File beignet.changes of Package beignet
242
1
-------------------------------------------------------------------
2
Thu Jan 24 12:09:11 UTC 2019 - alarrosa@suse.com
3
4
- Add a _constraints file to specify this package needs at least
5
6500 MB of memory to build. Also use %limit_build to be sure
6
parallel builds are taken into account.
7
8
-------------------------------------------------------------------
9
Sat Jan 19 19:17:40 UTC 2019 - Linnaea Lavia <linnaea@lavia.moe>
10
11
- Add Coffee Lake support, reusing Kaby Lake code.
12
* 0004-Enable-Coffee-Lake-support.patch
13
14
-------------------------------------------------------------------
15
Mon Jan 7 04:23:07 UTC 2019 - Linnaea Lavia <linnaea@lavia.moe>
16
17
- Fix build failure with llvm7 with patch from upstream Git:
18
* 0008-Add-preliminary-LLVM-7-support.patch (fate#325696)
19
- Fix self-test failures on some systems:
20
* beignet-disable-NegAddOptimization.patch
21
22
-------------------------------------------------------------------
23
Wed Oct 31 19:03:26 UTC 2018 - Bernhard Wiedemann <bwiedemann@suse.com>
24
25
- drop indeterministic pch files to make package build reproducible
26
27
-------------------------------------------------------------------
28
Wed Aug 8 15:38:43 UTC 2018 - alarrosa@suse.com
29
30
- Add patch beignet-llvm6.patch to fix build failure with llvm6
31
- Fix installation of broken symlink for README.md
32
- Use %license
33
34
-------------------------------------------------------------------
35
Wed Jan 31 14:54:42 UTC 2018 - msrb@suse.com
36
37
- Remove dependency on clang-devel-static. (bnc#1065464)
38
* It was removed, clang-devel now again provides everything
39
necessary as shared libraries.
40
41
-------------------------------------------------------------------
42
Thu Oct 26 11:28:51 UTC 2017 - mpluskal@suse.com
43
44
- Update to version 1.3.2:
45
* Add some Gemini Lake devices support.
46
* Add llvm4.0 and llvm5.0 support.
47
* Add cl_intel_required_subgroup_size extension support.
48
* Fix some other bugs.
49
- Drop no longer needed patches:
50
* 0001-Backend-Remove-old-llvm-support-code.patch
51
* 0002-Backend-Fix-an-include-file-error-problem.patch
52
* 0003-Backend-Refine-GEP-lowering-code.patch
53
* 0004-Backend-Refine-LLVM-version-check-macro.patch
54
* 0005-Backend-Refine-FCmp-one-and-une.patch
55
* 0006-utest-fix-image-qualifier-of-compiler_fill_gl_image-.patch
56
* 0007-Backend-Add-LLVM40-support.patch
57
58
-------------------------------------------------------------------
59
Thu May 4 19:47:06 UTC 2017 - mardnh@gmx.de
60
61
- Add patches for LLVM 4.0 support
62
* 0001-Backend-Remove-old-llvm-support-code.patch
63
* 0002-Backend-Fix-an-include-file-error-problem.patch
64
* 0003-Backend-Refine-GEP-lowering-code.patch
65
* 0004-Backend-Refine-LLVM-version-check-macro.patch
66
* 0005-Backend-Refine-FCmp-one-and-une.patch
67
* 0006-utest-fix-image-qualifier-of-compiler_fill_gl_image-.patch
68
* 0007-Backend-Add-LLVM40-support.patch
69
70
-------------------------------------------------------------------
71
Tue Mar 14 11:47:10 UTC 2017 - mpluskal@suse.com
72
73
- Update to version 1.3.1:
74
* OpenCL 2.0 support.
75
* OpenCL event and enqueue module re-implement.
76
* Other OpenCL runtime driver refine.
77
* LLVM 3.9 support.
78
* Extension cl_khr_gl_sharing support.
79
* Extension intel_subgroups_short support.
80
* Large kernel compiling speed up.
81
* Register allocation improvement.
82
* Enable OpenCL 2.0 where device supported.
83
* Optimize int64 multiply int64.
84
* Add cl_khr_gl_sharing example and document.
85
* Fix some bugs.
86
87
-------------------------------------------------------------------
88
Sun Jan 1 12:41:01 UTC 2017 - jengelh@inai.de
89
90
- Trim filler words from description.
91
92
-------------------------------------------------------------------
93
Sat Nov 12 10:44:33 UTC 2016 - mpluskal@suse.com
94
95
- Update to version 1.2.1
96
* Add some Apollo Lake devices support.
97
* Add llvm3.9 support.
98
* Fix some cl_intel_subgroups extension bugs.
99
* Avoid too many messages when the driver could not find good
100
work group size.
101
* Fix some other bugs.
102
103
-------------------------------------------------------------------
104
Tue Aug 30 16:48:42 UTC 2016 - mardnh@gmx.de
105
106
- Update to 1.2.0
107
* 6th generation Intel Atom Processors (Broxton) support.
108
* 6th generation Intel Core Processors (Kabylake) support.
109
* LLVM 3.7 and 3.8 support.
110
* ASM debug infomation and profiling support.
111
* Experimental double data type support for processors after 6th generation(Broadwell) .
112
* Full cl_intel_subgroups extension.
113
* OpenCL 2.0 workgroup built-in functions.
114
* Local copy propagation optimization and other optimizations
115
* intel_accelerator extension and basic intel_motion_estimation extension.
116
* Android build.
117
* Refined printf implementation.
118
* Bug fixes since last release.
119
- Drop no longer needed patches (fixed upstream)
120
* gcc6-support.patch
121
* llvm38-support.patch
122
* std-isnan-isinf.patch
123
124
-------------------------------------------------------------------
125
Mon Jul 4 18:29:03 UTC 2016 - mardnh@gmx.de
126
127
- Add patches to support build with GCC6.1
128
* gcc6-support.patch
129
* llvm38-support.patch
130
* std-isnan-isinf.patch
131
132
-------------------------------------------------------------------
133
Sun May 8 13:10:31 UTC 2016 - mpluskal@suse.com
134
135
- Update to 1.1.2
136
* Add Broxton support.
137
* Add llvm3.7 support.
138
* Fix event bugs.
139
* Fix the annoying "Failed to release userptr..." error message.
140
* Fix printf bugs.
141
- Drop no longer needed beignet-llvm37.diff
142
143
-------------------------------------------------------------------
144
Fri Dec 18 12:48:30 UTC 2015 - mpluskal@suse.com
145
146
- Update to 1.1.1
147
* Dump asm to a file.
148
* Use addRemappedFile to avoid creating temporary cl source file.
149
* add collectInsnNum to collect block instruction number.
150
* Fix piglit clLinkProgram fail.
151
* Calculate appropriate timestamps for cl profile.
152
- Use cmake macros
153
154
-------------------------------------------------------------------
155
Thu Oct 22 17:58:29 UTC 2015 - normand@linux.vnet.ibm.com
156
157
- add ExclusiveArch %{ix86} x86_64 to avoid build error other archi
158
159
-------------------------------------------------------------------
160
Tue Oct 13 22:12:17 UTC 2015 - mardnh@gmx.de
161
162
- add patch: beignet-llvm37.diff to support llvm 3.7
163
- add rpmlintrc
164
165
-------------------------------------------------------------------
166
Fri Jul 31 07:54:13 UTC 2015 - mardnh@gmx.de
167
168
- update to version 1.1.0
169
The highlighted improvements are as below:
170
* Added 6th generation Intel Core Processors (SKL) support.
171
* Added 5th generation Intel Atom Processors (BSW) support.
172
* Re-implement structurized control flow algorithm.
173
* Added half data type support after BDW.
174
* Implement a new BTI solution to support dynamic bti.
175
* Added extension cl_intel_subgroups support.
176
* Replace some built-ins with llvm intrinsic.
177
* Add Indirect structure argument read support.
178
* Enable cl_khr_spir extension to build and run from SPIR binary.
179
* Improvement some opencv test suite performance on all platforms.
180
* Several other bug fixes since last release.
181
- removed patch beignet-icd-name-fix.diff (included upstream)
182
183
-------------------------------------------------------------------
184
Tue Mar 17 20:30:46 UTC 2015 - mardnh@gmx.de
185
186
- update to version 1.0.2
187
It's a bug fix release including some bug fixes, minor new
188
features and some performance improvements.
189
The highlighted improvements are as below:
190
* Added LLVM 3.6 support.
191
* Fixed a bug in popcount().
192
* Refactor the legalize pass and const expression handling.
193
* Fixed a bug when hardware is inaccessible.
194
* Enabled multiarch 32bit and 64bit coexisting.
195
* Implemented some new device side clCopyImage kernels.
196
* Optimize clEnqueueCopyImageToBuffer for 16 aligned buffer.
197
198
-------------------------------------------------------------------
199
Mon Jan 19 18:06:27 UTC 2015 - mardnh@gmx.de
200
201
- update to version 1.0.1
202
* Enable userptr support by default thus it's possible to achieve zero-copy
203
when create a buffer object.
204
* Change to use TILING_Y mode on BDW platform which lead to a major performance
205
improvement for some cases.
206
* Fix one liveness bug which is the last such type of known bug in the
207
Gen Backend.
208
* Fix some builtin math function's accuracy bugs.
209
* Improve the strict mode sin/cos implementation, reduce the instruction count
210
from 1700 to 400.
211
* Use Clang native sampler and image types, now image/sampler are fully comply
212
with OpenCL spec.
213
* Fix bugs triggered by some popular applications such as darktable.
214
Now darktable works fine on all the supported platforms.
215
* Add support for old system which doesn't have c++11 features.
216
217
-------------------------------------------------------------------
218
Sat Nov 15 22:25:21 UTC 2014 - mardnh@gmx.de
219
220
- update to version 1.0.0
221
- removed patch fix_no-return-in-nonvoid-function-llvm_gen_backend.diff
222
since it is now included upstream
223
224
-------------------------------------------------------------------
225
Sat Oct 18 10:59:35 UTC 2014 - mardnh@gmx.de
226
227
- added patch: fix_no-return-in-nonvoid-function-llvm_gen_backend.diff
228
to address a build rpm-post build error
229
------------------------------------------
230
E: beignet no-return-in-nonvoid-function
231
backend/src/llvm/llvm_gen_backend.cpp:1108
232
------------------------------------------
233
- add ncurses-devel to BuildRequires
234
- added build flags CMAKE_BUILD_TYPE=Release
235
236
-------------------------------------------------------------------
237
Tue Oct 7 18:55:41 UTC 2014 - mardnh@gmx.de
238
239
- initial stable package, version 0.9.3
240
based on home:mnhauke:opencl:testing/beignet
241
242