Laptop for Machine Learning - 2023

[pascal]

Upgrade time!
The new laptop Asus Tuf F15 arrived.
Specs: i7 - 12650H, 16GB ram, SSD 512GB nvme, RTX 3070 8GB, 144Hz FullHD 15.6" display.

I installed Windows 11 / Office 2021 Pro ESD.
My first impressions:

- Cheaper than the competition.
- Nice look and very discret for a gamer laptop. I liked the black piano style.
- Robust and good material.
- Light wheight for a gaming laptop, only 2kg. The PSU is light too.
- Extremelly fast for home office use.
- It is cool and quiet for most of the time.
- Installed a new Kingston 960GB nvme (DC1000B) server grade SSD.
- Anaconda, CUDA and Pytorch working great, and it is 50% speedier than the previous laptop when running a Transformer model.

Very good!

 

[pascal]

More info
I use the laptop in silent mode, since the ML performance is not affected too much (less than 5%).
This keeps temperature (CPU and GPU) around 72c and 74c while training a Transformer model.
The CPU and GPU usage around 11% and 98%.

 

[pascal]

Three months now using this laptop.
One thing standout is how it being 1.1 pound (0,5 Kg) lighter than the previous gamer laptop makes me so happy!
Much less shoulder strain and easier to manipulate.

Here is a review. Not exactly the same model but is close:
https://www.tomshardware.com/reviews/asus-tuf-gaming-f15-2022

 

[pascal]

I tested the laptop with Chat RTX and it is working great.
https://www.nvidia.com/en-us/ai-on-rtx/chatrtx/

This laptop has the minimal configuration to run it.
The transformer model installed is Mistral 7B 4 int.

Wish I had a laptop with RTX 4080 to run the model LLaMa 2 13B 4 int.