Most mining software prioritizes using the GPU for computation automatically. To ensure the GPU is utilized first, the most important step is to confirm that your computer's graphics card drivers are correctly installed and that the operating system fully recognizes the hardware. Furthermore, on Unix-like systems such as Linux, if there isn't a pre-compiled version of the mining program available, you may need to install some essential libraries to execute the subsequent compilation commands. The process might involve additional steps for mining software utilizing CUDA or AMD OpenCL, as more libraries are required. The benefit is that a natively compiled mining program often delivers better performance. This guide walks through the installation and usage for each platform, highlighting any extra libraries needed for Mac and Linux systems.
Mining on Windows: The Straightforward Approach
Using Windows for mining is the most user-friendly option. If you plan to use only the CPU, running GUIMiner on Windows 7 typically requires no special libraries (though Windows XP users must install Microsoft's Visual C++ 2008 Redistributable Package). GUIMiner performs particularly well with AMD graphics cards that support OpenCL; users with AMD cards should definitely use it.
Visit the GUIMiner website (guiminer.org) and click the Download link to get the software. The downloaded file is a self-extracting 7-zip archive. Run it, and it will extract to the current directory by default. Inside the newly created GUIMiner folder, double-click the main GUIMiner executable to launch it. If your installed graphics card does not support OpenCL, a warning message will appear. This does not affect your ability to mine with either the GPU or CPU; simply click OK to continue.
The main GUIMiner interface allows you to select a mining pool server, with many of the world's major servers pre-configured, including Slush's Pool. Enter your worker name in the Username field, the corresponding password in the Password field, and select the computing device from the Device dropdown. The CPU Affinity checkbox lets you select which processor cores to use. Finally, click Start Mining to begin. Enable the Show summary and Show console options from the View menu to access additional tabs with more detailed information.
There's no trick—mining in a Windows environment really is this simple. It's worth noting that running GUIMiner within a virtual machine may cause some features to malfunction. The Mac version of GUIMiner has a very similar interface, which we will cover in the Mac section. Next, we explore the process on Mac and Linux platforms.
For those looking to expand their mining capabilities across different setups, it's crucial to have the right tools. 👉 Explore more strategies for efficient mining
Mining on Mac OS X: Leveraging the Terminal
This section uses Mac OS X 10.8. While most Mac users prefer graphical interfaces, the software installation process requires the Terminal. Since the author's machines with CUDA-supported cards are Macs, both command-line and graphical mining software will be covered.
The demonstration for OS X involves compiling the poclbm mining software. Essentially, poclbm is the core engine behind GUIMiner, just without the graphical wrapper. Before compiling poclbm, you must install Xcode, Apple's development tools, and then enable command-line tools within it. After setting up this environment, you can begin the compilation and mining execution.
Install Xcode directly from the App Store; it requires approximately 2GB of disk space. Once installed, open Xcode, go to Preferences in the main menu, switch to the Downloads tab, and find the Command Line Tools item under Components. Click the Install button next to it; Xcode will handle the download and setup automatically. Next, install the PyOpenCL library, a Python module.
OS X 10.7 and above generally have a recent enough version of Python. The installation is done primarily via GitHub. Open Terminal and input the following commands sequentially:
git clone http://git.tiker.net/trees/pyopencl.git
cd pyopencl
git submodule init
git submodule update
python configure.py
python setup.py buildAfter entering these commands, Python will compile PyOpenCL, creating a package ready for installation. You can ignore any warning messages during compilation as long as the main process doesn't error out. Once the command prompt returns, run make to compile. After compilation, execute sudo python setup.py install. You will be prompted for an administrator password; enter it correctly to begin installation. The process is complete when you see "Finished processing dependencies for pyopencl==2013.1", confirming the installation of the January 2013 version.
With PyOpenCL installed, users with Macs containing AMD Radeon graphics cards can utilize them. NVIDIA users must first install the latest CUDA drivers, which can be updated later from System Preferences. After ensuring the graphics drivers are installed, we proceed to run the poclbm miner via Terminal.
Download the newest NVIDIA CUDA drivers from NVIDIA's Developer Zone. After CUDA driver installation, a CUDA preference pane will appear in System Preferences. To ensure the discrete GPU remains active, disable the "Automatic graphics switching" option in the Energy Saver preferences.
Again, fetch the poclbm source code from GitHub. In Terminal, run:git clone https://github.com/m0mchil/poclbm.gitcd poclbm/
These commands download poclbm and navigate into its directory. Now, perform the most crucial step: connecting to the mining pool to start calculating. Within the poclbm folder, input the command:
python poclbm.py http://YourWorkerName:[email protected]:8332 –verbose
Execution begins the connection and mining process!
When mining starts, poclbm lists all available computing devices and automatically selects the GPU as the calculation source. To specify a device manually, use the -d parameter followed by the device number. For instance, if device 0 is the CPU and device 1 is the GPU (the default), but you want to force CPU calculation, use -d 0. The full terminal command becomes:
python poclbm.py http://YourWorkerName:[email protected]:8332 –verbose -d 0
To verify poclbm is working, check if the terminal is active. A typical message shows the communication port with the pool host, time, device, calculation speed (hashes processed per second), and data reception status. To maximize a single computer's calculation power, open two terminal windows and run poclbm simultaneously for both the CPU and GPU.
While OS X has a simple GUI, using the Terminal for these tasks can be slightly complex. For users preferring a more memory-intensive approach, the OS X version of GUIMiner is an option. Its interface is nearly identical to the Windows version.
Monitoring the terminal's mining status messages shows continuous updates. An "accepted" message indicates a阶段性 block verification was correct. The last "Rej" (Rejected) item in the status shows the number of failed verifications. The numerator is the number of failed blocks, and the denominator is the total processed blocks (increasing by one with each "accepted"), resulting in a percentage.
The Mac GUIMiner interface closely resembles Windows. Select your pool server, fill in the Worker name and password, choose the device, and start mining. The status bar displays current data processing情况和 hashrate. GUIMiner can handle multiple pools simultaneously; if you have enough processor cores, different cores can interact with different pools. If the status bar information seems insufficient, enable Summary from the menu for complete results, showing speed, accepted, and rejected counts. The Console tab usually lists pool mining statuses, like discovering new blocks.
Mining on Linux: For the Advanced User
The Linux ecosystem has numerous distributions with varying hardware support, making graphics driver setup challenging (especially for NVIDIA CUDA drivers). Most Linux users might not have high-end graphics cards. Generally, operations on Linux are very similar to using terminal commands on OS X. The primary task is installing the necessary packages using the distribution's package manager.
For example, to install the GCC compilation tools on Ubuntu, use sudo apt-get install g++. The version control tool git isn't installed by default either, so execute sudo apt-get install git. Ubuntu will automatically connect to online repositories, download, and set it up. Similarly, install PyOpenCL (the package name is often python-pyopencl). This guide focuses primarily on the CPU mining tool cpuminer.
Every Linux distribution has its own package management tool. Tools or libraries required for mining can usually be installed via command. For instance, Ubuntu uses apt-get, while Fedora uses yum.
The essential Linux packages for cpuminer are: make, gcc, libcurl, and jansson. Installing them with the correct command, such as sudo apt-get install make gcc libcurl4-openssl-dev libjansson-dev, sets up the proper compilation environment. These packages are common and were successfully installed on both Raspberry Pi's Raspbian and desktop Ubuntu (both Debian-based).
After downloading cpuminer from its source, configure and make commands will produce the minerd executable. Use parameters to communicate directly with a mining pool and start mining.
Download the Linux/BSD version of cpuminer-1.0.2 and extract it. The commands are:wget http://yyz.us/bitcoin/cpuminer-1.0.2.tar.gztar xfv cpuminer-1.0.2.tar.gz
Navigate to the extracted folder, configure the build environment, and compile the executable:cd cpuminer-1.0.2./configuremake
After compilation, the main program is minerd. To connect to the example Bitcoin.cz pool, the command is:./minerd –url http://api.bitcoin.cz:8332 –userpass WorkerName:Password
Upon successful execution, cpuminer utilizes all available processor threads. A message reading "PROOF OF WORK RESULT: true (yay!!)" means a mining result was correct and accepted by the pool.
Hardware Performance Dictates Mining Results
Mining pool websites usually provide a web interface or API for miners to check their results. The BitCoin.cz interface is simple; earned rewards are visible under the Total reward field. Total reward is the sum of Unconfirmed reward (yellow) and Confirmed reward (green). In contrast, the red Estimated reward initially overestimates and then递减 based on actual计算 efficiency, making it less reliable.
The Total reward clearly shows that mining is currently hard work, with amounts accumulating very slowly.
The Workers section显示每个矿工是否正常运行，通过最后提交计算数据的时间（Last Share）判断节点是否存活。计算效能由 Mhash/s* 值查看，数值越高越好。Found blocks indicate finding new blocks (which grants extra Bitcoin), but for low-power test machines, this is highly unlikely.
For Linux environments without a browser, the author created a simple Ruby script (requiring the JSON and JSONPATH RubyGems) to access the BitCoin.cz pool's API. This API returns JSON-formatted results, which can be parsed into readable information.
From the personal page on BitCoin.cz, click Manage API tokens to see two APIs provided by the pool, both reporting mining operations in JSON format. You can write your own program to display desired information.
The author's experience involved投入 all available computing resources during the狂热 period. Some devices, like the Raspberry Pi, were clearly wasting electricity despite low power draw. The most surprising result was the sheer power of GPUs. Using older hardware—a 2010 Mac mini (Core 2 Duo 2.53GHz) and a 2010 MacBook Pro (1st-gen Core i5 2.4GHz)—the GPU's advantage was evident.
The Mac mini's GeForce 9400 averaged 1.85 MH/s (Mega Hashes per second), while its CPU managed only about 1.5 MH/s. The MacBook Pro's GeForce GT 330M achieved a much better 5.1 MH/s. This demonstrates a significant advantage for GPU mining, making it worthwhile to repurpose闲置的中高阶 graphics cards.
The author also tested mining on cloud platforms: Amazon AWS EC2, Microsoft Azure, and a VPS from DirectSpace (using xVM virtualization). The results weren't as good as physical machines but served to benchmark the providers' performance, as IaaS platforms rarely disclose exact specifications.
The finding was that, purely in terms of performance, the value proposition of AWS or Azure was inferior to some VPS services. Each core performed below 1 MH/s, yet costs were nearly $50 per month. In contrast, a sub-$10/month VPS provided three cores, each performing better than a single AWS core though slightly worse than Azure. This comparison is for performance reference only, as IaaS providers offer higher availability guarantees, justifying their higher costs.
Therefore, using cloud services for mining is currently unprofitable; monthly earnings wouldn't cover the service fees. The free AWS Free Micro tier showed very poor stability, with hashrates fluctuating between 140 KH/s and 400 KH/s due to shared resources.
Running cpuminer on an Azure Linux Small instance yielded a stable hashrate of around 830 KH/s per core, but it only had one core. Azure's interface is user-friendly. A Small instance costs about NT$1,340 per month. An Amazon AWS EC2 Large Instance was stable, with two cores each achieving about 640 KH/s, but it was expensive at roughly $175 per month. A mid-tier Azure Linux virtual machine with two cores cost only NT$2,680 per month, significantly cheaper than AWS.
Professional "Industrial" Mining Rigs
Seeing the prices for Amazon and Azure makes cloud mining seem unappealing. The author avoided testing the more expensive Amazon GPU EC2 service, priced over $3 per hour, as it would also likely be unprofitable. Investing that money into mid-to-high-end graphics cards seems wiser (unless you have access to idle supercomputing resources, though using them for mining might lead to repercussions).
However, investing in high-end graphics cards necessitates another crucial upgrade: the power supply unit (PSU). A high-end card cannot be powered by a generic 350W PSU that often comes with a computer case; a 600W+ unit is essential. Connecting multiple cards in SLI or CrossFireX requires a top-tier 1,200W PSU, which can cost as much as a graphics card itself. A high-wattage PSU increases power consumption, raising monthly electricity bills and potentially causing circuit breaks in rooms with old wiring if a homemade mining rig is started.
The only way to achieve zero electricity cost might be to mine secretly in a school lab or office (engaging in such high-risk, semi-electricity-theft behavior is done at your own risk).
Bitcoin's popularity has spurred the development of "industrial" mining equipment. While connecting multiple graphics cards is effective, it's power-hungry and complex to set up. Companies are now developing Application-specific Integrated Circuits (ASICs) designed for a specific task.
For mining, these circuits are designed solely to solve the 256-bit hash function, unlike the complex, power-intensive CPU or GPU. The most renowned company in this field is the US-based BUTTERFLYLABS. Their products are dedicated mining machines. An entry-level unit costs $274 USD and delivers 5 GH/s (Giga Hashes per second), potentially a better investment than graphics cards. Their highest-spec model boasts 1,500 GH/s but is currently sold out. Its price would be astronomical, given that a 50 GH/s model costs $2,499 USD.
This 30x more powerful machine could reasonably sell for $75,000 USD. The fact that the 1,500 GH/s model is sold out and on pre-order indicates strong investor confidence in Bitcoin's future. Singapore's Business Times also reports local companies developing ASIC miners, turning mining into an arms race.
BUTTERFLYLABS's website provides minimal instructions (mentioning USB connection to a computer and using their software). A foreign user's test of the 5 GH/s unit involved a Raspberry Pi running MinePeon to create a private pool. It achieved an actual output of 5.4 GH/s, 12.5% higher than advertised.
The reported daily income was around 0.25 Bitcoin, roughly $32 USD. If sold for profit, the machine's cost could theoretically be recouped in less than a month. This makes cloud mining options like Amazon's GPU service, costing at least $72 per day, seem very unattractive in comparison.
BUTTERFLYLABS's 1,500 GH/s mining rig comes with an intelligent操作面板, requiring just power and an Ethernet cable to start mining. Other models are also available for pre-order.
An Unprecedented Virtual Currency Opportunity
The rise of the virtual currency Bitcoin has created a new gold rush, enticing both companies and players (perhaps better termed investors) into the mining competition. Getting involved now, while mining is still relatively viable, represents a lower-risk approach compared to直接投入现金进入股市. Even if you invest in an ASIC-based mining rig, a mining computer can later be used for gaming or sold online.
Is Bitcoin suited for long-term holding or short-term speculation? A future report will discuss market investment in Bitcoin.
Frequently Asked Questions
What is the most important first step before starting to mine?
The crucial first step is ensuring your graphics card drivers are correctly installed and that your operating system fully recognizes your hardware. This ensures the mining software can properly utilize your GPU for optimal performance.
Can I use both my CPU and GPU to mine simultaneously?
Yes, it is possible to use both your CPU and GPU for mining simultaneously. This typically involves running two instances of your mining software—one configured to use the CPU and another the GPU—to maximize your computer's total computational power.
Is cloud mining a profitable option?
Generally, using cloud services like AWS or Azure for mining is not currently profitable. The monthly costs of these services typically exceed the value of Bitcoin you would earn through mining. Investing in dedicated hardware usually offers a better return.
What makes ASIC miners superior to GPUs?
ASIC (Application-Specific Integrated Circuit) miners are designed from the ground up for the single task of mining Bitcoin. They solve the specific SHA-256 hashing algorithm far more efficiently than general-purpose GPUs, delivering vastly higher hashrates for a fraction of the power consumption.
How do I know if my mining hardware is working correctly?
You can verify your hardware is working by checking the mining software's console output for "accepted" shares, which indicate successful calculations. You can also monitor your worker's status on your mining pool's website, which shows your hashrate and the last time a share was submitted.
Why is choosing a mining pool important?
Mining pools combine the computational power of all their users to increase the chance of successfully mining a block. The rewards are then distributed among participants based on their contributed work. Choosing a reliable pool with low fees and a stable connection is key to consistent earnings.