CoinWorld reported:
Kuzco is a specialized LLM (Large Language Model) computing power mining network. This year, it was selected for a16z’s Crypto Startup Accelerator (CSX) Fall Accelerator Program launched in New York. Projects selected for this program will receive a minimum investment of $500,000 from a16z over an 8-week period, as well as guidance and support from the a16z operational team.
Written by: J1N, Techub News
Introduction
Today, the author is sharing about Kuzco, a specialized LLM computing power mining network. It was selected for a16z’s Crypto Startup Accelerator (CSX) Fall Accelerator Program in New York this year. Projects selected for this program will receive a minimum investment of $500,000 from a16z over an 8-week period, as well as guidance and support from the a16z operational team. Currently, there are approximately 2 weeks left until the end of the program.
Kuzco is still in the very early stages and belongs to the same GPU computing power network field as io.net, but it is taking a different development direction. io.net, backed by strong funding and background, has already completed airdrops to users earlier than Kuzco and has successfully launched on mainstream cryptocurrency exchanges such as Binance, Coinbase, and Bybit, gaining greater market exposure.
I personally participated in mining with io.net, and to be honest, the whole process felt like “making money while your graphics card lies idle”. You just need to hang your graphics card and it will automatically generate income for you, without even spending much on electricity. Each mining machine consumes only 1.2 kWh of electricity per day, totaling only $10.8 in electricity costs over three months. With the depreciation cost of used graphics cards and the miscellaneous costs of internet tools, the cost of $100 to $200 can easily be covered. In the end, I made a profit of $4000. However, no matter how you look at it, this project feels like a “castle in the air” that rises with the left foot stepping on the right foot.
The subsequent price decline has verified this point. In addition, I have also compiled an article titled “70% Price Crash: How did the AI computing power rental bubble burst?” This article further reflects the bubble in the AI computing power market. Whether it is the price in the rental market or the decline in project tokens, the market is correcting the true value of these projects. This price adjustment is in line with market rules and indicates that the future may develop in a more rational direction.
Kuzco’s mining mechanism is significantly different from the popular project io.net. io.net is a “castle in the air” that easily generates tens of times the profit by hanging graphics cards, but the actual value behind it is not obvious. Kuzco, on the other hand, is more down-to-earth and has received support from a16z. I spent several weeks carefully testing different configurations of mining machines, including single-card, multi-card, and different system setups. Through these practical operations, I will share these experiences with everyone, allowing them to understand a mining project that provides real value to users, rather than a false prosperity supported by bubbles.
Specialized LLM Computing Power Mining Network
Kuzco is a decentralized GPU network in the Solana ecosystem. It aims to utilize idle GPU resources of individual users to provide efficient and cost-effective computing power services for large-scale language models such as Llama3, Mistral, and Phi3. Users can use these models through an API compatible with OpenAI. Kuzco’s distributed architecture can integrate idle computing resources to run large-scale language models while incentivizing miners who provide computing power through a reward mechanism.
Project Operation Status
As of October 21st, the Kuzco project has 2,000 online GPU mining machines, with a peak of 6,000 machines. The most commonly used GPU models include 3090 and 3060. Miners currently receive KZO Point rewards, but these points cannot be cashed out yet, and the project has not announced its tokenomics model. The author speculates that with the progress of the a16z accelerator program, the project may have more new developments and updates in the future.
Deployment Instructions
Official Hardware Requirements
The Kuzco project can run on Mac, Windows, and Linux operating systems, supporting various hardware configurations. The minimum system requirements are 16GB of RAM, 30GB of available disk space, and at least 10MB/s of network bandwidth. Kuzco supports NVIDIA (N card) and AMD (A card) graphics cards with at least 8GB of VRAM. The minimum requirement for ordinary users’ N cards is GTX750, while most professional mining GPUs are supported.
However, the author does not recommend using A cards for mining as AMD graphics cards have poor compatibility, especially in AI tasks where they perform worse than N cards. If you must use A cards, you need to check the official compatibility support list. In the field of AI, the author has to say: AMD, NO!
The author’s five-card platform, ASUS Z490 motherboard
The author tested the mining efficiency of several hardware models, but due to network fluctuations, the efficiency may vary.
GTX1070: 20 tok/s
RTX2060: 30 tok/s
RTX2070S: 40 tok/s
RTX3080: 80 tok/s
RTX4060Ti: 50 tok/s
RTX4070S: 70 tok/s
M2: 20 tok/s
M3: 30 tok/s
Unit: Average Tokens/Second (average calculation power in Tokens completed per second)
Real-time monitoring of mining machine operation
Deployment Methods
Kuzco provides a client application that users can download and use to start mining, but this method is sometimes unstable and may experience disconnections without automatic restarts. The author recommends a more stable approach, which is to use the Linux system or the Windows WSL (Windows Subsystem for Linux) environment to start mining through the command line (CLI) or Docker containers. If multiple graphics cards are needed, you can specify a specific GPU for multi-card mining in Linux using Docker containers. For example, to specify GPU0: “docker run –rm –runtime=nvidia –gpus ‘”device=0″‘ -d kuzcoxyz/worker:latest –worker ." this will start multiple GPUs.
When mining with multiple cards, please pay attention to the following hardware devices:
Power supply: The power supply is crucial and should not be skimped on. It is recommended to purchase a power supply based on the standard of "1 RMB = 1 watt" and use a gold-rated power supply if possible. When mining, choose a power supply of 1500W to 2000W depending on the number of graphics cards, or use multiple power supplies. However, multiple power supplies require additional startup cables to connect to the motherboard for normal power supply. Otherwise, the system will not start properly.
Power consumption information
Cables: In high-power environments, power cables are prone to damage, so the author recommends using higher-quality power cables. In addition, different brands of modular power supplies use different cable interfaces and are not interchangeable. Using different brands of cables may cause equipment damage or burning. Therefore, the cables must be compatible with the brand of the power supply to ensure compatibility.
Aging damage to power cables
Motherboard: Each channel of the motherboard (x1, x8, x16) can only support one graphics card. For example, the number of channels determines the number of graphics cards that can be supported. The B85 motherboard that was popular during the Ethereum mining period is a good choice.
The B85 platform left by the author when participating in Ethereum mining, already scrapped
CPU: The more threads the CPU has, the better because it needs to handle multiple tasks simultaneously. When starting mining with Docker, it will initially occupy a large amount of CPU resources. If using multiple graphics cards, they must be started one by one in order, otherwise the system may crash or freeze.
Young people should go for i9, no need to show off with e5
Graphics cards: One Docker process occupies approximately 6GB of VRAM (official documentation states 8GB, but in reality, 6GB is sufficient for operation). If the graphics card has 12GB of VRAM, you can run two Docker processes on one graphics card. During mining, the workload of the graphics card will occupy 50% to 90%, and the temperature of the graphics card should be kept below 85 degrees Celsius for it to be reasonable and safe.
PCIe riser cables: It is recommended to use PCIe riser cables that convert x1 to x16, with x1 inserted into the motherboard and x16 inserted into the graphics card. If using 40 series graphics cards, an x16 extension cable is required.
Network connection quality: Network connection quality has a significant impact on mining efficiency. The author's tests have found that using the Singapore network node receives more computing tasks than the Hong Kong node, which means that choosing a better network node can improve mining efficiency.
If there are any malfunctions during mining, first check the running status of the motherboard and software. If the problem is at the software level, it can be determined by checking the terminal error messages. It may be a memory error, requiring a computer restart, or it may be due to the official files being updated and the local mining machine not updating the code in time. The solution is to change the node or update the code.
Runtime errors
At the hardware level, if the mining machine cannot start, first check the fault lights on the motherboard. Taking ASUS motherboards as an example, the most common issue is the VGA white light, which indicates a problem with the graphics card power supply. In this case, you can try re-plugging the PCIE and graphics card power cables. However, sometimes the white light may illuminate but the machine still runs normally.
Conclusion
Although the author referred to io.net as a "castle in the air," implying that its market value was severely overestimated, it did successfully raise $40 million in financing with a valuation of $1 billion. However, after io.net went live, many copycat projects appeared. These copycat products and the backers behind them cannot withstand rigorous scrutiny, indicating that the success of io.net is not easily replicable by every project.
Based on this observation, the author has been looking for a mining project with more practical value. Finally, the author discovered Kuzco. Kuzco first received incubation and support from a16z, which increases its credibility and potential. Secondly, Kuzco's mining mechanism truly provides computing power services through GPUs.
Furthermore, from a macro perspective, the field where Kuzco is located, large language models (LLMs), is currently the most widely used AI product by the public. Countless people around the world use these models every day, and these models require tremendous computing power support. The author believes that providing computing power for such a huge demand is not only meaningful but also has practical business value. Therefore, Kuzco is a project worth paying attention to.
Additionally, the cost of participating in the GPU computing power network is relatively low, especially with the stable prices of 40 series graphics cards in the second-hand market and low depreciation costs. However, the author advises against renting graphics cards as the cost of renting is much higher than buying second-hand cards. At the same time, the airdrop incentives of Kuzco are unclear, so there is a higher risk if renting cards on a large scale.
Moreover, mining machines themselves have high practical value. They can not only be used to mine Kuzco but also provide more stable and reliable returns compared to investing in high-risk altcoins directly. Due to the scalability of mining machines, they can be used to mine other GPU projects or become validators for blockchain nodes, running scripts and services to generate additional income. This means that even if you stop mining Kuzco, the mining machine can still continue to generate value.
Lastly, many people ask how much profit can be made in a day from mining. There is no definite answer to this question, and besides the project team, no one can accurately know the final profit. Mining returns have a great deal of uncertainty and can be higher or lower than expected, so it is impossible to determine in advance how much money can be earned.
