Opting for C over Python for Neural Network Development

Python is a comfortable lie. Don't get me wrong, I love Python. It’s easy and works great, especially for fast prototyping. But when I was building a neural network sometime ago, I decided to do it the hard way: removed all the abstractions I possibly could! No libraries. No shortcuts. People often talk about how heavy software can get. Python is great, but it’s like carrying the whole house wherever you go for a "just in case" situation. It makes you super slow and you don't even need all that stuff. C doesn't. It in fact, leaves you with the basics and lets you figure out life your way. I had to manage every single byte of memory myself. It was frustrating, and I spent hours chasing bugs. But the result? The C version was 100x faster. It goes back to what I said in my previous post: simple solutions only stay simple locally. Python is easy to write, but C is what lets the hardware fly. #ComputerScience #Coding #CProgramming #LearningInPublic #SoftwareEngineering

Asides from strengthening your fundamentals, what purpose would this achieve other than slowing everything down even more?

Python gives speed to developers. C gives speed to machines. Both have their place.

See more comments

To view or add a comment, sign in

Explore content categories