News Article:
Kuzco, a specialized GPU mining network that serves LLM (Large Language Model) computational power, has been selected for a16z’s Crypto Startup Accelerator (CSX) Fall Accelerator Program launched in New York this year. Projects selected for this program will receive a minimum investment of $500,000 from a16z within 8 weeks, as well as guidance and support from the a16z operations team.
Introduction:
Kuzco, the focus of today’s discussion, is a decentralized GPU network within the Solana ecosystem. Its aim is to utilize idle GPU resources of individual users to provide efficient and cost-effective computational power services for large language models such as Llama3, Mistral, and Phi3. Users can access these models through an API that is compatible with OpenAI. Kuzco’s distributed architecture allows for the integration of idle computational resources and enables the execution of large language models. Miners who provide computational power are incentivized through a reward mechanism.
Project Status:
As of October 21st, the Kuzco project has 2000 online GPU miners, with a peak of 6000 miners at one point. The most commonly used GPU models include 3090 and 3060. Miners currently receive KZO Point rewards, which cannot be cashed out yet, and the project has not yet announced its token economic model. It is speculated that with the progress of the a16z accelerator program, the project may have more developments and updates in the future.
Deployment:
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 a network bandwidth of at least 10MB/s. Kuzco supports NVIDIA (N card) and AMD (A card) graphics cards with at least 8GB of VRAM. The minimum requirement for ordinary N cards is GTX750, while most professional mining cards are supported.
However, it is not recommended to use A cards for mining as they have poor compatibility, especially in AI tasks where their performance is inferior to N cards. If you must use an A card, it is necessary to refer to the official compatibility support list. In the field of AI, it must be said: AMD, NO!
Monitoring Mining Operations:
Kuzco provides a client application that users can download and use to start mining. However, this method is sometimes unstable and may cause disconnections without automatic restart. A more stable approach is to use the Linux system or the Windows Subsystem for Linux (WSL) environment in Windows to start mining through the command line (CLI) or Docker containers. If multiple GPUs are required, multiple GPUs can be specified for multi-card mining in Linux through Docker containers. For example, to specify GPU0: “docker run –rm –runtime=nvidia –gpus ‘”device=0″‘ -d kuzcoxyz/worker:latest –worker ", multiple GPUs can be started.
When mining with multiple cards, the following hardware devices should be taken into consideration:
- 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 "1 RMB = 1 watt" and preferably a gold-rated power supply. For mining, a power supply of 1500W to 2000W is recommended depending on the number of graphics cards, or multiple power supplies can be used, but additional startup cables are required to connect the power supplies to the motherboard for normal power supply. Otherwise, the system will not start properly.
- Cables: In high-power environments, power cables are prone to damage, so it is recommended to use higher-quality power cables. Additionally, different brands of modular power supplies use different cable interfaces and are not interchangeable. Using different brands of cables may result in device damage or burning. Therefore, the cables must be consistent with the brand of the power supply to ensure compatibility.
- Motherboard: Each channel on the motherboard (regardless of x1, x8, x16) can only support one graphics card. For example, if there are several channels, they can support the same number of graphics cards. B85 motherboards, popular during the Ethereum mining period, are a good choice.
- CPU: As multiple tasks need to be processed simultaneously, the more threads the CPU has, the better. When starting mining with Docker, a significant amount of CPU resources will be initially occupied. If multiple graphics cards are used, they must be started one by one in order; otherwise, the system may crash or freeze.
- Graphics cards: Each Docker process occupies approximately 6GB of VRAM (official documentation states 8GB, but 6GB is sufficient in practice). If a graphics card has 12GB of VRAM, two Docker processes can run on one graphics card. During mining, the workload of the graphics card will range from 50% to 90%, and the temperature should be kept below 85 degrees Celsius for optimal performance and safety.
- PCIe riser cables: It is recommended to use x1 to x16 PCIe riser cables, with x1 connected to the motherboard and x16 connected to the graphics card. If using 40 series graphics cards, x16 extension cables are required.
Quality of network connection: The quality of the network connection has a significant impact on mining efficiency. Testing has shown that more computational tasks are received when using Singapore network nodes compared to Hong Kong nodes, indicating that choosing better network nodes can improve mining efficiency.
In case of any malfunctions during mining, it is necessary to first check the status of the motherboard and software. If the problem is at the software level, it can be determined by examining the terminal error messages. It may be due to a memory error, requiring a computer restart, or it may be caused by the failure to update the local miner code in a timely manner due to updates from the official files. In such cases, changing nodes or updating the code can solve the problem.
Conclusion:
Although io.net has been referred to as an "air castle," indicating a severe overvaluation of its market value, it has indeed raised $40 million in funding with a valuation of $1 billion. However, after its launch, many imitations of io.net appeared, and the products and backers behind these imitations could not withstand strict scrutiny, indicating that the success of io.net cannot be easily replicated by every project.
Based on this observation, I have been searching for a mining project with more practical value. Eventually, I discovered Kuzco. Kuzco has first received incubation and support from a16z, which adds credibility and potential. Furthermore, Kuzco's mining mechanism provides computational power through GPUs, which is a tangible value for users.
Moreover, from a macro perspective, large language models (LLMs) in the field where Kuzco operates are currently the most widely used AI products among the general public. Countless people around the world use these models every day, and these models require significant computational power support. I believe that providing computational power for such a massive demand is not only meaningful but also commercially valuable. Therefore, Kuzco is a project worthy of attention.
Furthermore, the cost of participating in the GPU mining network is relatively low, especially with the stable prices of 40 series graphics cards in the second-hand market and low depreciation costs, making it worth purchasing. However, it is advised not to choose card rentals as they are much more expensive than buying second-hand graphics cards. Additionally, Kuzco's airdrop incentives are not clear, so there may be higher risks in mass card rentals.
Furthermore, mining machines themselves have high practical value. They can be used not only for mining Kuzco but also for mining other GPU projects or becoming validators for blockchain nodes, running scripts and services to generate additional income. This means that even if Kuzco mining is no longer profitable, the mining machines can continue to generate value.
Finally, many people ask about the daily earnings from mining. There is no definite answer to this question, as no one besides the project team can accurately determine the final earnings. Mining returns are highly uncertain and can be either high or lower than expected, making it impossible to determine the exact amount of money that can be earned in advance.
