There is still half a month of preparation time before Epoch Two begins.

Written by: J1N, Techub News

Introduction: Epoch One to Two

Kuzco is a network dedicated to LLM large language model computing power mining. This year, it was selected for a16z's Crypto Startup Accelerator (CSX) fall accelerator program, which launched on September 9 in New York. Projects selected for this program can receive at least $500,000 in investment from a16z and guidance and support from the a16z operations team. The accelerator program has now concluded.

On November 16, Kuzco announced that the first phase (Epoch One) incentive program will end on November 18, 2024, all operations will be suspended, data snapshots will be permanently stored, and the final point rankings will be published on a new leaderboard.

Official disclosures indicate that Epoch One launched on March 6, 2024, with a peak device count exceeding 8,000. The network runs Meta's 8B specification Llama-3 AI large language model, with a total inference of over 1 trillion tokens.

It was also announced that financing information and project development roadmap will be released in the coming weeks, and the second phase (Epoch Two) incentive program will begin on December 9. Epoch Two will introduce new features, such as higher throughput and reliability of NVIDIA hardware; encouraging users to connect top computing power devices like A100 and H100; and supporting more image generation and multimodal language models (VLM).

There is still half a month of preparation time before Epoch Two begins. This article will explore:

• Sharing personal mining practices and results, from single machine to cluster transition.
• Demonstrating the entire process of obtaining financing through research and practice, and building high-spec machines.
• Discussing the compatibility of hardware configuration with project requirements and answering common investor questions.

Epoch One Review: Solo Operations

Configuration

My configuration list includes RTX series graphics cards 2060, 2070S, 3080, 4060, 4060Ti, as well as 4 x 4070S and 2 Apple M2, M3 devices. These devices are distributed across several desktops, laptops, and a dedicated mining machine.

Costs

It is worth mentioning that these graphics cards were originally purchased by me annually based on gaming needs, not specifically for mining. Therefore, when calculating costs, I did not include hardware purchase expenses, only the actual electricity costs of the mining machine. Here, I will take the mining machine assembled in the first article "a16z 'Disciple' Kuzco Practical Guide: How to Efficiently Conduct AI Computing Power Mining?" as an example.

The configuration of this mining machine:

• Motherboard: z490 (later replaced with an industrial board)
• CPU: 10th Gen I9
• Graphics Cards: 2060, 2070s, 3080, 4060ti, 4070s

Hand-assembled mining machine

The following image shows the electricity consumption of this mining machine in October and November, totaling 564 kWh, earning approximately 600 million points (KZO Point). All machines combined earned about 1.1 billion points. The specific electricity cost needs to be calculated based on the local electricity rates, and this is provided for reference only.

At the far right, a total of 1 billion points earned

Preparing for Epoch Two: Cluster Deployment

Based on my sharing in the first article and my rich operational experience in assembling, debugging, and deploying the environment, I successfully secured some funding support and invested it all in assembling high-performance mining machines to further enhance computing power scale and operational efficiency.

From solo hand-assembly to cluster deployment

Configuration and Selection Logic for High-Spec Machines

Combining my practical experience in Epoch One, I comprehensively optimized the motherboard, CPU, graphics card, power supply, platform, and network configuration, selecting a more compatible hardware combination. This not only improved overall operational stability, safety, and efficiency but also emphasized the liquidity of the second-hand market in hardware selection. This strategy effectively reduces actual investment costs and provides subsequent participants with higher cost-performance options.

Motherboard

I chose an industrial motherboard instead of the mainstream B85, mainly based on a comprehensive consideration of performance, stability, and cost-effectiveness.

In terms of performance, running Kuzco's Llama-3 model requires starting multiple Docker processes, and running these processes in parallel consumes a lot of CPU resources, which places high demands on CPU performance. The CPUs compatible with B85 cannot meet this requirement.

Additionally, industrial motherboards have significant advantages in long-term stable operation, high-temperature resistance, and manufacturer warranty, while also having stronger liquidity in the second-hand market, making them the optimal choice.

Graphics Card

I chose to use the 4070S as the main graphics card, based on the following points:

Advantages in AI computing performance: Compared to the 30 series graphics cards, the performance improvement of the 40 series graphics cards in AI computing far exceeds the improvement in gaming performance. The core reason is that AI computing power mainly relies on the number of CUDA cores in the graphics card, and the 40 series graphics cards have significantly more CUDA cores than the 30 series.

Energy efficiency advantage: I conducted detailed tests on multiple GPUs and calculated the average power consumption per token.

• 4060Ti (160W): 0.125 Tokens/W
• 3080 (330W): 0.22 Tokens/W
• 4090 (450W): 0.26 Tokens/W
• 4070S (220W): 0.38 Tokens/W

From the test results, the 4070S performs best in balancing performance and power consumption, and its higher energy efficiency directly reduces electricity costs, making it the most cost-effective choice.

Price and liquidity in the second-hand market: As a mid-to-high-end graphics card, the 4070S has high liquidity and value retention in the second-hand market, further reducing the holding costs of the equipment and providing flexibility for future hardware upgrades.

CPU

As mentioned earlier, Kuzco's Llama-3 requires starting multiple Dockers during operation, which significantly occupies CPU resources, especially in multi-card operations where CPU usage can reach 80%-90%. Therefore, multi-core and multi-threaded processing capability is particularly important. A high-performance, multi-threaded, stable CPU can effectively support multi-tasking and ensure the stability and efficiency of the entire mining process.

13th Gen i5 can reach over 70% utilization under full load

Network Environment

Soft router is the square box in the image

The network environment is also crucial in mining. Even with high-performance graphics cards, if the network is not optimized, computing power can be severely affected. According to my tests, insufficient network speed can lead to a drop in computing power by up to 30%, and low-quality network nodes may directly prevent connection to the Kuzco network. Both of these issues are unacceptable for mining. To address these problems, I adopted a soft routing solution, which is not only easy to configure but can also run efficiently with minimal manual intervention after setup, theoretically supporting an unlimited number of devices. For specific operational methods, I recommend readers consult relevant materials based on their needs.

Power Supply

Classic Great Wall 2000W nuclear power supply

When selecting a power supply, special attention must be paid to peak power consumption. This is why, even though the rated power consumption of 7 x 4070S is only 1540W, I still chose to use dual 2000W power supplies, totaling 4000W. This is not a waste of resources but a consideration for the stability and safety of device operation.

Graphics cards can experience peak power consumption during operation, where at certain moments, their actual power consumption may reach 1.5 times or more than the rated power, before dropping back to normal levels. If the power supply is insufficient to handle these peaks, it may trigger the power supply's forced shutdown mechanism, potentially damaging the graphics card. This poses a fatal threat to the normal operation of the mining machine.

4070S power consumption performance

Taking the 4070S as an example, although its rated power consumption is 220W, its peak power consumption may exceed 400W. The combined peak power consumption of 7 graphics cards may exceed 3000W, so configuring dual 2000W power supplies is to ensure the stable operation of the machine. Users configuring multiple 4090s should pay particular attention, as a single 4090 has a rated power consumption of 450W, while its peak power consumption may reach 770W. In multi-card situations, relying on just two power supplies may not meet the demand, and typically three power supplies are needed to ensure system stability.

4070S power consumption performance

Supplement

Regarding BIOS settings, hardware compatibility, and remote management issues, I will not elaborate too much here. There are plenty of free tutorials available online for reference, and following these tutorials can resolve most issues. It is recommended to conduct targeted research and handling based on your hardware configuration and needs for simplicity and efficiency.

Risks and Returns

To answer the question everyone is most concerned about: How much can you mine per day? Frankly, there is no clear answer to this question, as risks and returns always coexist. I can share a clear perspective: whether in the cryptocurrency space or traditional industries, if any project can accurately calculate daily returns, then by the time you get in, you are likely no longer able to make big money. Unless you possess certain monopolistic resources, such as extremely low electricity costs or very cheap mining equipment, you can gain an advantage in returns. However, such resources are not available to everyone.

I chose devices with good liquidity precisely to reduce investment risks and cost pressures. Taking Kuzco mining as an example, costs mainly focus on hardware depreciation and electricity expenses, so your maximum loss is limited to these fixed costs. If you are not participating under low-cost conditions, then any investment decision loses its meaning. It is important to emphasize that the characteristics of mining early coins determine that there is no clear return expectation, but this is also where the potential of early mining lies.

From a subjective judgment perspective, this track has enormous market prospects: on one hand, Kuzco has received investment support from a16z; on the other hand, the demand for LLM large language models is rapidly expanding. Think about it, almost no one would do without LLM, right? Platforms like OpenAI's ChatGPT, Meta's Llama, and Musk's XAI have seen rounds of high financing, clearly indicating the growth potential of this industry.

For ordinary people, directly participating in the AI industry is not an easy task. On one hand, the technical threshold for AI is high; on the other hand, training AI models requires massive resources and funding, which most people cannot afford. However, by joining the AI computing power network through Kuzco, ordinary people can easily participate in this high-growth field under controllable costs, contributing to AI computing power while also earning returns.

Additionally, Bitcoin's price is currently about to break $100,000, rising from $16,000 in 2022 to its current peak, which carries significant retracement risks. If you choose to directly purchase tokens from AI projects, you will also face similar high volatility risks. In contrast, participating in the AI computing power network is a more robust choice: not only are costs clearly controllable, but it also allows for entry into the high-growth trajectory of the AI industry with relatively low risk. This is one of the practically feasible ways for ordinary people to enter the AI field in the current environment.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。