• Russel Winder via Digitalmars-d (21/21) Aug 03 2014 The numba package (and llvmpy below it) is rapidly getting to production
• deadalnix (4/24) Aug 03 2014 Good data layout and avoiding to rely on AA is necessary to be

Russel Winder via Digitalmars-d <digitalmars-d puremagic.com> writes:

The numba package (and llvmpy below it) is rapidly getting to production
use stage, this means all those people using Python for data analysis
(and there are a lot of them) will no longer be searching for Cython/C/C
++/D to speed up their codes, they'll just  autojit their Python code to
generate LLVM based native code for the performance critical sections.

This will mean this arena of programming will only use C/C++/Fortran for
ready made libraries and not for anything new. I suspect this will make
PyD and similar more or less redundant.

On the up side, it is further emphasizing that LLVM is the short- and
medium-term future of native code generation and it is good that there
is LDC, and that it is (almost) up to date with D versions.
--=20
Russel.
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D
Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder ekiga.n=
et
41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel winder.org.uk
London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder

"deadalnix" <deadalnix gmail.com> writes:

On Sunday, 3 August 2014 at 10:01:42 UTC, Russel Winder via 
Digitalmars-d wrote:
 The numba package (and llvmpy below it) is rapidly getting to 
 production
 use stage, this means all those people using Python for data 
 analysis
 (and there are a lot of them) will no longer be searching for 
 Cython/C/C
 ++/D to speed up their codes, they'll just  autojit their 
 Python code to
 generate LLVM based native code for the performance critical 
 sections.

 This will mean this arena of programming will only use 
 C/C++/Fortran for
 ready made libraries and not for anything new. I suspect this 
 will make
 PyD and similar more or less redundant.

 On the up side, it is further emphasizing that LLVM is the 
 short- and
 medium-term future of native code generation and it is good 
 that there
 is LDC, and that it is (almost) up to date with D versions.

Good data layout and avoiding to rely on AA is necessary to be 
really fast. Pyhon won't perform as fast as C/C++/D.