First of all I don't play games at all and I still quite new to deep learning.I was using Alex-net(transfer learning actually) in MATLAB to classify images in my current laptop(i5-3230,without any GPU). And it was taking roughly 25-30 hours to finish a training and I don't dare using gooLeNet. I have to buy a laptop. How fast will be if I use GPU? And I am not that rich to buy 2080 or 1080ti that easily. Actually I am poor and I have struggle financially if the price exceeds $1200.I have intention to do research on signal and medical image processing using deep learning in future. Will it be enough for me if I buy a laptop with GTX 1060 6GB or is it essential/overkill to buy one with RTX2060?
Feel free to give any other option that could fill my requirements...

  • 5
    $\begingroup$ Try out Google's Colab first. They allow you to use one of Google's GPU (or TPU) for a certain amount of time. This will give you an idea of the speed-up that you gain using a GPU, which should be significant. $\endgroup$ – nbro May 20 '19 at 22:31
  • $\begingroup$ But unfortunately I am using MATLAB and I am not very good python in case of DL. Can I use MATLAb code in Google's Colab? $\endgroup$ – Fazla Rabbi Mashrur May 20 '19 at 23:23
  • $\begingroup$ Mainstream GPUs like the RTX2060 aren't efficient enough for Deeplearning. Professional scientists are using the GreenArrays GA144 chipset (stackbased transputer) for training neural networks. $\endgroup$ – Manuel Rodriguez May 21 '19 at 8:21
  • $\begingroup$ Some neural networks will be too big to run on a GPU with 6Gb of memory. $\endgroup$ – Lahav May 26 '19 at 3:59

I think you should redirect your focus. Learn and play with ML, and only when compute becomes the main bottleneck for your learning, invest in hardware.

I've recently engaged into a ML project for which I 've assembled a machine with gtx1080, installed gui-less ubuntu, configured ssh, drivers etc, and than spend 3 months on data collection. And the dataset was so small (30k images), that I could probably learn mobilenet v3 on my CPU instead.

I think that in most cases focusing on hardware is a form of procrastination when it comes to machine learning. Also, a laptop with really strong GPU makes little sense (unless it's really well cooled like Alienware r3 that I use at work, but it's to heavy to be really considered a laptop...). It will throttle, It will be less durable due to constant overheating when you learn models.

So I'd say, go for gtx1060 laptop and only when you really need compute (you think you have a chance to actually win a competition due to a great idea you have, or you have a client that has seen a demo and really want's to pay for your model if you improve it), invest into standalone PC or cloud to tune your prototype model.

| improve this answer | |

If you are just starting out with Deep Learning, then a laptop with GTX 1060 is enough. I am using a GTX 1060 myself and I find it adequate for many of my personal projects, training large datasets and participating in most (not all) Kaggle competitions as well. But you say that you want to do research work as well. In that case, you may want to contact someone who has worked in that field. What kind of hardware are they using, the efficiency, training time, and all such questions. You should have a pretty decent idea after that.

In the meantime, keep searching for RTX laptops in \$1200 range. I think that a few laptops have popped up in the market that is giving RTX GPUs for \$1200.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.