Prerequisites

Before setting up your miner, verify your hardware and software environment meets these requirements.

Hardware Requirements

Component	Minimum	Recommended
GPU VRAM	48GB (A6000)	48GB (A6000)
System RAM	32GB	64GB
Storage	100GB SSD	500GB NVMe SSD
Network	Stable broadband	Dedicated server connection

GPU compatibility: NVIDIA A6000 (48GB) is the reference hardware. Other GPUs with 48GB+ VRAM will also work.

Software Requirements

• Python 3.10 or later — earlier versions are not supported
• CUDA 12.x — must be compatible with your GPU driver
• Git — for cloning the repository
• pip — for installing dependencies

Verify your CUDA version:

nvidia-smi
nvcc --version

Bittensor Wallet Setup

You need a Bittensor wallet with a coldkey and hotkey to participate in the subnet.

Create a new wallet (Bittensor is installed as part of the project dependencies — no separate install needed):

btcli wallet new_coldkey --wallet.name my-wallet
btcli wallet new_hotkey --wallet.name my-wallet --wallet.hotkey my-hotkey

Check your wallet:

btcli wallet list

TAO Registration

You must register your hotkey on the BlockZero subnet to participate. Registration requires a small amount of TAO (the Bittensor network token).

btcli subnet register --wallet.name my-wallet --wallet.hotkey my-hotkey --netuid <BlockZero_UID>

Check the current registration cost:

btcli subnet info --netuid <BlockZero_UID>

warning

Registration costs TAO. Check the current recycle cost before registering. Registration is permanent for the lifetime of your hotkey on this subnet.

Optional: Weights & Biases

For training monitoring, you can set up a free Weights & Biases account. Enable with wandb: true in your config file.

pip install wandb
wandb login

Checklist

Before proceeding to Installation:

• GPU with ≥48GB VRAM confirmed (A6000 or equivalent)
• Python 3.10+ installed
• CUDA 12.x installed and working (nvidia-smi shows GPU)
• Bittensor wallet created (coldkey + hotkey)
• TAO obtained for registration
• Subnet registration completed