Figure 1 AI image set showcasing FLUX.1 on an AMD Radeon™ RX 7900 XTX graphics.
Amuse 2.1 Beta is now available for Ryzen™ AI and Radeon™ GPU users and can be downloaded from TensorStack’s website. The new Amuse 2.1 Beta brings support for the newly launched FLUX.1 Schnell model on applicable AMD hardware. It also includes various bug fixes and even more Stable Diffusion models for users for enhanced experiences.
FLUX.1 Schnell is a state-of-the-art image generation model launched by Black Forest Labs. It can create stunning images in just 4 steps and is supported on the GPU through Amuse. Support for popular Stable Diffusion1.5 based Latent Consistency Models, like Dreamshaper v8 has also been added in this update.
Figure 2 AI image set showcasing FLUX.1 on an AMD Radeon™ RX 7900 XTX graphics.
Amuse 2.1 Beta also includes an update that should improve the image quality for all models as well as bug fixes that were impacting users on older hardware. While the software is still very much in a “beta” state and provided as is, the experience should be more stable with this update.
- Click here to download Amuse 2.1 Beta
- Recommended Driver: AMD Software: Adrenalin Edition™ – 24.8.1
In order to run FLUX.1 Schnell on a Radeon™ RX 7900 XTX GPU, head over to Advanced Mode and download the FLUX.1 Schnell model from the gallery. You can then generate images using the “text to image” tab. Please note that FLUX.1 Schnell is a very large model and your generation times will vary based on your system configuration and are affected by components such as drive speeds, memory size, speed and configurations etc. FLUX.1 support in Amuse 2.1 Beta is experimental and feedback is welcome.
Figure 3 AI image-set showcasing FLUX.1 on an AMD Radeon™ RX 7900 XTX graphics.
Full Change log:
- Preliminary and experimental support for FLUX.1 added (Advanced Mode only).
- 23 various stable diffusion models added to the model gallery.
- Image quality improvement for scheduler.
- Fix incorrect device display order in some cases.
- Fix for users affected with corrupt performance counter registries.
- Minor bug fixes affecting users with older hardware.
System requirements
|
Model |
Recommended Hardware |
|
New LCM and SDXL Turbo models |
AMD Ryzen™ AI 300 Series, AMD Ryzen™ 8040 Series or AMD Ryzen™ 7040 Series processors |
|
FLUX.1 Schnell |
AMD Radeon™ RX 7900 XTX, AMD Radeon™ PRO W7900 and AMD Radeon™ PRO W7800 graphics cards Minimum system RAM requirement: 32 GB System Memory
Performance will vary based on drive and memory speeds. |
|
FLUX.1 Schnell (Minimum Memory mode) |
Radeon RX 7900 XT |
Performance Reference
AMD Radeon™ RX 7900 XTX
|
Model |
Prompt |
Resolution |
Inference Step |
Time Elapsed (sec) |
|
Flux_schnell-f16-onnx |
cat wearing a cape and t-shirt that reads “FLUX + AMUSE” |
1024x1024 |
4 |
49.9s |
Endnotes:
Please note that Amuse 2.1 is a Beta release and experimental software. It is being provided on an “as is” basis. As such, there may be bugs and the user experience may be unstable, or cause unintended behavior.
GD-97 - Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.
GD-220c - Ryzen™ AI is defined as the combination of a dedicated AI engine, AMD Radeon™ graphics engine, and Ryzen processor cores that enable AI capabilities. OEM and ISV enablement is required, and certain AI features may not yet be optimized for Ryzen AI processors. Ryzen AI is compatible with: (a) AMD Ryzen 7040 and 8040 Series processors except Ryzen 5 7540U, Ryzen 5 8540U, Ryzen 3 7440U, and Ryzen 3 8440U processors; (b) AMD Ryzen AI 300 Series processors, and (c) all AMD Ryzen 8000G Series desktop processors except the Ryzen 5 8500G/GE and Ryzen 3 8300G/GE. Please check with your system manufacturer for feature availability prior to purchase.
RX-1122 -Testing done by AMD Labs, as of August 2024, on system configured AMD Ryzen 9 9950X processor and AMD Radeon RX 7900 XTX,64GB RAM, Driver package 24.7.1. Microsoft Windows 11. System was tested using Amuse 2.0.20 beta: Quality with model Flux_schnell-f16-onnx – ‘low memory mode’ to generative text to image prompt: “cat wearing a cape and t-shirt that reads "FLUX + AMUSE"” System manufacturers may vary configurations, yielding different results. Performance may vary. RX-1122.