22
22
23
23
Non-comprehensive list of changes in this release
24
24
=================================================
25
- .. NOTE
26
- For small 1-3 sentence descriptions, just add an entry at the end of
27
- this list. If your description won't fit comfortably in one bullet
28
- point (e.g. maybe you would like to give an example of the
29
- functionality, or simply have a lot to talk about), see the `NOTE` below
30
- for adding a new subsection.
31
25
32
26
* The ``Redirects `` argument of ``llvm::sys::ExecuteAndWait `` and
33
27
``llvm::sys::ExecuteNoWait `` was changed to an ``ArrayRef `` of optional
@@ -48,17 +42,6 @@ Non-comprehensive list of changes in this release
48
42
* Preliminary support for Sanitizers and sibling features on X86(_64) NetBSD
49
43
(ASan, UBsan, TSan, MSan, SafeStack, libFuzzer).
50
44
51
- * Note..
52
-
53
- .. NOTE
54
- If you would like to document a larger change, then you can add a
55
- subsection about it right here. You can copy the following boilerplate
56
- and un-indent it (the indentation causes it to be inside this comment).
57
-
58
- Special New Feature
59
- -------------------
60
-
61
- Makes programs 10x faster by doing Special New Thing.
62
45
63
46
Changes to the LLVM IR
64
47
----------------------
@@ -72,19 +55,17 @@ Changes to the LLVM IR
72
55
to enable/disable specific floating-point optimizations. Making the
73
56
optimizer respond appropriately to these flags is an ongoing effort.
74
57
58
+
75
59
Changes to the AArch64 Target
76
60
-----------------------------
77
61
78
- During this release:
62
+ * Enabled the new GlobalISel instruction selection framework by default at `` -O0 ``.
79
63
80
- * Enabled the new GlobalISel instruction selection framework by default at ``-O0 ``.
81
64
82
65
Changes to the ARM Target
83
66
-------------------------
84
67
85
- During this release the ARM target has:
86
-
87
- * Got support for enabling SjLj exception handling on platforms where it
68
+ * Support for enabling SjLj exception handling on platforms where it
88
69
isn't the default.
89
70
90
71
@@ -93,12 +74,12 @@ Changes to the Hexagon Target
93
74
94
75
* The Hexagon backend now supports V65 ISA.
95
76
96
- * The ``-mhvx `` option now takes an optional value that specified the ISA
77
+ * The ``-mhvx `` option now takes an optional value that specifies the ISA
97
78
version of the HVX coprocessor. The available values are v60, v62 and v65.
98
79
By default, the value is set to be the same as the CPU version.
99
80
100
81
* The compiler option ``-mhvx-double `` is deprecated and will be removed in
101
- the next release of the compiler. Programmers should use ``-mhvx-length ``
82
+ the next release of the compiler. Programmers should use the ``-mhvx-length ``
102
83
option to specify the desired vector length: ``-mhvx-length=64b `` for
103
84
64-byte vectors and ``-mhvx-length=128b `` for 128-byte vectors. While the
104
85
current default vector length is 64 bytes, users should always specify the
@@ -130,8 +111,8 @@ Fixed numerous bugs:
130
111
* Corrected the encoding of movep for microMIPS32r6.
131
112
* Fixed an issue with the usage of insert instructions having an invalid set of
132
113
operands.
133
- * Fixed an issue where TLS symbols where not marked as such.
134
- * Enabled the usage of register scavanging with MSA, due to its' shorter offsets
114
+ * Fixed an issue where TLS symbols were not marked as such.
115
+ * Enabled the usage of register scavenging with MSA, due to its shorter offsets
135
116
for loads and stores.
136
117
* Corrected the ELF headers when using the DSP ASE.
137
118
@@ -152,10 +133,6 @@ Deprecation notices:
152
133
* microMIPS64R6 support was been deprecated since 5.0, and has now been
153
134
completely removed.
154
135
155
- Changes to the PowerPC Target
156
- -----------------------------
157
-
158
- During this release ...
159
136
160
137
Changes to the SystemZ Target
161
138
-----------------------------
@@ -206,30 +183,24 @@ During this release the X86 target has:
206
183
207
184
* Improved documentation for SSE/AVX intrinsics in intrin.h header files.
208
185
209
- Changes to the AMDGPU Target
210
- -----------------------------
211
-
212
- During this release ...
213
-
214
- Changes to the AVR Target
215
- -----------------------------
216
-
217
- During this release ...
218
186
219
- Changes to the OCaml bindings
220
- -----------------------------
221
-
222
- During this release ...
223
-
224
-
225
- Changes to the C API
226
- --------------------
187
+ External Open Source Projects Using LLVM 6
188
+ ==========================================
227
189
228
- During this release ...
190
+ LDC - the LLVM-based D compiler
191
+ -------------------------------
229
192
193
+ `D <http://dlang.org >`_ is a language with C-like syntax and static typing. It
194
+ pragmatically combines efficiency, control, and modeling power, with safety and
195
+ programmer productivity. D supports powerful concepts like Compile-Time Function
196
+ Execution (CTFE) and Template Meta-Programming, provides an innovative approach
197
+ to concurrency and offers many classical paradigms.
230
198
231
- External Open Source Projects Using LLVM 6
232
- ==========================================
199
+ `LDC <http://wiki.dlang.org/LDC >`_ uses the frontend from the reference compiler
200
+ combined with LLVM as backend to produce efficient native code. LDC targets
201
+ x86/x86_64 systems like Linux, OS X, FreeBSD and Windows and also Linux on ARM
202
+ and PowerPC (32/64 bit). Ports to other architectures like AArch64 and MIPS64
203
+ are underway.
233
204
234
205
JFS - JIT Fuzzing Solver
235
206
------------------------
@@ -257,21 +228,6 @@ import of .h symbols - even inline functions and macros. Zig uses LLD combined
257
228
with lazily building compiler-rt to provide out-of-the-box cross-compiling for
258
229
all supported targets.
259
230
260
- LDC - the LLVM-based D compiler
261
- -------------------------------
262
-
263
- `D <http://dlang.org >`_ is a language with C-like syntax and static typing. It
264
- pragmatically combines efficiency, control, and modeling power, with safety and
265
- programmer productivity. D supports powerful concepts like Compile-Time Function
266
- Execution (CTFE) and Template Meta-Programming, provides an innovative approach
267
- to concurrency and offers many classical paradigms.
268
-
269
- `LDC <http://wiki.dlang.org/LDC >`_ uses the frontend from the reference compiler
270
- combined with LLVM as backend to produce efficient native code. LDC targets
271
- x86/x86_64 systems like Linux, OS X, FreeBSD and Windows and also Linux on ARM
272
- and PowerPC (32/64 bit). Ports to other architectures like AArch64 and MIPS64
273
- are underway.
274
-
275
231
Additional Information
276
232
======================
277
233
0 commit comments