FazBrowse GitHub Viewer | Trending |
URL:
| Home
Tools: [Download Repo ZIP]   [Original HTTPS Page]

Update image sig purpose and strategy (#126) · oneapi-src/oneAPI-tab@9239da6 · GitHub

This repository was archived by the owner on Feb 5, 2024. It is now read-only.
  1. oneapi-src
  2. oneAPI-tab

Repository navigation

Commit 9239da6

Browse files
melonakos
authored
Update image sig purpose and strategy (#126)
* Updated the Image SIG description, adding a purpose and strategy to the document.
1 parent bfb4b42 commit 9239da6

1 file changed

Lines changed: 102 additions & 35 deletions

File tree

‎image/README.rst‎

Lines changed: 102 additions & 35 deletions
Original file line numberDiff line numberDiff line change
@@ -1,46 +1,113 @@
1-
================================
2-
oneAPI Community Forum Image SIG
3-
================================
4-
5-
The Image SIG hosts discussions and presentations for fundamental
6-
image processing routines for high-performance computing, engineering,
7-
financial, and other applications.
8-
9-
This SIG discusses the oneAPI Image Processing Library (oneIPL),
10-
currently in the planning and spec development stage. Its goal is to
11-
become an extensive library of ready-to-use, highly optimized image
12-
processing functions. As part of oneAPI, oneIPL is designed to allow
13-
execution on various computational devices: CPUs, GPUs, and other
14-
accelerators. The functionality is subdivided into several domains:
15-
filers, geometry transformations, and color and type conversions.
1+
====================================
2+
Image SIG - a oneAPI Community Forum
3+
====================================
4+
5+
Introduction
6+
============
7+
8+
The Image SIG hosts discussions and presentations for fundamental image
9+
processing routines for high-performance computing, engineering, financial, and
10+
other applications.
11+
12+
This SIG discusses the oneAPI Image Processing Library (oneIPL) in the planning
13+
and spec development stage. Its goal is to become an extensive library of
14+
ready-to-use, highly optimized image processing functions. As part of oneAPI,
15+
oneIPL is designed to allow execution on various computational devices: CPUs,
16+
GPUs, and other accelerators. The functionality is subdivided into several
17+
domains: filers, geometry transformations, and color and type conversions.
1618

1719
Its royalty-free APIs help developers:
1820

19-
1. Use vectorization and SIMD (single instruction, multiple data)
20-
instructions on CPU and GPU.
21+
1. Use vectorization and SIMD (single instruction, multiple data) instructions
22+
on CPU and GPU.
23+
24+
2. Use CPU threading and GPU SIMT (single instruction, multiple threads).
25+
26+
3. Improve the performance of computation-intensive applications with pipelines
27+
involving image processing and other operations provided as DPC++ kernel or
28+
oneAPI libraries.
29+
30+
In addition to domain-specific functionality, the Image SIG may also discuss
31+
overall design features like the execution model, memory model, or error
32+
handling. At times, the open source oneIPL Interfaces project, which implements
33+
the oneIPL specification may also be discussed.
34+
35+
Purpose of the Group
36+
====================
37+
38+
The Image Special Interest Group (Image SIG) is committed to the collaborative
39+
development, understanding, and promotion of the open standard oneIPL. We aim
40+
to build a comprehensive image-processing library to set a new standard for
41+
image processing in the oneAPI ecosystem.
42+
43+
1. Collaborate: Gather key industry stakeholders to contribute to the oneIPL
44+
standard. Engineers and software developers from across the industry will
45+
share insights, knowledge, and resources to collaboratively shape the future
46+
of the oneIPL specification.
2147

22-
2. Use CPU threading and GPU SIMT (single instruction, multiple
23-
threads).
48+
2. Innovate: Encourage innovative thinking and problem-solving to advance the
49+
future of image processing within the oneAPI ecosystem.
2450

25-
3. Improve the performance of computation-intensive applications with
26-
pipelines involving image processing and other operations provided
27-
as DPC++ kernel or oneAPI libraries.
51+
3. Educate: Share experiences and best practices related to oneIPL with the
52+
wider oneAPI community.
2853

29-
In addition to domain-specific functionality, the Image SIG may also
30-
discuss overall design features like the execution model, memory
31-
model, or error handling. At times, the open source oneIPL Interfaces
32-
project, which implements the oneIPL specification, may also be
33-
discussed.
54+
Strategy for Achieving Goals
55+
============================
56+
57+
To achieve these goals, the following strategic approach will be taken:
58+
59+
1. Community Building: Foster a vibrant and active community of engineers and
60+
software developers who meet regularly to discuss, debate, and determine the
61+
direction of oneIPL. This will be achieved through online events and a clear
62+
communication channel for group members.
63+
64+
2. Open Collaboration: Promote an open-source mindset to enable active
65+
contributions from all members. The group will focus on collaboration,
66+
maintaining a transparent process where all members have access to resources
67+
and are encouraged to provide input on development.
68+
69+
3. Training and Education: Organize training sessions, workshops, and webinars
70+
to educate members and the broader oneAPI ecosystem about the features,
71+
advantages, and potential of the oneIPL library. This will also be an
72+
opportunity to gather feedback and requirements from end users.
73+
74+
4. Regular Updates and Reviews: Schedule periodic meetings for group members to
75+
share their experiences, progress, and challenges with the oneIPL
76+
library. These sessions will act as a platform for knowledge sharing,
77+
brainstorming solutions, and shaping the roadmap for the future of the
78+
oneIPL specification.
79+
80+
5. Integration with the oneAPI Ecosystem: Work closely with the broader oneAPI
81+
community to ensure the oneIPL library seamlessly integrates with the other
82+
elements of the oneAPI ecosystem. This will help ensure the standard is
83+
accessible, beneficial, and valuable to the community.
84+
85+
Through this purpose and strategic approach, the Image Special Interest Group
86+
will shape the future of image processing via the oneIPL image processing
87+
library in alignment with the broader goals of the oneAPI ecosystem. We believe
88+
in the power of collaborative innovation to drive the future of image
89+
processing and look forward to contributing to this journey.
90+
91+
---
3492

3593
The Image SIG is led by John Melonakos.
3694

37-
To find out how to join the Image SIG `get in
38-
touch. <https://www.oneapi.io/community/>`__
95+
Upcoming Meetings
96+
=================
97+
98+
To join the Image SIG, email oneapi@codeplay.com to register for our quarterly
99+
meetings. You will be added to the mailing list and notified of upcoming
100+
events.
39101

40-
Meeting notes:
41-
==============
102+
Previous Meetings
103+
=================
42104

43-
| `November 29, 2021 <minutes/2021_11_29_Minutes.rst>`__
44-
| `December 16, 2021 <minutes/2021_12_16_Minutes.rst>`__
45-
| `February 3, 2022 <minutes/2022_02_03_Minutes.rst>`__
46-
| `February 17, 2022 <minutes/2022_02_17_Minutes.rst>`__
105+
| June 21, 2023 - `presentation <presentations/2023-06-21_Slides.pdf>`__
106+
| February 17, 2022 - `notes <minutes/2022_02_17_Minutes.rst>`__, \
107+
`presentation <presentations/2022-02-17_Slides.pdf>`__
108+
| February 3, 2022 - `notes <minutes/2022_02_03_Minutes.rst>`__, \
109+
`presentation <presentations/2022-02-03_Slides.pdf>`__
110+
| December 16, 2021 - `notes <minutes/2021_12_16_Minutes.rst>`__, \
111+
`presentation <presentations/2021-12-16_Slides.pdf>`__
112+
| November 29, 2021 - `notes <minutes/2021_11_29_Minutes.rst>`__, \
113+
`presentation <presentations/2021-11-20_Slides.pdf>`__

0 commit comments

Comments
 (0)
Back | FazBrowse Home | New Git URL