Laptop for Visualizations and Comp Sci (Grad student)

[Omnitect]

So I'm trying to pick out a new laptop for school and have the fortunate opportunity of not really having a budget, due to my professor's almost surplus of funding. My research area is visualization of artificial intelligence, and I work in subfields of artificial intelligence that benefit from both cpu performance (through search) and gpu performance (machine learning). I can offload a lot of my work to our super computer, but I like to be able to train models and do search on my local machine. So, I'm looking for a high end laptop that can do at least some of this work, but still be thin(ner) and light(er), since I'm going to be carting it around everyday to and from my office.

 

I've looked around and come across a few results, with pros and cons on a few choices.

 

Macbook Pro 2018

 

Good:

Core i9

decent graphics card for what I need (can train on another dedicated machine)

I actually like Mac OS

best screen in the business

they finally have 32 GB RAM a good battery

 

Bad:

The thermal throttling on the VRMs (main concern and my only withholding on getting this device)

Expensive storage

 

Dell XPS 9570

 

Good:

Core i9

32 GB RAM

1050ti graphics (not entirely sure how it pairs up with the radeon pro 550X)

Cheaper storage

 

Bad:

Possibly thermal throttling? Not sure on it's state compared to the MBP 2018

Windows

 

Aero 15X

 

Good:

32GB RAM

GTX 1070 (!)

Cheaper storage

 

Bad:

Core i7 (only bad vs the i9's)

Style of the laptop

Windows

 

 

As far as OS is concerned, I would prefer Mac OS and don't really like windows, at least on a laptop. I'd probably use linux again and dual boot on either of the windows machines, but it's my second choice to the MBR. These are the three I've looked at the most, so if anyone has any better suggestions that I haven't seen yet, that would be amazing! Also, if you think I should wait for Apple to respond to the thermal throttling issue (haha), tell me! Thanks for any insight you lovely people might have.

[DrMacintosh]

If I were you I would only be looking at 8th Gen Core i7 laptops. The Dell will thermal throttle just like the MacBook Pro with the Core i9. 

 

I recommend the MacBook Pro. Get the Core i7, 32GB of RAM and the Radeon Pro 560X GPU upgrade.

[TLCH723]

Aero 15x is hard to get linux on.

 

XPS15 and MBP both have thermal issue.

 

If you really want a Mac then get a Mac.

 

btw, what ml framework are you using? tensorflow only support CUDA so only nvidia

 

[Omnitect]

Supports*? They don't do opencl?

[TLCH723]

2 minutes ago, Omnitect said:

Supports*? They don't do opencl?

Not offically last time I check.

 

And this is what I found https://github.com/tensorflow/tensorflow/issues/22 

[Omnitect]

Well, I do have a dedicated machine that I can run bigger projects on. Idk yet though. 

[TLCH723]

I would suggest a desktop at the office and a light-weight laptop with eGPU at home

[Omnitect]

Alright. I can probably do that. I might just do an X11 forwarding type setup then. 

[genexis_x]

3 hours ago, Omnitect said:

Core i7 (only bad vs the i9's)

What if I tell you that 8950HK is actually not much faster (sometimes even slower) than 8750H?

 

If you prefer macOS, the choice is obvious