Displaying 1 result from an estimated 1 matches for "add164".
Did you mean:
add14
2011 Nov 14
0
[LLVMdev] algebraic (de)optimizations form long chains of dependent operations
...f.else146 ], [ 2047, %if.end137 ]
%conv154 = trunc i32 %val8.0 to i16
store i16 %conv154, i16* %add.ptr36, align 2
%add = add i32 %val1.0, %sum.1
%add161 = add i32 %add, %val2.0
%add170 = add i32 %add161, %val3.0
%add167 = add i32 %add170, %val4.0
%add173 = add i32 %add167, %val5.0
%add164 = add i32 %add173, %val6.0
%add176 = add i32 %add164, %val7.0
%add179 = add i32 %add176, %val8.0
%cmp181 = icmp eq i32 %rem, 56
br i1 %cmp181, label %if.then183, label %for.inc
The problem with this bitcode is that all additions form one long chain of dependent additions, rather than a tre...