NVIDIA A100 40GB PCIe

Rate $1.29 / hr. Eligible wallets are not charged. Runtime is one hour.

Specs

VRAM

40GB HBM2e

CUDA cores

6912

Tensor cores

432

Memory bandwidth

1555 GB/s

FP32

19.5 TFLOPS

FP16

312 TFLOPS

Used for

Training, LLM fine-tuning, Batch inference

Images on this GPU

PyTorch
PyTorch with CUDA, cuDNN, and JupyterLab.
pytorch:2.1.0-py3.10-cuda11.8.0-devel-ubuntu22.04
8888/http 22/tcp
TensorFlow
TensorFlow 2.x with CUDA and JupyterLab.
tensorflow/tensorflow:latest-gpu-jupyter
8888/http 22/tcp
LLM Inference
Text Generation WebUI.
pytorch:2.1.0-py3.10-cuda11.8.0-devel-ubuntu22.04
7860/http 22/tcp
CUDA
Ubuntu with CUDA toolkit. SSH only.
nvidia/cuda:12.1.0-devel-ubuntu22.04
22/tcp

Start

locked

Hold any amount of $GPU to start a treasury-funded session. Every session is capped at 1 hour.

Node name

Runtime 1 hour (fixed)

GPU NVIDIA A100 40GB PCIe

Ports 8888/http,22/tcp

Cost ~$1.29 covered if eligible

SSH public key

Use your public key only. Create a key locally with ssh-keygen -t ed25519, then paste the contents of ~/.ssh/id_ed25519.pub.

Connect a wallet on Robinhood Chain.