Pierre Augier
Pierre Augier
There is nothing on cross-compilation in Pythran documentation. I understand that it is a subject in itself but sometimes, adding `-march=native` really speedup computations, but then the binary is no...
Hi Serge, i hope you are fine at home. I investigated about the bad result for Pythran mentioned in https://github.com/Thierry-Dumont/BenchmarksPythonJuliaAndCo/wiki/5-The-FeStiff-benchmark (see the resulting PR https://github.com/Thierry-Dumont/BenchmarksPythonJuliaAndCo/pull/12). I isolated two issues. The...
nbabel Pythran code is 10 % slower with Numpy 1.20.1 than with Numpy 1.19 ! 3 seconds is a bit too much for just conversion between Python-Numpy objects to Pythran...
I get a new PR in the project https://github.com/paugier/nbabel for a more efficient Julia implementation. It's now 30% faster to what we get with Pythran. 30% is of course not...
Basic support of `numpy.linalg.solve` (https://numpy.org/doc/stable/reference/generated/numpy.linalg.solve.html) would be very useful. I find it quite often in codes not written by me that would benefit from acceleration with Pythran. @serge-sans-paille do you...
I'm surprised that xsimd makes a function slower + the implementation with explicit loops is faster: ``` In [1]: run microbench_simd.py In [2]: %timeit advance_positions(positions, velocities, accelerations, time_step) 4.16 µs...
Did you read this paper published in Nature: [The Ecological Impact of High-performance Computing in Astrophysics](https://arxiv.org/pdf/2009.11295.pdf) ? One of the few figures show that Python-Numpy is very inefficient for a...
https://docs.python.org/3.8/library/collections.html#collections.Counter can be very useful in numerical kernels. Since they are very similar to dict, it may be not too difficult to implement, especially a simplified version without the extra...
Sorry to fill up Pythran issue tracker. As you can see, I'm working with Pythran these days. Same thing than this old closed issue https://github.com/serge-sans-paille/pythran/issues/413 > It looks like setdefault...
The [support page](https://pythran.readthedocs.io/en/latest/SUPPORT.html) could be improved. Each time I have to read it, I find it unclear. Too much "guess what it means". It's not at all important and I...