Hello, I encountered an issue with GPU architecture incompatibility while trying to reproduce your code using a 5070ti GPU. I installed CUDA12.8 version compatible with 50 series GPU and Nightly version corresponding to PyTorch, but encountered an error during compilation due to incompatible architecture.
Error message:
NVIDIA GeForce RTX 5070 Ti Laptop GPU with CUDA capability sm_120 is not compatible...
The current PyTorch install supports CUDA capabilities sm_50 sm_60 sm_70 sm_75 sm_80 sm_86 sm_90.
error: no suitable conversion function from "const at::DeprecatedTypeProperties" to "c10::ScalarType" exists
ps：Only partial information was intercepted
Is there a way to avoid these problems, or can we just wait for the optimization of the 50 series GPU by various software packages in the future?
Looking forward to your reply
Hello, I encountered an issue with GPU architecture incompatibility while trying to reproduce your code using a 5070ti GPU. I installed CUDA12.8 version compatible with 50 series GPU and Nightly version corresponding to PyTorch, but encountered an error during compilation due to incompatible architecture.
Error message:
NVIDIA GeForce RTX 5070 Ti Laptop GPU with CUDA capability sm_120 is not compatible...
The current PyTorch install supports CUDA capabilities sm_50 sm_60 sm_70 sm_75 sm_80 sm_86 sm_90.
error: no suitable conversion function from "const at::DeprecatedTypeProperties" to "c10::ScalarType" exists
ps：Only partial information was intercepted
Is there a way to avoid these problems, or can we just wait for the optimization of the 50 series GPU by various software packages in the future?
Looking forward to your reply