Advice for Purchasing laptop

If you really have some money for a laptop, go and check the MSI Titan series. It comes with 2 NVIDIA GPUs 1080 TI.

But it’s not for everyone!

Link: https://ca.msi.com/Laptop/GT83VR-7RF-200CA-Titan-SLI.html

Steeve

P.S.: I have one, it comes with a backpack!

1 Like

I’m planning to buy the new 15" Surface Book when it’s out next month. 6GB GPU in a slim convertible form factor - but at a price!

Maybe the MSI Titan should come with a forklift?.. :stuck_out_tongue:

4 Likes

Amazon has GTX1070 model starting at $1699 with 512GB SATA SSD.

Seems that Eluktronics, Sager, XOTICPC, are resellers of Pro-Star Clevo brand?

This configuration has an upgrade to a 4K screen and an overclockable 7820HK CPU. A 7820HK is 3% faster than a 7700HQ at standard speeds.

You can custom configure a similar system at https://www.pro-star.com/p650hs-g.html

Apparently i’m also planning to buy a laptop ,to end my misery over working in deep learning with a core i3 processor ,4gb ram laptop.so i request you if you find the right laptop for deep learning which is available in india and under 1000$ please refer it to me .

1 Like

Yes, I’ve found 3 that I really liked, I’m still searching for a better deal if I can get hold of one.

I’ve found 3 good ones in the price(prices in INR for your convenience) bracket of:

  1. 88k: MSI GL62
    It has a 4 gig vRAM. @Jeremy suggests to use one with 6gig atleast.

  2. 1lakh: Acer 300 Helios:
    However this is an imported version, not sure if Amazon will provide warranty on it.

  3. 1.25 lakhs: Dell Inspiron 7757:
    I’ll be getting this one most probably, because I want one that’d be able to get me through 3-5 years at least there’s one hiccup in all of its reviews that the battery, weight, screen, keyboard all suck. But I’m fine with that as long as the Hardware is promising enough.

Edit: However, I must mention that everyone stresses that a custom assembled rig would be way better than a laptop in terms of price and performance. I’m being given a grant from my college hence I’m going to put in some money from my side and get it. If it’s possible, you should stick with the option of assembling a rig.

Regards.

1 Like

thank you very much for a detailed explanation,and for a kind advice i really appreciate it.

I have Hp omen (Gaming laptop) with GTX 1050 4GB and 8GB of ram . ( currently with HDD). Will this laptop be useful instead of AWS or Paperspace?

Imo it wouldn’t be too bad. However I believe a greater amount of vRAM is suggested.
But we have the AWS treats to get us through the course so no worries until then :smiley:

I bought this laptop in April so I could take this course. Costs $1049, has a GTX1060 6GB, and 16GB RAM. https://www.amazon.com/Acer-Predator-Helios-GeForce-G3-571-77QK/dp/B06Y4GZS9C/ref=sr_1_1?ie=UTF8&qid=1499524653&sr=8-1&keywords=acer+laptop+gtx+1060

Just another thought: have a think about whether you really want a laptop, rather than buying a DL desktop and a cheap laptop to connect to it. You can often get more processing power for less money with a desktop, and can add more GPUs to it later.

9 Likes

Yup. Jeremy is right.

I’m looking to build a custom 12 NVIDIA 1080 TI. You will find really fast the limit of a laptop in the deep learning realm ;).

I already have :slight_smile: while attempting to run the dogs vs. cats ensemble model from Part 1v1. My goal for buying this was to quickly learn while saving up to build a headless multi-GPU desktop rig.

1 Like

Portability is the main benefit of a laptop over a desktop. After much researching, I’m inclined to prefer a DL desktop because of performance and value.

I’ve considered many DL desktop options such as Ryzen, Threadripper, Coffee Lake. I’m now focusing on a Intel 8700K system. Additionally, I’m backordered on a Intel Optane SSD 900P (480GB, AIC PCIe 4.0, 20nm, 3D XPoint) which labels NVMe as so 2016.

Tom’s Hardware review of Intel Optane 900P 3D XPoint: http://www.tomshardware.com/reviews/intel-optane-ssd-900p-3d-xpoint,5292-3.html

Finally! Christmas is near and I will get my laptop soon! :smiley: (My college administration is too slow, I know)

  1. I have left my final decision to be one from Gtx-1070 one vs Gtx-1060 one. The price difference is about 800USD. Is it a good idea to spend that much?

(The pricing is according to India, Can’t negotiate with that :frowning:)

  1. The RAM is 16GB on the laptop. Upgrading to 32 GB will cost me about 300 USD, should I do that right away or wait a little while and upgrade it later when the price for RAM drops? (I ask this because, I will definitely take up the Part 2 next)

I’m convinced (almost) that the hot setup is to use a cheap notebook to remote into a machine learning desktop GPU server. Put your money into the desktop GPU server. Desktops are cheaper and more powerful than notebook machine learning systems.

1 Like

I actually have the same laptop. I’m looking for instructions/tutorials to setup a working environment on this without having to resort to an option like Paperspace/AWS. I’d appreciate being directed to such a thing.

1 Like

You can use the same script as the one for setting up paperspace.
But do check if you’d want to comment out a few commands in it.

In my opinion, you should buy a desktop workstation, and a cheap laptop (but be sure not to neglect IPS display and a matte coating for it) to remotely log in into your ws.

You will have both at the same price of a fancy GPUed laptop, but you will end up with a much more flexible solution with no heat-related issues, and you laptop will be lighter to carry around. And you won’t lose an eye if you throw a cup of coffee on it.

Minimum desktop spec: gtx 1070, dual core, 16gb ram, but if you can, buy something able to handle more pcie lanes for future expansions (e.g. xeon e5, amd threadripper or epyc) and ECC ram.

Minimum laptop spec: like I said, IPS and matte coating for the display, and a good battery.

Note: ECC ram

@jeremy: what do you think about it? ECC ram could prevent data corruption in the long run. Do you have it onto you fast.ai or enlitic production workstations?

1 Like

Consider a Titan V instead of 12 gtx 1080ti. You will have roughly the same processing power (~110 tflops) for less than half the price, without parallelization issues.