[llvm] [AArch64][CostModel] Increase the cost of illegal SVE int-to-fp converts (PR #130756)

[David Sherwood via llvm-commits]

================
@@ -1546,93 +1546,93 @@ define i32 @casts_no_users() {
   %r168 = fptoui <16 x double> undef to <16 x i64>
   %r169 = fptosi <16 x double> undef to <16 x i64>
 
-  %r170 = uitofp <2 x i1> undef to <2 x float>
-  %r171 = sitofp <2 x i1> undef to <2 x float>
-  %r172 = uitofp <2 x i8> undef to <2 x float>
-  %r173 = sitofp <2 x i8> undef to <2 x float>
-  %r174 = uitofp <2 x i16> undef to <2 x float>
-  %r175 = sitofp <2 x i16> undef to <2 x float>
-  %r176 = uitofp <2 x i32> undef to <2 x float>
-  %r177 = sitofp <2 x i32> undef to <2 x float>
-  %r178 = uitofp <2 x i64> undef to <2 x float>
-  %r179 = sitofp <2 x i64> undef to <2 x float>
+  %r170 = uitofp <2 x i1> poison to <2 x float>
+  %r171 = sitofp <2 x i1> poison to <2 x float>
+  %r172 = uitofp <2 x i8> poison to <2 x float>
+  %r173 = sitofp <2 x i8> poison to <2 x float>
+  %r174 = uitofp <2 x i16> poison to <2 x float>
+  %r175 = sitofp <2 x i16> poison to <2 x float>
+  %r176 = uitofp <2 x i32> poison to <2 x float>
+  %r177 = sitofp <2 x i32> poison to <2 x float>
+  %r178 = uitofp <2 x i64> poison to <2 x float>
+  %r179 = sitofp <2 x i64> poison to <2 x float>
 
-  %r180 = uitofp <2 x i1> undef to <2 x double>
-  %r181 = sitofp <2 x i1> undef to <2 x double>
-  %r182 = uitofp <2 x i8> undef to <2 x double>
-  %r183 = sitofp <2 x i8> undef to <2 x double>
-  %r184 = uitofp <2 x i16> undef to <2 x double>
-  %r185 = sitofp <2 x i16> undef to <2 x double>
-  %r186 = uitofp <2 x i32> undef to <2 x double>
-  %r187 = sitofp <2 x i32> undef to <2 x double>
-  %r188 = uitofp <2 x i64> undef to <2 x double>
-  %r189 = sitofp <2 x i64> undef to <2 x double>
+  %r180 = uitofp <2 x i1> poison to <2 x double>
+  %r181 = sitofp <2 x i1> poison to <2 x double>
+  %r182 = uitofp <2 x i8> poison to <2 x double>
+  %r183 = sitofp <2 x i8> poison to <2 x double>
+  %r184 = uitofp <2 x i16> poison to <2 x double>
+  %r185 = sitofp <2 x i16> poison to <2 x double>
+  %r186 = uitofp <2 x i32> poison to <2 x double>
+  %r187 = sitofp <2 x i32> poison to <2 x double>
+  %r188 = uitofp <2 x i64> poison to <2 x double>
+  %r189 = sitofp <2 x i64> poison to <2 x double>
 
-  %r190 = uitofp <4 x i1> undef to <4 x float>
-  %r191 = sitofp <4 x i1> undef to <4 x float>
-  %r192 = uitofp <4 x i8> undef to <4 x float>
-  %r193 = sitofp <4 x i8> undef to <4 x float>
-  %r194 = uitofp <4 x i16> undef to <4 x float>
-  %r195 = sitofp <4 x i16> undef to <4 x float>
-  %r196 = uitofp <4 x i32> undef to <4 x float>
-  %r197 = sitofp <4 x i32> undef to <4 x float>
-  %r198 = uitofp <4 x i64> undef to <4 x float>
-  %r199 = sitofp <4 x i64> undef to <4 x float>
+  %r190 = uitofp <4 x i1> poison to <4 x float>
+  %r191 = sitofp <4 x i1> poison to <4 x float>
+  %r192 = uitofp <4 x i8> poison to <4 x float>
+  %r193 = sitofp <4 x i8> poison to <4 x float>
+  %r194 = uitofp <4 x i16> poison to <4 x float>
+  %r195 = sitofp <4 x i16> poison to <4 x float>
+  %r196 = uitofp <4 x i32> poison to <4 x float>
+  %r197 = sitofp <4 x i32> poison to <4 x float>
+  %r198 = uitofp <4 x i64> poison to <4 x float>
+  %r199 = sitofp <4 x i64> poison to <4 x float>
 
-  %r200 = uitofp <4 x i1> undef to <4 x double>
-  %r201 = sitofp <4 x i1> undef to <4 x double>
-  %r202 = uitofp <4 x i8> undef to <4 x double>
-  %r203 = sitofp <4 x i8> undef to <4 x double>
-  %r204 = uitofp <4 x i16> undef to <4 x double>
-  %r205 = sitofp <4 x i16> undef to <4 x double>
-  %r206 = uitofp <4 x i32> undef to <4 x double>
-  %r207 = sitofp <4 x i32> undef to <4 x double>
-  %r208 = uitofp <4 x i64> undef to <4 x double>
-  %r209 = sitofp <4 x i64> undef to <4 x double>
+  %r200 = uitofp <4 x i1> poison to <4 x double>
----------------
david-arm wrote:

When making cost model changes I've typically avoided converting undef to poison to avoid the github pre-commit errors to make patches easier to review. If you do want to update the cost model tests to use poison I think that's best done as a NFC patch separately?

https://github.com/llvm/llvm-project/pull/130756