davrot/pytutorial
1
0
Fork 0
mirror of https://github.com/davrot/pytutorial.git synced 2025-04-18 21:26:41 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Update README.md

Browse source 
Signed-off-by: David Rotermund <54365609+davrot@users.noreply.github.com>
This commit is contained in:
David Rotermund 2023-12-28 18:01:43 +01:00 • committed by GitHub
parent 648b7a85d3
commit c1c867a98c
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 46 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

63
numpy/numba/README.md
View file

@ -20,25 +20,54 @@ pip install numba
## [Numba basic types](https://numba.pydata.org/numba-doc/dev/reference/types.html)
For the example that will show you the options of optimization we need to understand the numba naming schema.
### Numbers
For the function signatures we need to be able to translate the usual np.dtype into numpy.types.
For doing so we just replace **np.** by **numba.types.** .
|Type name(s)| Shorthand| Comments|
|---|---|---|
|boolean |b1| represented as a byte|
|uint8, byte |u1| 8-bit unsigned byte|
|uint16 |u2| 16-bit unsigned integer|
|uint32 |u4| 32-bit unsigned integer|
|uint64 |u8| 64-bit unsigned integer|
|int8, char |i1| 8-bit signed byte|
|int16 |i2| 16-bit signed integer|
|int32 |i4| 32-bit signed integer|
|int64 |i8| 64-bit signed integer|
|intc |–| C int-sized integer|
|uintc |–| C int-sized unsigned integer|
|intp |–| pointer-sized integer|
|uintp |–| pointer-sized unsigned integer|
|float32| f4| single-precision floating-point number|
|float64, double| f8| double-precision floating-point number|
|complex64| c8| single-precision complex number|
|complex128| c16| double-precision complex number|
|numba.types.boolean |b1| represented as a byte|
|numba.types.uint8, byte |u1| 8-bit unsigned byte|
|numba.types.uint16 |u2| 16-bit unsigned integer|
|numba.types.uint32 |u4| 32-bit unsigned integer|
|numba.types.uint64 |u8| 64-bit unsigned integer|
|numba.types.int8, char |i1| 8-bit signed byte|
|numba.types.int16 |i2| 16-bit signed integer|
|numba.types.int32 |i4| 32-bit signed integer|
|numba.types.int64 |i8| 64-bit signed integer|
|numba.types.intc |–| C int-sized integer|
|numba.types.uintc |–| C int-sized unsigned integer|
|numba.types.intp |–| pointer-sized integer|
|numba.types.uintp |–| pointer-sized unsigned integer|
|numba.types.float32| f4| single-precision floating-point number|
|numba.types.float64, double| f8| double-precision floating-point number|
|numba.types.complex64| c8| single-precision complex number|
|numba.types.complex128| c16| double-precision complex number|
### Arrays
If we have arrays in the function signature, which is a very likely senario, we might want to give as much information to numpy as possible about the numpy.ndarray. In some cases it is very benificial to make a np.ndarray an array with C memory layout and tell numba about it.
We can use the numpy function **numpy.ascontiguousarray for** converting a numpy array into a C memory layout.
We can also check a numpy array, let's call it X, if it is already in the C memory layout. This is done by looking at **X.flags['C_CONTIGUOUS']**.
Some example for array signatures are:
|||
|---|---|
|numba.types.float32[:]| 1d array of float32 with no particular memory layout|
|numba.types.float32[:,:]| 2d array of float32 with no particular memory layout|
|numba.types.float32[:,:,:]| 3d array of float32 with no particular memory layout|
|numba.types.float32[::1]| 1d array of float32 with C memory layout|
|numba.types.float32[:,::1]| 2d array of float32 with C memory layout|
|numba.types.float32[:,:,::1]| 3d array of float32 with C memory layout|
|numba.types.float32[::1,:]| 2d array of float32 with Fortran memory layout|
|numba.types.float32[::1,:,:]| 3d array of float32 with Fortran memory layout|
## An example (up to 437x faster)