Modular is an artificial intelligence company that a few weeks ago presented a programming language called Mojo, of which a lot has started to be said about it because it has two very striking characteristics: combines the ease of use of Python with the speed of C.
There are numerous projects underway to make Python faster, such as Jax and, more recently, a python compiler called codon. And that’s not to mention alternative languages oriented to data science like Julia.
The programmer career in 2017 and in the future (with Javier Santana)
35,000 times faster than Python
Mojo manages to distinguish itself from other Python improvement efforts because according to its creators is “35,000 times faster than Python” when running numerical algorithms like Mandelbrot thanks to hardware acceleration.
In the words of its creators, “Mojo combines the parts of Python that researchers love so much with system programming capabilities that require the use of C, C++, and CUDA.”
mojo allows the use of custom CPUs, GPUs, TPUs and ASICs, according to the strengths of each type of hardware, according to its creators. “Mojo’s high-level syntax and semantics are comparable to those of Python, making it easier for experienced Python developers to learn and use.”
To be more agile, Mojo is compiled statically, which is translates into faster execution times and better optimizationsince the code is compiled before execution.
Data scientist Jeremy Howard, co-founder of Fast.ai, has gone so far as to say that “Mojo may be the biggest programming language advance in decades.”
According to him, this language tries to address the bifurcated reality of AI: while AI models are developed in Python due to the richness of the ecosystem, Python programmers often end up plugging their code into modules in more performing languages, such as C/ C++ and Rust. AND this “two language” approach makes it more difficult to profile, debug, learn and deploy machine learning applications.
Who’s after Mojo
chris lattner is one of its forerunners. He previously worked at companies like Apple, Google and Tesla and co-developed the LLVM compiler toolchainco-founded the MLIR compiler framework and spearheaded the development of the Swift programming language.
Another of its precursors is Tim Davis, co-founder of Modular and former head of Google ML, where he oversaw machine learning APIsthe compilers and the execution infrastructure of the Internet giant.
Image | Modular