The Nvidia 4*** graphics card is ignored

[SSar2026]

I'm using an Nvidia 4060 Ti graphics card for brute force on Win10, but no matter what I do, it only uses the CPU instead of the GPU.
I installed the NVIDIA GPU Computing Toolkit (CUDA) v13.2, but it didn't help.
Next, I installed visual studio community and recompiled the binary from the source code using a bat file, but it still didn't work!
In the end, I spent several hours trying to get the software to see the graphics card, but the process always ran only on the CPU, no matter what I did.
I'm using a source bin file (not a wav file).
What else do I need to do to make the software see the graphics card? Can you recompile the release with support for the NVIDIA GeForce RTX 4*** ?

[felixx-sp]

Oh! Let me check the problem.
Could you tell me what you get when you run the command "nvidia-smi" in your terminal and send me the result?

Thanks!

[SSar2026]

Oh! Let me check the problem. Could you tell me what you get when you run the command "nvidia-smi" in your terminal and send me the result?

Thanks!

[felixx-sp]

Hey! I've added a couple of fixes and improved error handling. Could you try again with the new version (v0.3.1)? If it still doesn't work, please let me know here along with a screenshot of the error you're getting. Thanks for your help!

[SSar2026]

Hey! I've added a couple of fixes and improved error handling. Could you try again with the new version (v0.3.1)? If it still doesn't work, please let me know here along with a screenshot of the error you're getting. Thanks for your help!

Congratulations! You've fixed it, but ... NO
The CPU usage is 60-70% (i5 13600KF), and the video card usage is 0-1% (4060 Ti). How did that happen?)

If you have a Telegram contact (or private messages), I can help you more quickly with tests and a few other automation-related questions.

[felixx-sp]

Hey! The Task Manager showing 0% is totally normal for CUDA workloads on consumer GPUs — it only tracks the 3D engine, not Compute. Use nvidia-smi to check real GPU usage, that's the one that matters. If nvidia-smi shows activity you're good! Feel free to reach me on Telegram @s6 if you need help.

[felixx-sp]

Also, I've made some changes in v0.3.2 that should fix the GPU being ignored on RTX 4xxx cards. Give it a try and let me know what keys/s you're getting!

[SSar2026]

Also, I've made some changes in v0.3.2 that should fix the GPU being ignored on RTX 4xxx cards. Give it a try and let me know what keys/s you're getting!

Now the use of the video card has turned on, it is immediately noticeable because yesterday it was 0% and now it is 100% and it is difficult to confuse it, but there is a problem the speed of 40M is extremely low even for my CPU, not to mention GPU.

PS Does your utility remember the checkpoints when iterating? If you close it and then run it again the next day, for example.

[felixx-sp]

I've added your suggestion to the issues list. I've also improved the CUDA code in v0.3.3 to make better (and more efficient) use of GPU memory. On my laptop's 3050ti, I've gone from 11M keys to 97M

[felixx-sp]

By the way, my Telegram username is S66S6S6S