-
Bug
-
Resolution: Fixed
-
High
-
Code Generation Tools
-
CODEGEN-5777
-
-
-
Avoid mixing single and double-precision when using float multiplication, or use compiler option --legacy
-
The compiler may place an M-unit instruction too close after either a MPYSP2DP or MPYSPDP instruction. This corrupts the functioning of one of the M-unit instructions.
Compiler schedules an M-unit instruction too soon after MPYSP2DP or MPYSPDP