Fix: VisoMaster Won't Launch on RTX 5090 / 5080 / 5070 — CUDA Compatibility Error

You just got your shiny new NVIDIA RTX 5090 (or 5080, 5070), installed VisoMaster, hit launch, and... a wall of red text.

If you're seeing an error message like this:

NVIDIA GeForce RTX 5090 D with CUDA capability sm_120 is not compatible with the current PyTorch installation.
The current PyTorch install supports CUDA capabilities sm_50 sm_60 sm_61 sm_70 sm_75 sm_80 sm_86 sm_90.

Followed by:

RuntimeError: CUDA error: no kernel image is available for execution on the device

Don't worry — you're not alone. This is one of the most common issues people run into after upgrading to an RTX 50 series GPU. The good news? There's a simple, one-click fix.

Why Does This Happen?

In simple terms, your GPU is too new for the software libraries that VisoMaster was previously bundled with.

NVIDIA's new RTX 50 series cards (RTX 5070, 5070 Ti, 5080, 5090, 5090 D) are built on the Blackwell architecture, which uses a new CUDA compute capability called sm_120. The older version of PyTorch that came with some VisoMaster packages only knows about older architectures (up to sm_90, which covers RTX 40 series cards).

When PyTorch tries to run a computation on your GPU and can't find the right instructions for your chip, it throws that "no kernel image is available" error and crashes.

Which GPUs Are Affected?

This issue affects all NVIDIA RTX 50 series desktop and laptop graphics cards, including:

• NVIDIA GeForce RTX 5070
• NVIDIA GeForce RTX 5070 Ti
• NVIDIA GeForce RTX 5080
• NVIDIA GeForce RTX 5090
• NVIDIA GeForce RTX 5090 D

If you have an RTX 40 series or older card, you won't see this error.

The Fix: VisoMaster All-in-One Package

We've updated VisoMaster's all-in-one package to fully support the RTX 50 series Blackwell architecture. The package comes with all the correct dependencies pre-installed — including a version of PyTorch that natively supports sm_120.

You don't need to:

• Manually install or update PyTorch
• Install CUDA drivers or toolkits yourself
• Touch any Python environments or command lines
• Downgrade your GPU drivers

All you need to do is download the updated package.

How to Get It

1. Visit the official VisoMaster website: visomaster.com
2. Download the latest all-in-one package from the homepage
3. Extract the downloaded zip file to any folder on your PC
4. Double-click VisoMaster.exe to launch

That's it. The new package detects your RTX 50 series card and runs correctly out of the box. No setup wizards, no environment configuration, no command line required.

What If I Already Have an Older Version Installed?

If you previously installed VisoMaster and it was working fine with your old GPU, but now you've upgraded to an RTX 50 series card:

1. Download the new all-in-one package from visomaster.com
2. Extract it to a new folder (don't overwrite your old installation)
3. Launch from the new folder

Your old installation won't interfere. You can delete it once you've confirmed the new version works.

Frequently Asked Questions

Will the new package still work with my RTX 40 series / 30 series card?

Yes. The updated package supports both older and newer NVIDIA architectures. Whether you have an RTX 3060 or an RTX 5090, it will work.

Do I need to install CUDA or cuDNN separately?

No. The all-in-one package includes everything. Just download, extract, and run.

I see the error on a laptop with an RTX 5080 Mobile. Does this fix apply to me?

Yes. Laptop variants of the RTX 50 series use the same Blackwell architecture and will benefit from the same fix.

What if I still get an error after downloading the new package?

Make sure you downloaded the latest version from visomaster.com. If the problem persists, join our community for support — links are available on the website.

---

Upgrading your GPU should be exciting, not frustrating. We're committed to keeping VisoMaster running smoothly on the latest hardware. If you've hit this error, grab the updated package and get back to creating in minutes.

Download the latest VisoMaster all-in-one package → visomaster.com