Geforce RTX 4070 Ti Super support for Tensorflow-gpu

General Discussion
,
#1

Hi,
Could you please let me know whether “Geforce RTX 4070ti Super” support Tensorflow-gpu.
In below link https://developer.nvidia.com/cuda-gpus
it is specified that Geforce RTX 4070 ti is capable for CUDA tensorflow-gpu. But it didn’t mentioned anything about “4070ti Super”. “RTX 4070 Ti” and “RTX 4070 Ti Super” have different numbers of CUDA cores and VRAM, so I’m concerned about potential issues when configuring environments for TensorFlow-GPU in the future. Please confirm whether Geforce RTX 4070ti Super support tensorflow-gpu.

#3

Hi @hoya9802, As RTX 4070ti super is cuda enabled you can use this GPU for tensorflow. Thank You.

1 Like
#5

Thank you so much! Then, should I download it to match the “Compute Capablilty 8.9 of the RTX 4070ti” that is on the link even though I buy “RTX 4070ti super”?

#6

Hi @hoya9802, If you install tensorflow using pip install tensorflow[and-cuda] the cuda(8.9) and cudnn(12.2) versions supported for tensorflow 2.15 will automatically download and you can start using GPU with tensorflow. Thank You.

1 Like
#7

thank you for your response!!

#9

The GeForce RTX 4070ti Super should support TensorFlow-gpu, as it falls under the RTX 4070 series specified for child recovery modified by moderator CUDA compatibility. While specific mentions may not exist for the “Super” variant, the architecture similarities should enable TensorFlow functionality.