|
|
|
|
|
|
by nine_k
2140 days ago
|
|
|
Hey, Pandas wrap heavily optimized C and Fortran code, with a thin layer of Python interface on top. This thing is pure Julia. This being any faster than Pandas is a huge compliment to Julia the language in general and its JIT compiler in particular. |
|
|
Indeed. This 'whole stack under the same language' is an important feature to have. You get to reap the advantages of an improved JIT. End to end autodiff is easier. Some of these things are a problem, for example, in PyPy because of the Python <-> C bridge.