Support Matrix#
Supported Hardware#
Minimum System Hardware Requirements#
CPU Cores |
RAM (GB) |
NVMe SSD Storage (GB) |
GPU Count |
|---|---|---|---|
4 |
16 |
15 |
1 GPU from the list of Supported NVIDIA GPUs |
Note
The NVMe SSD Storage requirement of 15GB accounts for the RFdiffusion NIM image size plus additional space for Docker layer caching and temporary files during runtime.
Supported NVIDIA GPUs#
The RFdiffusion NIM requires NVIDIA GPUs with at least 12 GB of GPU Memory. The RFdiffusion NIM is configured to run on a single GPU and should run on any NVIDIA GPU that meets this minimum memory requirement and has compute capability >7.0. The following configurations have been specifically tested for optimal performance and stability. We recommend using one of the following tested configurations:
GPU |
GPU Memory (GB) |
Microarchitecture |
Precision |
GPU Count |
TensorRT Optimized |
|---|---|---|---|---|---|
H100 |
80 |
Hopper |
FP32/FP16 |
1 |
✓ |
A100 |
40-80 |
Ampere |
FP32/FP16 |
1 |
✓ |
L40S |
48 |
Ada Lovelace |
FP32/FP16 |
1 |
✓ |
A10G |
24 |
Ampere |
FP32/FP16 |
1 |
✓ |
Note
Starting with the 2.0 release, RFdiffusion NIM is optimized using NVIDIA Warp and NVIDIA TensorRT frameworks, providing up to 2x speedup. Pre-compiled TensorRT engines are included for A100, A10G, L40S, and H100 GPUs. When running on other GPUs with compute capability >7.0, TensorRT engines will be built at runtime.
Testing Locally Available Hardware#
Begin with Docker-supported operating system
Install Docker - minimum version: 23.0.1
Install NVIDIA Drivers - minimum version: 535
Install the NVIDIA Container Toolkit - minimum version: 1.13.5
Verify your container runtime supports NVIDIA GPUs by running
docker run --rm --runtime=nvidia --gpus 1 ubuntu nvidia-smi
Example output:
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 550.144.03 Driver Version: 550.144.03 CUDA Version: 12.4 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 NVIDIA GeForce ... Off | 00000000:01:00.0 Off | N/A |
| 41% 30C P8 1W / 260W | 2244MiB / 11264MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
+-----------------------------------------------------------------------------+