GPU what is it in a computer? Mining on a GPU Graphics Card - A Complete Guide

If you can boast of a computer with a good video card, then you can start mining bitcoins right now. However, in order to increase profits and do this professionally, we recommend purchasing several video cards (optimally from 4 to 6) and assembling your own.

By installing several video adapters, you can increase the performance level of your computer, which will count much faster because... It is video cards that provide all the computing power for mining (for most algorithms).

Note that taking into account the current prices for video cards in 2019, cloud GPU mining has become more profitable, which cheaper than buying own equipment, much more flexible in the amount of investment and does not require maintenance of its farms. We're done with the rating best services based on the results of the last few years.

Material navigation:

How much can you earn on one video card?

Having some idea of ​​how, you should know which video adapters are best for mining. An excellent option would be AMD Radeon fifth series and above. Since the final amount of your earnings will depend on the performance of this device, it is recommended to acquire flagships.

There is an opinion that when using video cards manufactured by AMD, the calculation speed is slightly higher than that of analogues under the nVidia brand. But this greatly depends on the calculation algorithm on which each specific cryptocurrency is built. Accordingly, the more calculations per second, the more significant your earnings.

It is important to remember that if you decide to mine using one, even the top-end video card, its performance will not be enough. You will spend a lot of time before you get 1 BTC. For example, if the Radeon HD 7970 GPU is used, your result will be about 555 MH/s, and the daily production will be at the level of 0.0031 BTC, or 80 cents. It should be taken into account that during mining, electricity consumption increases. Therefore, this method of mining cryptocurrency is considered inappropriate.

Which video card to choose - performance on different algorithms

Knowing how to make money on a video card, the future miner faces a difficult choice between one or another video adapter, which must be purchased for a homemade farm.

GTX 1080

GTX 1050

The models given in the table pay for themselves much faster than their analogues and, judging by the reviews of experts, are best choice for mining in 2019. For example, GTX 1070/1060 and RX 480/470 will pay for themselves in 5-6 months. Also, do not forget that mining cryptocurrency is becoming more and more difficult every day, but its value is constantly growing, which allows you to maintain the necessary balance, attracting new people to mining.

The most popular for cryptocurrency mining are solutions from the red camp - AMD. This is due to the design features of video cards, which is why users primarily buy solutions such as Radeon RX 470 and higher. Nevertheless, today the gap in mining power for video adapters from Nvidia is not colossal, so such popular models, like GTX 1060 and above are selling like hotcakes.

By the way, there is a conspiracy theory about increased interest in cryptocurrency on the part of video card manufacturers, since Nvidia is preparing specialized video cards for mining called P104-100 and P106-100. These solutions differ from classic video adapters in that they do not have video outputs, are equipped with poor cooling and have a limited warranty. That's why to the average user It’s still more cost-effective to purchase a top-end solution in the form of the same GTX 1070, because it can always be sold to a gamer, thereby partially recouping its initial cost.

Calculation of the payback of video cards

As you can see from the table above, the most profitable for mining on at the moment are Nvidia video cards Geforce GTX 1060 3gb and Radeon RX 470 4gb. It is worth noting that this is only true at the moment and on the currently most profitable Equihash algorithm; what will happen next, what algorithms will appear in the future and how these video cards will behave on them is not known.

If you are interested in mining profitability right now, then you can choose the Geforce GTX 1060 if you are a fan of Nvidia or the Radeon RX 470 if you prefer AMD GPUs. They are used to calculate the profitability of mining (the data in them can vary greatly even within one day, since the exchange rate can change dramatically).

Overclocking video cards to increase hashrate during mining

Increasing the performance of GPU cards for mining is an integral part of setting up a farm. Overclocking can be done using special programs working from operating system, for this we recommend MSI Afterburner. It can also be achieved by flashing the BIOS.

In detail, there are also video instructions for AMD cards and Nvidia (their overclocking principles are slightly different).

It is worth noting that the warranty on these video cards is only 3 months.. Launching video cards for mining on Nvidia GTX expected mid-June, but timing may vary by supplier. The P104-100 graphics chip is claimed to have increased performance/watt by 30% compared to the GTX 1060 3GB. And the P106-100 chip gives a 10% increase in comparison with the same video card. Both video cards are released without video interfaces.

Nvidia P104-100 graphics card uses the same design as Nvidia GeForce GTX 1080. But it provides much more performance per 1 watt of electricity consumed, since the adapter is specially modified for mining on Nvidia GTX. Cards from different manufacturers come with different chip frequencies, while this base model runs on base frequency 1607 MHz. The boost frequency is 1733 MHz with 10 GB/s GDDR5X memory bandwidth and a 256-bit bus width.

Power is supplied via a single 8-pin connector and power consumption is approximately 180 W. Basic model The graphics adapter is planned to be shipped at a price of $350, but a model from the manufacturer Inno3D is announced at a price of $370 - Inno3D P6D-N104-1SDN P104-100 Twin X2 8GB GDDR5X. These prices are significantly less than their gaming counterparts ($499).

The stated mining performance on the Nvidia GTX P104-100 will be approximately 60 MH/s, but this performance will only be achieved after BIOS updates adapter for new firmware.

The Nvidia P106-100 model uses the same design as the Nvidia GeForce GTX 1060. This made it possible to customize graphics adapter mining on Nvidia GTX is much more efficient. The adapter operates at a base frequency of 1506 MHz, the turbo frequency is 1708 MHz, and throughput 8 Gb/s GDDR5 memory with 192-bit bus. Power will be transmitted through a 6-pin connector, and power consumption will be 120 W.

The price of the base model in the US market will be $200, which is $49 cheaper than the gaming analogue GeForce GTX 1060 6 GB. Here is the price of a specific Inno3D adapter (N5G-N106-4SDN P106-100 Twin X2 6GB GDDR5) – $235.

We all know that a video card and a processor have slightly different tasks, but do you know how they differ from each other in the internal structure? Like CPU central processing unit), and GPU (English - graphics processing unit) are processors, and they have a lot in common, but they were designed to perform different tasks. You will learn more about this from this article.

CPU

The main task of the CPU, speaking in simple words, this is the execution of a chain of instructions in the maximum possible time short time. The CPU is designed to execute several such chains simultaneously, or to split one stream of instructions into several and, after executing them separately, merge them back into one. in the right order. Each instruction in a thread depends on the ones that follow it, which is why the CPU has so few execution units, and the entire emphasis is on execution speed and reducing downtime, which is achieved using cache memory and a pipeline.

GPU

The main function of the GPU is rendering 3D graphics and visual effects, therefore, everything is a little simpler in it: it needs to receive polygons as input, and after carrying out the necessary mathematical and logical operations, output pixel coordinates. Essentially, the work of a GPU comes down to operating on a huge number of tasks independent of each other; therefore, it contains a large amount of memory, but not as fast as in a CPU, and a huge number of execution units: in modern GPUs there are 2048 or more of them, while like a CPU, their number can reach 48, but most often their number lies in the range of 2-8.

Main differences

The CPU differs from the GPU primarily in the way it accesses memory. In the GPU it is coherent and easily predictable - if a texture texel is read from memory, then after a while the turn of neighboring texels will come. The situation is similar with recording - a pixel is written to the framebuffer, and after a few clock cycles the one located next to it will be recorded. Also, the GPU, unlike general-purpose processors, simply does not need cache memory large size, and textures require only 128–256 kilobytes. In addition, video cards use more fast memory, and as a result, the GPU has many times more bandwidth available, which is also very important for parallel calculations operating with huge data streams.

There are many differences in multithreading support: the CPU executes 1 – 2 threads of calculations per processor core, and the GPU can support several thousand threads for each multiprocessor, of which there are several on the chip! And if switching from one thread to another costs hundreds of clock cycles for the CPU, then the GPU switches several threads in one clock cycle.

In a CPU, most of the chip area is occupied by instruction buffers, hardware branch prediction, and huge amounts of cache memory, while in a GPU, most of the area is occupied by execution units. The above described device is shown schematically below:

Difference in computing speed

If the CPU is a kind of “boss” that makes decisions in accordance with the instructions of the program, then the GPU is a “worker” that performs a huge number of similar calculations. It turns out that if you feed independent protozoa to the GPU math problems, then he will cope much faster than CPU. This difference is successfully used by Bitcoin miners.

Mining Bitcoin

The essence of mining is that computers located in different points Earth, solve mathematical problems, as a result of which bitcoins are created. All Bitcoin transfers along the chain are transmitted to miners, whose job is to select from millions of combinations a single hash that is suitable for all new transactions and secret key, which will ensure the miner receives a reward of 25 bitcoins at a time. Since the computation speed directly depends on the number of execution units, it turns out that GPUs are much better suited for executing of this type tasks than the CPU. The greater the number of calculations performed, the higher the chance of receiving bitcoins. It even went so far as to build entire farms out of video cards.

Hello everyone, GPU is the designation of a video card, or more precisely, a graphics processor. This word, well, that is, the abbreviation can often be found in some characteristics, well, for example, in the characteristics Intel processor There is such a thing as Integrated GPU, which means built-in video card. Well, that's right, it's actually built-in, the video chip sits right in the processor, this is not news, as it were

That is, we have already drawn the conclusion that the GPU is a video device. But what else is important to understand? I wrote that the GPU is found in the characteristics, everything is correct, but in addition to this it can also be found in programs that show the temperature. I think that you know such programs.. Well, or you don’t know, in short, in any case, what I’m going to write now will be useful for you to know. So we are talking about GPU temperature. Many people claim that the video camera can work at 80 degrees, but I declare that this is too high a temperature! And in general, I think that above 70 is not the norm!

By the way, GPU stands for Graphics Processing Unit

And here I am graphics chip, well, that is, the GPU, so I pointed it out on the board with arrows:

But what is the normal temperature then? Up to 60 degrees, well, a maximum of 66, well, 70 degrees is already the ceiling... But above that, I think that this is no longer very good, it’s just that such a temperature will definitely not extend the service life, do you agree with me? Well, there's more interesting point, in short, if the video card gets quite hot, then damn it also throws out its heat into the case, well, it obviously won’t be cool in it, and then the process will get hot, in short, fun! Remember that it is TEMPERATURE that can reduce the life of the device! Here on old motherboards from high temperature electrolytic capacitors exploded.. If you don’t believe me, you can look it up on the Internet for yourself..

A German researcher on the use of GPU computing in econophysics and statistical physics, including for analyzing information on the stock market. We present to your attention the main points of this material.

Note: The article in the magazine is dated 2011, since then new models of GPU devices have appeared, however, the general approaches to using this tool in the infrastructure for online trading have remained unchanged

Requirements for computing power are growing in various areas. One of them is financial analysis, which is necessary for successful trading in the stock market, especially with HFT funds. In order to make a decision to buy or sell shares, the algorithm must analyze a significant amount of input data - information about transactions and their parameters, current quotes and price trends, etc.

The time that will pass from creating a buy or sell order to receiving a response about its successful completion from the exchange server is called a round-trip (RTT). Market participants are doing their best to reduce this time, in particular, technology is used for this direct access to the exchange, and servers with trading software are located in a colocation facility next to the exchange’s trading engine.

However, the technological possibilities for reducing the roundtrip are limited, and after they are exhausted, traders are faced with the question of how else they can speed up trading operations. To achieve this, new approaches to building infrastructure for online trading are being used. In particular, FPGAs and GPUs are used. We wrote earlier about accelerating HFT trading using programmable hardware, today we will talk about how GPUs can be used for this.

What is GPU

When a C program using CUDA extensions calls a GPU kernel, copies of that kernel, or threads, are numbered and distributed to available multiprocessors, where their execution begins. For this numbering and distribution, the core network is divided into blocks, each of which is divided into different threads. Threads in such blocks execute simultaneously on available multiprocessors. To manage a large number of threads, the SIMT (single-instruction multiple-thread) module is used. This module groups them into “packs” of 32 threads. Such groups are executed on the same multiprocessor.

Analysis of financial data on GPU

Hurst exponent

Subsequently, the Hurst exponent began to be used in economics, in particular, in technical analysis to predict trends in the movement of price series. Below is a comparison of the speed of calculating the Hurst exponent on the CPU and GPU (acceleration indicator β = total time calculations on CPU / total calculation time on GPU GeForce 8800 GT):

Ising model and Monte Carlo method

Each vertex of the crystal lattice (not only three-dimensional, but also one- and two-dimensional variations are considered) is associated with a number called spin and equal to +1 or −1 (“field up”/“field down”). Each of 2^N possible options The arrangement of spins (where N is the number of lattice atoms) is assigned the energy resulting from the pairwise interaction of the spins of neighboring atoms. Next, for a given temperature, the Gibbs distribution is considered - its behavior is considered for a large number of atoms N.

In some models (for example, with dimensions > 1), a second-order phase transition is observed. The temperature at which the magnetic properties of a material disappear is called critical (Curie point). In its vicinity, a number of thermodynamic characteristics diverge.

Initially, the Ising model was used to understand the nature of ferromagnetism, but later it became more widespread. In particular, it is used to make generalizations in socio-economic systems. For example, a generalization of the Ising model determines the interaction of financial market participants. Each of them has a behavioral strategy, the rationality of which may be limited. Decisions about whether to sell or buy shares and at what price depend on a person's previous decisions and their outcome, as well as on the actions of other market participants.

The Ising model is used to model the interaction between market participants. To implement the Ising model and simulation, the Monte Carlo method is used, which makes it possible to construct mathematical model for a project with uncertain parameter values.

Below is a comparison of simulation performance on CPU and GPU ( NVIDIA GeForce GTX 280):

There are implementations of the Ising model using during analysis various quantities spins Multi-spin implementation allows you to load several spins in parallel.

Acceleration with multiple GPUs

In the case of a Monte Carlo simulation of the Ising model, the results indicate that performance increases almost linearly when adding more GPU

Conclusion

Many people have seen the abbreviation GPU, but not everyone knows what it is. This component, which is part of video cards. Sometimes it is called a video card, but this is not correct. The GPU is busy processing teams that form three-dimensional image. This is the main element on whose power depends performance the entire video system.

Eat several types such chips - discrete And built-in. Of course, it’s worth mentioning right away that the first one is better. It is placed on separate modules. It is powerful and requires good cooling. The second one is installed on almost all computers. It is built into the CPU, making energy consumption several times lower. Of course, it can’t compare with full-fledged discrete chips, but at the moment it shows pretty good results.

How the processor works

GPU is engaged processing 2D and 3D graphics. Thanks to the GPU, the computer's CPU is freer and can perform more important tasks. Main feature GPU that he tries his best increase speed calculation of graphic information. The chip architecture allows for greater efficiency process graphic information, rather than central CPU PC.

GPU installs location 3D models in the frame. Engaged in filtering triangles included in them, determines which ones are visible, and cuts off those that are hidden by other objects.