GPU performance(time taken) same as CPU in GeForce GT 755 M

i have windows 10 ,
core i7 4th gen lenovo y510p,
8GB ram,
nvidia geforce gt 755 m,
i installed all the things it requires like cuda 8,
cudnn 6,
visual studio 2015,
g++ compiler,
anaconda,
python 3.5

and run the script given in tesorflow.org and i got this

(tensorflow-gpu) C:\Users\Admin>set “KERAS_BACKEND=tensorflow”

(tensorflow-gpu) C:\Users\Admin>python tensorflow_self_check.py
TensorFlow successfully installed.
The installed version of TensorFlow includes GPU support.

but when i run on CNN of dog-cat dataset its speed is same as CPU (BOTH ARE TAKING APPROX SAME TIME FOR EPOCH)

then whats the benifit of installing this cuda and all of this
or there any way to speed up this

the GT755M chip is quite a potato (lower mid range laptop GPU, also technologically outdated)

Instead, try running this on a desktop PC with a decently powerful GPU (my definition of “powerful” starts at GTX1070)

For professional use there are Tesla P4 and TeslaP100 products that are specifically designed to speed up deep learning inference (not training). Both are quite expensive.

Christian