Ethereum Mining Rig

IMG_2932

After weeks of planning and acquiring all the necessary hardware, I finally completed assembling my first Ethereum mining rig. This is no walk in the park. Everything I learnt about building a gaming PC rig was upturned. I had to un-learn and re-learn a lot on what it took to build a PC. Building a computer system with 5 graphics cards for a highly specific purpose of mining Ethereum requires a different set of skills. So how do I get from building gaming PCs for leisure to building mining rigs to make some cash?

I first learned about cryptocurrency about a month ago. I received chatter on social media and on the news online about the rise of cryptocurrency. Of course, the whole concept of cryptocurrency isn’t new. Bitcoin has been around for quite some time. But it remained a very obscure notion to a lay person. Mine block chains and get rewarded for it in a form of electronic currency? Sounds puzzling.

Nonetheless, I soldiered on and read as much about cryptocurrency as I can. I watched various YouTube videos on what block chains are, how it works, how it is applied and the potential usefulness in the future. Along the way, I was surprised that there are several dozen forms of cryptocurrency floating around online! It comes in weird and futuristic sounding names, promising miners and holders of these digital currencies of a surging value in the near future when its use becomes widespread and mainstream. However, as I wrote this entry, the most popular and valuable (in terms of market capitalisation) are Bitcoin, Ethereum, Zcash, Dash, Litecoin and a just couple of others.

From then on, I was intensely curious as to how to build one. Again, I scoured the internet and read up on how to build a multi-GPU PC system. And when I said multi-GPU PCs, I am referring to 6 to 8 GPUs in a single system.

It actually sounds pretty straightforward. You are simply building a PC without a case, (since you can hardly find a PC case large enough to fit all 6 or 8 GPU inside) buy a bunch of GPUs, a compatible motherboard, the cheapest CPU, RAM and SSD you can find. The only difference is that you need to buy a bunch of GPU risers, which allows you to connect your GPU indirectly from the motherboard’s PCI-E slots. This way, you can slot more GPU within the PCI-E slots without cramping the motherboard’s real estate. Oh, and don’t forget a simple on-off switch that connects to the switch jumpers on the motherboard to switch your computer on and off.

Getting the GPUs was challenging! There was a massive shortage of GPU everywhere. I mean it, literally, everywhere. At the moment I was prepared to get all the hardware, it seemed that everyone else was doing it as well. I wasn’t the only crazy one to invest 2.7K on a new PC that does nothing but mine for crypto coins. But eventually, I did get those darn GPUs. It took a while, as all my GPU were on pre-order status and I had to wait for it to arrive. But I finally got 4 units of GTX 1060 mining GPU and 1 unit of an ordinary GTX 1060 GPU for general purpose use. The reason why I got just 1 ordinary unit of GTX 1060 was for video output capabilities. This is essential as I need to see what I’m doing with my PC, especially the first step of any PC building step after hardware assembly: OS installation. As for the rest of the GPUs, I got the mining GPU, cheaper and slightly more power efficient. It doesn’t have any video output at the back of the GPU, which resulted in some cost savings (but at the expense of resale value, as it only exist to serve one function: mine for coins)

I had difficulties with the driver installations as well. I didn’t fully realise that there is actually a certain version of graphics drivers that the mining GPU and the OS were able to recognise each other. I was close to giving up as I was at my wit’s end to figure out why Windows 10 couldn’t recognise the mining GPU, despite the fact that the mining GPU is basically a stripped down version of a Nvidia GTX1060 GPU. Zotac and Biostar weren’t helpful in providing information or FAQ about the correct driver installed.

But worked it did and it worked wonders. I was already familiar setting up the ethereum mining software (Claymore miner, mining block chains in a pool from Nanopool) by testing it on my gaming rig and letting my GTX 1070 churn through the cryptographic algorithm, solving complex puzzles. But the speed and productivity of a single 1070 pale in comparison to having a herd of 5 1060s working in tandem. My hash rate shot up through the roof after I got my mining rig running.

Now everything is in order and quietly humming away in another room. The whole set up can make the room pretty warm though since you have using the GPU to work hard and mine the block chains. The rig is going to run 24/7, maximising my return. I am pretty confident nothing will go wrong, at least for the foreseeable future. The whole set up looks crude, since it’s an open setup, with all the cables exposed and crisscrossing each other. Cost saving was a major deciding factor when it comes to building this mining rig, so I wasn’t keen on spending any more money to make the whole set up look neat (specialised case, rigs, frames to neatly hold the GPUs together).

IMG_2933

I am still closely monitoring the situation now that it has been up and running for 24 hours. But looking at the Claymore console for ethereum mining, it seems everything is going really smoothly.

My ‘Dark Rage’ gaming rig is not only for gaming

OLYMPUS DIGITAL CAMERA

It is also doing good, productive work in the name of science. Every evening, after coming back from work, I would naturally switch on my gaming PC. On some nights, I don’t play games at all. On many occasions, my PC is just idling, or working minimally, while I browse the web and listen to my favourite music.

It is on these occasions, that I fired up BOINC. BOINC stands for Berkeley Open Infrastructure for Network Computing. What it basically does is that it acts as a hub to coordinate volunteer computer resources around the world. Rather tham having my PC idling, doing nothing, I made sure that my gaming PC is hard at work, receiving packets of information that requires computational power and mathematical number crunching either via the CPU or the GPU, and then returning those results back to their central servers.

For my case, I selected 2 projects, one taking advantage of the computer power of the CPU, while the other, the GPU. They are SETI@Home and World Community Grid. There are tons of other projects out there, from simulating weather pattern, to proving or disproving a mathematical theory or conjecture, to searching for larger prime numbers or deepening our understanding of quantum physics through mathematical modelling, there is definitely a project that one might be interested in participating.

SETI@Home is a scientific initiative aimed at listening and picking up radio signals out in space for signs of extra-terrestrial life. The amount of data that this projects received is tremendous and would take years for a single cluster of supercomputers to sieve out the signal from the noise (if there is a signal -indicative of intelligent life out there, in the first place). This is where distributed computing comes in. My computer that is idling, and millions around the world, are willing to contribute their spare computing power to process all these information from SETI@Home. The combined computational power of millions of computers pales in comparison to the largest supercomputer out there right now. Together, we can crunch those data into meaningful information much faster.

World Community Grid is a collection of biological and life science research projects that requires huge amounts of computing power, such as the search for compounds and ligands that would interact with a particular HIV protein, potentially finding a cure or a treatment. Malaria is another project that is underway, as well as Zika and cancer research. Finding a potential compound, ligand or simply understanding how it interacts with a target protein at a molecular level requires a lot of mathematical modelling and simulations that only, through a network of computers, would achieve tangible results faster that any supercomputer out there. And the cost is almost nothing to the researchers, as they need not maintain and manage expensive clusters of computers. They simply seek volunteers, like me who would be willing to contribute spare computing power for the greater good.

untitled

I have been doing this for 3 months, contributing anywhere between 15 minutes to several hours, sometimes leaving the computer overnight for at least once a week. I have made progress through the number of points I received based on the number of work units done. So far based on the stats taken from http://stats.free-dc.org which keeps tracks on users progress on a number of projects, I am currently in top 200 (out of 7300) contributors within my country, Singapore, in contributing to SETI@home. I am in the top 400 in my country in contributing to World Community Grid. I joined Crunching@EVGA, a group with members that collectively contributes their computing power under one banner (presumably these members also have some from of EVGA products, most likely EVGA graphics card). Within the team, I am in the top 50 (out of 350) for contributing to SETI@home, and top 200 for the combined points received for contributing to various projects as a whole. Although comparing with volunteers from around the world, and on an individual levels there are tens of thousands of people ahead of me, I have made a lot of progress so far, especially when I am representing my country, Singapore.

I wish that more Singaporeans like me would participate in BOINC. I know that there is a sizeable community of PC gaming and gaming rig building enthusiasts in Singapore. If only more participate, we together could make an impact for the good of science. For me, I will continue to contribute whatever spare, idling computing power for BOINC and the projects I am passionate about (so long as my GPU and CPU continues to survive and function properly in the years to come. My gaming computer now does more than just gaming.

Tipping point

I am at a tipping point where I am very close to buying all the parts needed to build my very own DIY PC. I think I can safely say that the tipping point for that decision came about when I impulsively bought 2 EVGA GTX 1070 SC Gaming ACX 3.0 GPU at a relatively competitive price on Amazon. I saw a good deal on the website and decided to try my luck to buy and sell it here locally. Firstly, Singapore doesn’t sell EVGA GPUs. There is no official distributor for EVGA GPUs. But (and its a big but) EVGA GPUs are covered by an international warranty and they have pretty detailed RMA instruction from their official website should there be a need to return a defective GPU back to the manufacturer. Coupled with these two factors and on top of the fact that GPUs sold in the US are generally much cheaper, I decided to try my luck to buy 2 of those and hopefully make a tiny profit selling it back on the local market here. And if they do sell well, I might buy several more in the near future. I don’t need to make huge markups, just enough to cover the initial cost and pocket some little extra cash. Nothing lavish.

Personally, I love the EVGA GPU design. I think they have the best heatsink and fan design among the other GPU vendors. It is simple, angular, with little to none of the copper of nickel plated heatsink pipes protruding on the sides of the GPU. Their designs and colours are not too flashy either, but they do come with a very prominent LED-backed wording panel that displayed the EVGA brand and type of GPU that you are using. This is especially prominent if you have CPU case with a clear side, allowing you to show off your new GPU.

But that aside, the very fact that I purchased those 2 GPUs may have tipped my decision in building a DIY PC of my own. I actually spent close to SGD 1.5k just like that. In addition to the GPUs, I also bought 2, 480GB SSD, from OCZ, a subsidiary of Toshiba. Again, after much research, retail prices of SSDs here were surprisingly very expensive. And the variety of SSD brands and models were surprisingly thin. The SSD I bought is suitable for entry level, budget conscious users, like me, who wishes to just simply instal and boot up my OS and use it as a day to day storage device in my PC build. I realised that I did not need extra performance or speed in terms of read and write speeds. SSDs are naturally much faster than traditional hard drive. So unless I am using my PC to move large files, process thousands of images or various 4K video files all at once, I wouldn’t be able to notice any substantial improvements in a more professional grade SSD than what I would ultimately use for my PC, just a simple decent one for a fraction of the cost. This cost savings would then allow me to invest in other parts of the PC that I may find it more useful.

I have spent months researching about all the PC parts that are needed for my newly built DIY PC. I have reached a point where I know in detail most parts, devices and peripherals that would suit my needs and the variety of choices present in front of me. I know what to look out for in terms of cross compatibilities, and I now have the knowledge and technical know-how to confidently build my own PC and even make informed choices on which parts are the best for my current needs. The choices are endless, the permutations of parts you can select are infinite. But looking harder and studying the technical aspect of each computer part, all boils down to just a handful of items that not only be useful for me, but comes at just the right price. Everything else is just noise. At the end of the day, its not about investing in the best in the system, but investing in a system and suits you best.

Right now I have multiple builds written down in great detail, down to the exact costs that is going to take to build on. And I have even created multiple builds based on my numerous needs and wants. From HTPC setup, to mini ITX builds, each with various components mixed and matched (like I said, the permutations to mix and match are endless), each with its own set of pros and cons. But looking closer, a number of parts listed in those builds don’t differ much and that, by setting a budget and listing down your needs, you simply have to set your mind to listing the components you need to just a handful of them.

I have just listed my final build for my PC. I have already purchased some of the components listed online and should arrive in the later half of August. As for the rest that I intend to buy locally, there is no rush. I should be able able to start building my PC towards the end of August. I will post my final build soon and perhaps show pictures of all my purchases in future updates.