1
1
# numba-dppy
2
2
3
- ## Numba + dpCtl + dpNP = numba-dppy
3
+ ## Numba + dpPy + dpCtl + dpNP = numba-dppy
4
4
5
5
` numba-dppy ` extends Numba with new backend for support compilation
6
6
for Intel CPU and GPU architectures.
7
7
8
8
For more information about Numba, see the Numba homepage:
9
- http://numba.pydata.org
9
+ http://numba.pydata.org .
10
+
11
+ Note: ` numba-dppy ` requires patched version of Numba.
12
+ See https://github.com/IntelPython/numba .
10
13
11
14
For more information about dpCtl, see the dpCtl homepage:
12
15
https://intelpython.github.io/dpctl/
@@ -16,6 +19,47 @@ https://intelpython.github.io/dpnp/
16
19
17
20
## Dependencies
18
21
19
- * numba
20
- * dpCtl
21
- * dpNP (optional)
22
+ * numba >=0.51 (IntelPython/numba)
23
+ * dpCtl >=0.3.8
24
+ * dpNP >=0.3 (optional)
25
+ * llvm-spirv (SPIRV generation from LLVM IR)
26
+ * llvmdev (LLVM IR generation)
27
+ * spirv-tools
28
+
29
+ ## dpPy
30
+
31
+ dpPy is a proof-of-concept backend for NUMBA to support compilation for
32
+ Intel CPU and GPU architectures.
33
+ The present implementation of dpPy is based on OpenCL 2.1, but is likely
34
+ to change in the future to rely on Sycl/DPC++ or Intel Level-0 driver API.
35
+
36
+ ## Installation
37
+
38
+ Use setup.py or conda (see conda-recipe).
39
+
40
+ ## Testing
41
+
42
+ See folder ` numba_dppy/tests ` .
43
+
44
+ Run tests:
45
+ ``` bash
46
+ python -m numba.runtests numba_dppy.tests
47
+ ```
48
+
49
+ ## Examples
50
+
51
+ See folder ` numba_dppy/examples ` .
52
+
53
+ Run examples:
54
+ ``` bash
55
+ python numba_dppy/examples/sum.py
56
+ ```
57
+
58
+ ## How Tos
59
+
60
+ Refer the HowTo.rst guide for an overview of the programming semantics,
61
+ examples, supported functionalities, and known issues.
62
+
63
+ ## Reporting issues
64
+
65
+ Please use https://github.com/IntelPython/numba-dppy/issues to report issues and bugs.
0 commit comments