<html>
<head>
<base href="http://llvm.org/bugs/" />
</head>
<body><table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Bug ID</th>
<td><a class="bz_bug_link
bz_status_NEW "
title="NEW --- - Loop Vectorizer strips address space id when generating a bitcast to/from a vector type"
href="http://llvm.org/bugs/show_bug.cgi?id=16306">16306</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>Loop Vectorizer strips address space id when generating a bitcast to/from a vector type
</td>
</tr>
<tr>
<th>Product</th>
<td>libraries
</td>
</tr>
<tr>
<th>Version</th>
<td>trunk
</td>
</tr>
<tr>
<th>Hardware</th>
<td>PC
</td>
</tr>
<tr>
<th>OS</th>
<td>Linux
</td>
</tr>
<tr>
<th>Status</th>
<td>NEW
</td>
</tr>
<tr>
<th>Severity</th>
<td>normal
</td>
</tr>
<tr>
<th>Priority</th>
<td>P
</td>
</tr>
<tr>
<th>Component</th>
<td>Loop Optimizer
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>pekka.jaaskelainen@tut.fi
</td>
</tr>
<tr>
<th>CC</th>
<td>llvmbugs@cs.uiuc.edu
</td>
</tr>
<tr>
<th>Classification</th>
<td>Unclassified
</td>
</tr></table>
<p>
<div>
<pre>$ cat copy.c
char Y[4*10000] __attribute__((address_space(1)));
char X[4*10000];
int main() {
for (int i = 0; i < 4*10000; ++i)
X[i] = Y[i] + 1;
return 0;
}
$ clang -O3 copy.c -S -emit-llvm -c -O3 -o -
...
define i32 @main() #0 {
vector.ph:
br label %vector.body
vector.body: ; preds = %vector.body,
%vector.ph
%index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ]
%0 = getelementptr inbounds [40000 x i8] addrspace(1)* @Y, i64 0, i64 %index
%1 = bitcast i8 addrspace(1)* %0 to <16 x i8>*
%wide.load = load <16 x i8>* %1, align 16
%.sum13 = or i64 %index, 16
%2 = getelementptr [40000 x i8] addrspace(1)* @Y, i64 0, i64 %.sum13
%3 = bitcast i8 addrspace(1)* %2 to <16 x i8>*
; ^-------------- breaks here
%wide.load10 = load <16 x i8>* %3, align 16
%4 = add <16 x i8> %wide.load, <i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8
1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1>
%5 = add <16 x i8> %wide.load10, <i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1,
i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1>
%6 = getelementptr inbounds [40000 x i8]* @X, i64 0, i64 %index
%7 = bitcast i8* %6 to <16 x i8>*
store <16 x i8> %4, <16 x i8>* %7, align 16
%.sum14 = or i64 %index, 16
%8 = getelementptr [40000 x i8]* @X, i64 0, i64 %.sum14
%9 = bitcast i8* %8 to <16 x i8>*
store <16 x i8> %5, <16 x i8>* %9, align 16
%index.next = add i64 %index, 32
%10 = icmp eq i64 %index.next, 40000
br i1 %10, label %for.end, label %vector.body
for.end: ; preds = %vector.body
ret i32 0
}
...
Is there a nice trick to disable loop vectorizer at the opt command line when
-O3 is used?</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are on the CC list for the bug.</li>
</ul>
</body>
</html>