Is macbook air m2 good for A.I ML or should I go with HP Victus(16gb ram,rtx 3050A,i7-12th gen)
I am getting the 16gb ram 8 core CPU and 8 GPU on air m2
Need help guys? Which one will run TensorFlow,PyTorch etc. better?
HP Victus,you can use CUDA of 3050,but can you switch to 4060?8G ram of GPU better than 4G too much,not expensive too much,3050 is real dogshit for both game and ML.and 16g ram is not enough too,now ram is cheapper than before,so choose 32G,or you can choose R7-7840 or 8845 CPU,AMD better than intel on laptop.
I used a gaming laptop during my engineering days (i7 8th gen, 1060 6GB, 16GB RAM, 1TB HDD + 256GB SSD) and now I use M2 Macbook pro as an AI/Software engineer.
Here are the facts -
M2 Pro or higher models of M2 Air are much superior devices and set the bar really high for laptop devices, they generally easily last 20+ hours in terms of battery, extremely portable and good build quality, if you throw it at someone, the guy might crack their skull but your laptop would be just fine, have impressive performance even on battery, however their GPU is pretty much non-existent when it comes to performance. The display and speakers are nice too and if by chance you have an iPhone/Airpods, the experience is even better.
These gaming laptops have really good CPU and GPU performance however it really tanks as soon as they are not plugged into power. You would be lucky to get more than 2 hours of battery life from them after a year, they are bulky and my gaming laptop's charger was heavier than a macbook air. The GPU is good to have and back in the day (2019), 6GB of VRAM used to be a lot. The display is usually very bad (exceptions are Acer Predator back in the day) and the speakers are also horrible.
You can barely do any shit on 6GB VRAM nowadays. Even common models from hugging face use 8GB + VRAM atleast.
However one thing that these gaming laptops offers is customisability and experimentation.
You could install linux, windows or just whatever you want on them. Really important for engineering as most of the software supports either windows or linux. MacOS will be a real pain in the ass once you realize you are stuck in the few software that MacOS allows you to use. Also Macs have much smaller disk space compared to these gaming laptops which is another pain.
So if you are a student I recommend that you always experiment, install 5 different IDEs and try them all out. Install compilers for 10 different languages and try them all out. Install 2-3 linux distros and experiment with them. Break shit, fix shit. Only way to learn and become a good engineer. There will be a lot of times your linux installation will break (graphics drivers gone, wifi not working, audio not working). Learning how to fix these issues is crucial to problem solving later down your engineering path.
MacBooks however do not let you do anything like this. You have limited disk space which is just enough to install that 1 good IDE you already know how to use and love and 1 tech stack. You can expect your macbook to just work when you open the lid. As such you loose out a lot as you never learned to troubleshoot stuff.
Macbooks are meant for people who know their shit and have made up their minds on what tools they wanna use vs Gaming laptops which allow people to experiment with whatever they want to do.
Get a gaming laptop if you are a student, get a macbook when you are a professional and stay away from Macbook air base models unless you are just someone who does light work on google sheets and watches movies.
Coming to AI and ML, honestly both M1/M2 air and a 3050 is absolutely garbage. The base model Macbook air with 8GB of RAM will start shitting it's pants even with IDEs like Android Studio or the other Jetbrains stuff and 4GB of VRAM in 2024 is honestly not even a matter of discussion as even back in 2019 a 1060 with 6GB VRAM was considered decent.
Neither of these laptops are good for AI/ML. Your best bet is Google Colab but even that is not a great experience as our sessions will be shut down after a few hours and GPU hours per week is limited. So honestly at your budget you just can't get a decent device to run AI/ML locally unless you invest in a PC and grad a good deal on a second hand GPU from the 3000 series. Your best option is to try and juggle between kaggle notebooks and google colab free tier if your budget cannot be increased.