-
Notifications
You must be signed in to change notification settings - Fork 68
/
ChangeLog
79 lines (70 loc) · 2.61 KB
/
ChangeLog
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
2012-08-02
----------
Changes:
-Added new parallel GPU Huffman encoder for compute capabilities >= 2.0
-Rewritten GPU Huffman decoder
-Rewritten GPU forward DCT, preserving precision of fixed point implementation
and gaining better performance than NPP implementation.
(Both old implementations were thus removed.)
-Rewritten GPU encoder preprocessor to remove unnecessary operations
-Minor performance improvement in GPU encoder postprocessor
2012-03-29
----------
Changes:
-Optionally the encoder input or decoder output can be loaded/store
from/to OpenGL texture.
2012-03-09
----------
Changes:
-Refactored color spaces transformations, added new color spaces
-Added new implementation of DCT/IDCT on CUDA, it is slightly slower
than the NPP implementation, but the new IDCT doesn't cause color space
change. It can be turned on by Makefile option. By default is NPP
version used.
2012-02-24
----------
Changes:
-Slightly improved performance of encoder.
-Correction of segment info for large images
2012-02-21
----------
Changes:
-Added option to encoder for generating segment info into APP13 header that is used
by decoder to perform fast stream parsing.
-Renamed --interleving option to --interleaved and --chroma-subsampling to --subsampled.
2012-02-16
----------
Changes:
-Added option for verbose output from console application.
-Library now stores all coder time durations inside coder structure and it
can be accessed from userspace.
-Console application now prints GPU allocation info (when --verbose is used).
-Correction of encoding/decoding for large images (e.g. 4320p or 8K)
2012-02-07
----------
-Improved preprocessor and postprocessor performance.
-Added recovering from error inside JPEG stream to decoder.
-Library provides function for listing CUDA devices.
-Library provides option for decoder to output result right into OpenGL PBO resource
which improves performance when the result should be displayed on the same GPU.
-Correction of build warnings.
-Correction of other errors.
2012-01-04
----------
Changes:
-Added option to libgpujpeg Makefile for moving gpu huffman coder tables into constant memory.
-Option is default set to true and makes better performance on older GPUs.
2012-01-03
----------
Changes:
-Added interleaved mode (optionally).
-Added subsampling (optionally) - better performance when used
(because of smaller amount of encoded/decoded data).
2011-12-07
----------
First public release of gpujpeg library and console
application.
Features:
-Baseline JPEG, non-interleaved mode on CUDA-enabled GPUs.
-Performance - realtime HD and 4K encoding/decoding on NVIDIA GTX 580.
-Using CUDA and NPP library.