Displaying 1 result from an estimated 1 matches for "storemerge6".
Did you mean:
storemerge
2012 Jan 14
0
[LLVMdev] Vector ops out of loops
...i32* @i, align 4
%c.promoted = load <2 x i16>* @c, align 4
br label %for.body
for.body: ; preds = %entry,
%for.body
%inc7 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
%0 = phi <2 x i16> [ %c.promoted, %entry ], [ %6, %for.body ]
%storemerge6.off0 = phi i16 [ 0, %entry ], [ %extract.t, %for.body ]
%arrayidx = getelementptr inbounds [50 x <2 x i16>]* @a, i16 0, i16
%storemerge6.off0
%1 = load <2 x i16>* %arrayidx, align 4
%2 = extractelement <2 x i16> %1, i32 0
%arrayidx1 = getelementptr inbounds [50 x <...