CODEWITHSUNDEEP
tensorflowgpunvidia
Link L
Link L

Reputation: 439

Tensorflow: can it run ResNet-101 on a GTX 1070 (8GB)?

With tensorflow, is a GTX 1070 (8GB RAM) enough to run ResNet-101 ?

As per the ResNet paper, ResNet is "simpler" than VGG-16. Since VGG-16 already runs well in my current 4-GB GPU (GTX 960), a bigger 8GB 1070 GPU should have no problem with ResNet right?

Upvotes: 0

Views: 6318

Answers (1)

Guillem Cucurull
Guillem Cucurull

Reputation: 1691

According to this benchmark:

We benchmark all models with a minibatch size of 16 and an image size of 224 x 224; this allows direct comparisons between models, and allows all but the ResNet-200 model to run on the GTX 1080, which has only 8GB of memory.

All ResNet models but the 200 layers deep one fit in the 8GB 1080 GTX. As the 1070 GTX has also 8GB of memory, the ResNets up to 152 layers will fit into the GPU's memory, but not the deeper ones.

Upvotes: 4

Related Questions