I get to choose a new laptop for work (data processing, data visualization, deep learning). I'd like to switch to Linux
(probably PureOS) as I think it can do what a Mac does, and it's better at some things. What are people using?
I bought the Dell XPS 13 (9650) about 8 months ago. It comes preloaded with Ubuntu so you don't have spend too much time debugging drivers. This was my main concern since I was going to be running Linux as my only OS. I've only had a few problems with wifi drivers and ethernet over USB C so far. Costco occasionally has the high end model (preloaded with windows of course) on sale for $400 less than the dell site which is a great deal.<p>Resources:<p>1. <a href="http://www.dell.com/ca/p/xps-13-9350-laptop/pd" rel="nofollow">http://www.dell.com/ca/p/xps-13-9350-laptop/pd</a><p>2. <a href="https://wiki.archlinux.org/index.php/Dell_XPS_13_(9350)" rel="nofollow">https://wiki.archlinux.org/index.php/Dell_XPS_13_(9350)</a>
Lenovo Thinkpads are top tier. Shoot for X and T series for best build quality.<p>You want high resolutions, so customize it to the highest you can get. Some people may say it makes stuff look too small, but software will continually will improve to handle HDPI scaling, but that's easier than trying to switch out a screen yourself. Remember you'll at least have an editor and docs on the screen at the same time.<p>You said you want to do deep learning / big data stuff, Tensorflow can work with CUDA. Feel free to go for the NVIDIA laptops (T and P series), which can fall back on intel integrated graphics for power savings. NVIDIA's Linux and FreeBSD drivers are nice.<p>These laptops work well with Linux and FreeBSD.<p>On FreeBSD, newer intel chipsets graphics and wifi may have issues. If there's an issue with wifi support, consider a USB wifi like "ASUS (USB-N10) wireless-N USB Adapter" or "TP-LINK TL-WN725N Wireless N Nano USB Adapter", which have chipsets that are supported on FreeBSD and Linux very well.<p>Consider outlet.lenovo.com for deals on these laptops. Older generation laptops work well<p>Steer clear of anything with *40 at the end. T440, X240, etc. are the scourge of the Thinkpad line. They removed the physical keys from the buttons in the mouse and the trackpad itself is horribly poor quality. <a href="http://www.laptopmag.com/images/uploads/4020/g/lenovo-thinkpad-x240-g15.jpg" rel="nofollow">http://www.laptopmag.com/images/uploads/4020/g/lenovo-thinkp...</a> is bad. <a href="http://www.laptopreleasedate.com/wp-content/uploads/2016/01/Lenovo-ThinkPad-X260-Intel-Skylake-3.jpg" rel="nofollow">http://www.laptopreleasedate.com/wp-content/uploads/2016/01/...</a> is good.<p>You can go for X series and grab an external monitor / base station to plug into. This gives you may portability. But I still advise you go for 1080p resolution, if you compromise for less, it may be too squished to program in.
I am using Thinkpad T series. Waiting for Thinkpad P51 [0]<p>[0] - <a href="http://shop.lenovo.com/us/en/laptops/thinkpad/p-series/p51/" rel="nofollow">http://shop.lenovo.com/us/en/laptops/thinkpad/p-series/p51/</a>
I'm a firm believer that all laptops below the 2000 range are terrible so just pick one that has enough RAM and solid build quality. Probably Dell XPS or a Thinkpad
Laptops embody engineering tradeoffs.<p>If |deep learning| implies a discreet GPU then battery life will suffer. It will particularly suffer under Linux because the GPU and CPU manufacturers don't have not created Linux drivers that allow dynamic switching between the GPU and CPU graphics core when running the display. The CPU graphics cores generally use less power than the GPU and the switching provides longer durations running on battery power.<p>A similar tradeoff comes in terms of processing power versus weight and bulk. Fast multicore processors need to dissipate more heat. Heat sinks add weight. Fans require passages to move air and add bulk.<p>If portability and light weight is the killer feature then there are one set of alternatives. If 64GB of ECC; two teraflop of GPU; and a quad core Xeon are killer features then there are another set of alternatives.
From what I've read you want a good Nvidia GPU for machine learning so you can run cuda. Now you can rent GPU instances from AWS but it's probably cheaper to buy your own system if you are doing it often enough. As others said you may want to consider a more powerful desktop machine with 2x Nvidia GPU and just network that up to your laptop to offload workloads.<p>Personally I'm going to try AWS route first to see how much I enjoy the machine learning route. Then I can make a better decision without dropping a few thousand first.<p>I'm running MacBook pro 15 with touch bar. I prefer the older models that still have real USB ports. I also really dislike the new keyboard and giant mousepad. Never use the touch bar but that's just me. One last note... On new macs the hardware is impossible to modify. Hard drive and ram are soldered into the boards...
For me its also a MacBook Pro 15". It's beautiful, durable hardware. You can use it for many years and sell it afterwards for a good price, if you want to upgrade. The OS combines the best of Linux and Windows, but looks better. The terminal practically takes the same commands as in linux, very powerful. The OS is well thought through as opposed to the mess you see with Windows sometimes.
I currently use a Dell E7470 running Linux (elementaryOS specifically), and a Razer Blade running Windows 10.<p>Really like both. The Dell has a great range of ports built in (including ethernet), so no carrying around a bunch of dongles.<p>The Razer has a faster CPU (quad core vs dual core) and the best keyboard (laptop or not) I have ever used.
Dell XPS 13, 9360. Can ship with Linux preinstalled, and you can of course install your distribution of choice. I run Debian unstable on it and use it as my main development machine. Works with the Display-Link USB 3.0 dell docking station (haven't tried the Type C Usb yet). Even supports dual-head. But you have to install the drivers from the display link homepage. It is not a beast when you need performance. At the office my boss provides me a Dell XPS 15, with i7, 16gig and nvidia graphics card, works with the docking station too.
I recently purchased a 2016 Macbook. Reasoning is, they last longer.<p>My sony vaio stopped turning on after 2 years and my dell xps didn't even last 1.5 years. My 2009 Macbook still runs today, but it's no longer powerful enough.
I'd go with a Dell Precision M6800 or M4800, depending on your desired size. It can run linux natively, has a full sized keyboard, and all components are replaceable (even the CPU is a pretty doable replacement).
My ideal laptop I haven't found it yet.<p>Major latest GNU/Linux distributions works out of box(I usually had to configure something manually to make it work)<p>Decent CPU, Memory, Storage(This is the easiest requirements to satisfy because my work is mostly relies on CLI tools and web browsers)<p>No power hungry GPUs(especially nVidia's)<p>15 inches builtin 4k display(this is hard. I can find 12 inches with no dedicated GPU or 15 with dedicated GPU laptop)<p>DisplayPort(Most laptops only have HDMI port)
Ethernet(I don't want to carry USB-Ethernet adaptor)
4 USB ports(I don't want to carry USB hub.)
If you ever want to develop iOS apps, you pretty much have to get an Apple laptop unless you're willing to spend ages messing with hackintosh and trailing behind Xcode updates.<p>Not sure why all the hate for MacBooks. You can triple boot OS X, Linux and windows if you want. And since so many people use macs, there's plenty of community support for getting Linux to work.
what about this one: <a href="http://www.dell.com/en-us/shop/productdetails/xps-15-9560-laptop/dncwxb1609s" rel="nofollow">http://www.dell.com/en-us/shop/productdetails/xps-15-9560-la...</a><p>The new XPS 15 with 32GB RAM. I'm just worried about battery time since more RAM mean less battery time.<p>Any one have this? Impressions?
A laptop is probably not the best idea for deep learning, but you can certainly use one as a terminal to connect to a much more powerful gpu-optimized server. I'm curious to hear what others are using for their deep learning setup.
Thinkpad T-series which I bought 5 years ago for $2K.<p>I am amazed how it still holds under a daily heavy use at home and in the office.<p>Hope the new ones are as reliable.
that would be the new MacBook Pro
<a href="http://www.apple.com/macbook-pro/" rel="nofollow">http://www.apple.com/macbook-pro/</a>
released 20 minutes ago
<a href="http://www.apple.com/mac/" rel="nofollow">http://www.apple.com/mac/</a><p>Although "for work" I'd recommend getting a desktop computer, especially if you wanted to do deep learning.