The reason that I am having problems understanding this is because when I imagine crypto mining, I imagine basically an infinite command prompt constantly running. In my uneducated mind I don’t seeing this as a graphics drain I see it as taxing the gpu and ram. I am below beginner in my knowledge of computers. Thanks in advance!
In: 45
A graphics card is hundreds of tiny CPUs working together to solve a bigger problem
When being used as a graphics adaptor, they process maths and display the result on the screen.
When being being used for mining, it does the same thing, but the output is not a graphical output. the result is fed back to the mining program.
>I see it as taxing the gpu and ram
Assuming you meant cpu here.
The thing to understand here is that computing graphics involves doing loads and loads of the same set of relatively limited calculations. GPUs are optimized for this in their architecture. Think of it as wanting to solve a million simple math problems — who would complete the total workload faster, a decorated maths professor or several classrooms worth of high schoolers? The high school students win, because despite being less versatile their sheer parallel capacity gives them the edge. That’s the CPU and GPU respectively.
It just so happens that crypto also involves loads and loads of the same calculations, so GPUs are very well suited for it.
Cryptocurrency is based on blockchain, which relies on a verifiable chain of computational work to keep it secure. The blockchain network can provide more computation than a bad actor trying to change things so the right chain can be known because it is the longest.
In order to add on to the block chain a bunch of computation needs to be done, the vast majority of which is wasted guesses at the solution to a math problem. You can do this on a CPU but a GPU (graphics card) is designed to do many relatively simple math problems simultaneously. This makes GPUs a much better platform for all those random guesses than the more flexible CPU.
A GPU is optimized to run the same computations on lots of different data at once. For example doing trigonometry on all the 3D points in a scene or calculating the exact color for each pixel. While a CPU might have a handful of compute cores doing different computations a GPU have thousands of cores but they all do the same thing in sync. In cryptomining you are calculating the hash of a random number hoping to get the right result. The more times you can do this the more likely you are to win. And while a CPU core is usually much faster at this then a GPU core there are thousands of GPU cores so you can do all the hashes in parallel. This means you are kind of tricking the GPU into doing cryptographic hashes instead of graphics rendering. In the old days when concepts like this were starting you literally had to trick the GPU as you had to make your problem into a graphical problem that could be solved by the GPU. But pretty soon GPU manufacturers came out with drivers and toolkits to do any calculations on the GPU directly.
Several years ago an engineer at Nvidia decided to try using their graphics GPU as a parallel processor.
The GPU is designed to process large arrays of numbers and perform several types of calculations with those arrays.
He was successful. Thus the Tesla supercomputer was born.
This project showed that the GPU could be used for other purposes. The Tesla was a scalable architecture that could out perform any standard CPU based computer.
Since then, others have capitalized on this success by writing code that would run on the GPU for other purposes. So now, instead of buying a Tesla and spending all that cash, you can just buy graphics cards for certain application specific projects. Projects such as crypto mining.
Others have written other applications for this type of setup. But basically, it is mostly used for crypto mining.
When you render an image, you have to do thousands of little tasks in a short time frame all in parallel. Due to this, graphics cards excel in calculations where you have to do many things in parallel. CPUs are different and excel more in few, hard tasks.
Mining is the brute forcing of a solution, a lot of little tasks that you can complete in parallel.
Crypto mining is, typically, a very simple algorithm that involves guessing what the right input is going to be then running the algorithm and seeing if you were right. It takes a bunch of computer work to do it, but when you’re done you’ve either got the right answer and win some crypto coins, or the wrong answer and you get nothing.
A graphics card is a piece of hardware that does a neat trick: you give it a lot of different inputs and it runs the same algorithm on all of them at the same time.
For crypto mining, you can make a lot of guesses of what the right input is, then have the graphics card try them all at the same time. This is much faster than trying each guess one at a time.
Graphics card are like this because a lot of computer graphics processing has a similar “many inputs, one algorithm” kind of problem to solve.
Graphics cards used to be for displaying and drawing things only and for connecting a monitor. Early graphics cards were little more than memory (framebuffer) which contained the currently displayed image and made sure it was repeatedly sent to the monitor.
Over the years they became much more versatile. Modern GPUs (graphics processing unit) can do all kinds of mathematical operations. What distinguishes them from CPUs is that they can do those operations on lots of data in parallel. This makes them very suitable for some kinds of calculations. The graphics cards in crypto mining computers often don’t even have a monitor connector. They are purely used for mathematical/cryptographic calculations only.
Crypto mining boils down to mathematics with huge numbers. CPUs are not the best for this job because it takes a lot of clock cycles to multiply 2 numbers to begin with and they can only have up to 16-ish cores so they can do at most 16 calculations at a time. GPUs are specifically tailored to do just that: parallelised fast mathematics. Each multiplication is done in just a couple of clock cycles and and they have hundreds of cores so they can make hundreds of multiplications at the same time.
A lot of people already explained the difference between CPU and GPU, but I didn’t see one touch on your comment about “the infinite command prompt”.
The truth is that everything in the PC works together. When you play a game, the hard drive has to push data to the RAM, and the CPU does this, then it takes data that’s meant for the GPU and tosses into the GPU’s RAM, then the GPU processes it and throws it on the screen (as the graphics you see).
So you can’t just have a GPU, you need everything.
BUT the data the GPU processes does not necessarily need to be dumped on the screen. Crypto uses the GPU to process the data, and then sends the processed data back to the hard drive, re-encoded with the solutions. You don’t need to see that. All you need is for a prompt to tell you how the process is going, so just a tiny portion of the program will be for providing a visual event log of what’s happening, and that’s what you see as the “infinite command line”
Latest Answers