Machine learning and notebook

OK, this might seem like an oxymoron, but if you are doing machine learning you kind of want to do it when mobile or in poor connectivity areas yet also want something that doesn’t weight 7 pounds. I tried the MSI gaming notebook, but at 17 inches and 6 pounds it is just too big. So what if you spend most of your time computing but still want something for the road the main issue is lack of nVidia graphics. The answers are surprisingly large thanks to an overview by ultranotebookreview.com and the latest CES announcements, but the criteria are:

  • Ultrathin (0.8 inches)
  • Ultralight (4.5 pounds in this case like the macbook pro)
  • 2K display (at least, 1080p is just too litle)
  • nVidia graphics with 4GB for all those big models
  • 16GB memory
  • Skylake processor
  • Durable

Here are some choices:

  • Dual boot an older MacBook Pro 2014. Now the MacBook Pro 15 is about the right size and it weights in at 4.4 pounds and the older models (not the new MacBook Pro 2015) has an nVidia GT750M on it, so you can dual boot your MacBook with Ubuntu. It’s scary as hell, but apparently Ubuntu Grub can dual boot. This isn’t Skylake, but it has everything else
  • ASUS Zenbook GX500 or 
  • ASUS ROG G501VW. Just updated to Skylake, but I haven’t read a review. 4.4 pounds, 0.8″ thick, 
  • Gigabyte Aorus X3 Plus. This was just updated to v5 with Skylake, HDMI 2.0, USB C/3.1 and 4x PCIe M.2 SSDs. It is 4.1 pounds, the same 3200×1800 display that Razor has 16GB ram and 2x m.2 slots and 970M with 6GB VRAM. There is also a cheaper X3 model with a 2500 panel. Like many of these things, it has a huge power brick and also the hinge looked flimsy, so beware.

These lack a key feature.

  • Razor Blade. This is a ultrathin notebook with a nVidia graphics GTX970M with a 14″ 3200×1800 display with i7 quad core and 16GB is just about a perfect dedicated machine. It is expensive though and weighs 4.5 pounds as well. It hasn’t been updated to Skylake though, so although it has great build quality, the Aorus has better specs.
  • Gigabyte G34. This is the budget version of the Aorus X3 Plus and was also just updated to the v5 variant with Skylake, USB C/3.1, 2GB VRAM GTX965M, 4x M.2. It has a 2.5″ drive slot as well. The biggest reason it is cheaper is that it has a FHD (1080p) display and less VRAM and weighs 3.75 pounds and is much cheaper.
  • ASUS Zenbook UX303UB. It is a pretty small display at 13″ (I find 15″ about the perfect compromise for developers). This is the Skylake update with GTX940M so not truly fast, 12GB of memory and the SSD is in a 2.5″ bay so easy to replace. Overall a nice entry system that is good for typing and then if you are on the road and want to test your GPU aware machine learning application you still can. It is more MacBook Pro 13 class and weighs just 3.3 pounds. Durability seems to be the main issue with this one as well as the 12GB vs 16GB.