1Analysis Opportunities: 2 3//===---------------------------------------------------------------------===// 4 5In test/Transforms/LoopStrengthReduce/quadradic-exit-value.ll, the 6ScalarEvolution expression for %r is this: 7 8 {1,+,3,+,2}<loop> 9 10Outside the loop, this could be evaluated simply as (%n * %n), however 11ScalarEvolution currently evaluates it as 12 13 (-2 + (2 * (trunc i65 (((zext i64 (-2 + %n) to i65) * (zext i64 (-1 + %n) to i65)) /u 2) to i64)) + (3 * %n)) 14 15In addition to being much more complicated, it involves i65 arithmetic, 16which is very inefficient when expanded into code. 17 18//===---------------------------------------------------------------------===// 19 20In formatValue in test/CodeGen/X86/lsr-delayed-fold.ll, 21 22ScalarEvolution is forming this expression: 23 24((trunc i64 (-1 * %arg5) to i32) + (trunc i64 %arg5 to i32) + (-1 * (trunc i64 undef to i32))) 25 26This could be folded to 27 28(-1 * (trunc i64 undef to i32)) 29 30//===---------------------------------------------------------------------===// 31