<html>
<head>
<base href="https://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 --- - Poor VBROADCASTF128 / VBROADCASTI128 support"
href="https://llvm.org/bugs/show_bug.cgi?id=26235">26235</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>Poor VBROADCASTF128 / VBROADCASTI128 support
</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>Windows NT
</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>Backend: X86
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>llvm-dev@redking.me.uk
</td>
</tr>
<tr>
<th>CC</th>
<td>ahmed.bougacha@gmail.com, llvm-bugs@lists.llvm.org, spatel+llvm@rotateright.com
</td>
</tr>
<tr>
<th>Classification</th>
<td>Unclassified
</td>
</tr></table>
<p>
<div>
<pre>The use of the VBROADCASTF128 128-bit vector broadcast instruction is only
available through the __builtin_ia32_vbroadcastf128_ps256 /
__builtin_ia32_vbroadcastf128_pd256 intrinsics.
The VBROADCASTI128 128-bit vector broadcast instruction appears to be converted
to native IR but I haven't found a case that lowers to it.
This was mentioned in PR23886.
define <8 x i32> @broadcast_4i32(<4 x i32>* %p) {
%v128 = load <4 x i32>, <4 x i32>* %p, align 16
%v256 = shufflevector <4 x i32> %v128, <4 x i32> undef, <8 x i32> <i32 0, i32
1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
ret <8 x i32> %v256
}
define <4 x i64> @broadcast_2i64(<2 x i64>* %p) {
%v128 = load <2 x i64>, <2 x i64>* %p, align 16
%v256 = shufflevector <2 x i64> %v128, <2 x i64> undef, <4 x i32> <i32 0, i32
1, i32 0, i32 1>
ret <4 x i64> %v256
}
define <8 x float> @broadcast_4f32(<4 x float>* %p) {
%v128 = load <4 x float>, <4 x float>* %p, align 16
%v256 = shufflevector <4 x float> %v128, <4 x float> undef, <8 x i32> <i32 0,
i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
ret <8 x float> %v256
}
define <4 x double> @broadcast_2f64(<2 x double>* %p) {
%v128 = load <2 x double>, <2 x double>* %p, align 16
%v256 = shufflevector <2 x double> %v128, <2 x double> undef, <4 x i32> <i32
0, i32 1, i32 0, i32 1>
ret <4 x double> %v256
}
(llc -mtriple=x86_64-unknown -mattr=+avx2):
broadcast_4i32: # @broadcast_4i32
vmovaps (%rdi), %xmm0
vinsertf128 $1, %xmm0, %ymm0, %ymm0
retq
broadcast_2i64: # @broadcast_2i64
vmovaps (%rdi), %xmm0
vinsertf128 $1, %xmm0, %ymm0, %ymm0
retq
broadcast_4f32: # @broadcast_4f32
vmovaps (%rdi), %xmm0
vinsertf128 $1, %xmm0, %ymm0, %ymm0
retq
broadcast_2f64: # @broadcast_2f64
vmovaps (%rdi), %xmm0
vinsertf128 $1, %xmm0, %ymm0, %ymm0
retq</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>