen of the game is split down, and the polygon data, mapping, action and physical calculation of the map are all floating-point calculation. These are just the specialties of GPU, so it's natural that image data is processed by GPU
to make a digression, the reason why GPU is used in mining is that blockchain computing is based on floating-point numbers. So we all use GPU to mine, not CPU.
In the current AI training scenario of data center, data migration between computing and storage has become a bottleneck, and a new generation of AI chip architecture based on 3D stack storage technology has become a trend. The demand for data bandwidth in AI chips will further promote the universal application of 3D stacked memory chips in AI training chips. And brain like computing chip will further promote its development in the search for more suitable applications. In the training scenario of data center, AI dedicated chips will challenge the absolute dominance of GPU. The AI chip architecture that can fully reflect domain specific will be more reflected in many edge scenarios
In addition to this, the biggest trend of 2019 is that simple deep learning is mature, and graph neural network combined with deep learning combines end-to-end learning with inctive reasoning, which is expected to solve a series of problems such as relational reasoning and interpretability that deep learning can not handle. The powerful graph neural network will be similar to the human brain formed by neurons and other nodes. Machines are expected to become AI with common sense, understanding and cognitive ability