@@ -178,24 +178,33 @@ Optimization:
178
178
Codegen:
179
179
180
180
* POWER10 support
181
- * Added PC Relative addressing
182
- * Added __int128 vector bool support
181
+
182
+ * Added PC Relative addressing
183
+ * Added __int128 vector bool support
184
+
183
185
* Security enhancement via probe-stack attribute support to protect against stack clash
184
186
* Floating point support enhancements
185
- * Improved half precision and quad precision support, including GLIBC
186
- * constrained FP operation support for arithmetic/rounding/max/min
187
- * cleaning up fast math flags checks in DAGCombine, Legalizer, and Lowering
187
+
188
+ * Improved half precision and quad precision support, including GLIBC
189
+ * constrained FP operation support for arithmetic/rounding/max/min
190
+ * cleaning up fast math flags checks in DAGCombine, Legalizer, and Lowering
191
+
188
192
* Performance improvements from instruction exploitation, especially for vector permute on LE
189
193
* Scheduling enhancements
190
- * Added MacroFusion for POWER8
191
- * Added post-ra heuristics for POWER9
194
+
195
+ * Added MacroFusion for POWER8
196
+ * Added post-ra heuristics for POWER9
197
+
192
198
* Target dependent passes tuning
193
- * Updated LoopStrengthReduce to use instruction number as first priority
194
- * Enhanced MachineCombiner to expose more ILP
199
+
200
+ * Updated LoopStrengthReduce to use instruction number as first priority
201
+ * Enhanced MachineCombiner to expose more ILP
202
+
195
203
* Code quality and maintenance enhancements
196
- * Enabled more machine verification passes
197
- * Added ability to parse and emit additional extended mnemonics
198
- * Numerous bug fixes
204
+
205
+ * Enabled more machine verification passes
206
+ * Added ability to parse and emit additional extended mnemonics
207
+ * Numerous bug fixes
199
208
200
209
AIX Support Improvements:
201
210
0 commit comments