RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 19:23:23 +01:00
Code Issues Projects Releases Wiki Activity

Generalize @llvm.ctlz, @llvm.cttz, and @llvm.ctpop to work on vectors of integers, and fix the one optimization pass that I'm aware of that needs updating for this. At least one current target, ARM NEON, can implement these operations on vectors directly.

Browse Source 
llvm-svn: 134265
This commit is contained in:
Owen Anderson 2011-07-01 21:52:38 +00:00
parent c3fee5e2c7
commit dccc4e4b9a
2 changed files with 27 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
docs/LangRef.html
View File

@ -6626,7 +6626,8 @@ LLVM</a>.</p>
<h5>Syntax:</h5>
<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
width. Not all targets support all bit widths however.</p>
width, or on any vector with integer elements. Not all targets support all
bit widths or vector types, however.</p>
<pre>
declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
@ -6634,6 +6635,7 @@ LLVM</a>.</p>
declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
declare &lt;2 x i32&gt; @llvm.ctpop.v2i32(&lt;2 x i32&gt; &lt;src&gt;)
</pre>
<h5>Overview:</h5>
@ -6642,10 +6644,12 @@ LLVM</a>.</p>
<h5>Arguments:</h5>
<p>The only argument is the value to be counted. The argument may be of any
integer type. The return type must match the argument type.</p>
integer type, or a vector with integer elements.
The return type must match the argument type.</p>
<h5>Semantics:</h5>
<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable, or within each
element of a vector.</p>
</div>
@ -6658,7 +6662,8 @@ LLVM</a>.</p>
<h5>Syntax:</h5>
<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
integer bit width. Not all targets support all bit widths however.</p>
integer bit width, or any vector whose elements are integers. Not all
targets support all bit widths or vector types, however.</p>
<pre>
declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
@ -6666,6 +6671,7 @@ LLVM</a>.</p>
declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
declare &lt;2 x i32&gt; @llvm.ctlz.v2i32(&lt;2 x i32&gt; &lt;src;gt)
</pre>
<h5>Overview:</h5>
@ -6674,11 +6680,13 @@ LLVM</a>.</p>
<h5>Arguments:</h5>
<p>The only argument is the value to be counted. The argument may be of any
integer type. The return type must match the argument type.</p>
integer type, or any vector type with integer element type.
The return type must match the argument type.</p>
<h5>Semantics:</h5>
<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
zeros in a variable. If the src == 0 then the result is the size in bits of
zeros in a variable, or within each element of the vector if the operation
is of vector type. If the src == 0 then the result is the size in bits of
the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
</div>
@ -6692,7 +6700,8 @@ LLVM</a>.</p>
<h5>Syntax:</h5>
<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
integer bit width. Not all targets support all bit widths however.</p>
integer bit width, or any vector of integer elements. Not all targets
support all bit widths or vector types, however.</p>
<pre>
declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
@ -6700,6 +6709,7 @@ LLVM</a>.</p>
declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
declase &lt;2 x i32&gt; @llvm.cttz.v2i32(&lt;2 x i32&gt; &lt;src&gt;)
</pre>
<h5>Overview:</h5>
@ -6708,11 +6718,13 @@ LLVM</a>.</p>
<h5>Arguments:</h5>
<p>The only argument is the value to be counted. The argument may be of any
integer type. The return type must match the argument type.</p>
integer type, or a vectory with integer element type.. The return type
must match the argument type.</p>
<h5>Semantics:</h5>
<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
zeros in a variable. If the src == 0 then the result is the size in bits of
zeros in a variable, or within each element of a vector.
If the src == 0 then the result is the size in bits of
the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
</div>

8
lib/Transforms/InstCombine/InstCombineCalls.cpp
View File

@ -355,7 +355,9 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) {
case Intrinsic::cttz: {
// If all bits below the first known one are known zero,
// this value is constant.
const IntegerType *IT = cast<IntegerType>(II->getArgOperand(0)->getType());
const IntegerType *IT = dyn_cast<IntegerType>(II->getArgOperand(0)->getType());
// FIXME: Try to simplify vectors of integers.
if (!IT) break;
uint32_t BitWidth = IT->getBitWidth();
APInt KnownZero(BitWidth, 0);
APInt KnownOne(BitWidth, 0);
@ -372,7 +374,9 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) {
case Intrinsic::ctlz: {
// If all bits above the first known one are known zero,
// this value is constant.
const IntegerType *IT = cast<IntegerType>(II->getArgOperand(0)->getType());
const IntegerType *IT = dyn_cast<IntegerType>(II->getArgOperand(0)->getType());
// FIXME: Try to simplify vectors of integers.
if (!IT) break;
uint32_t BitWidth = IT->getBitWidth();
APInt KnownZero(BitWidth, 0);
APInt KnownOne(BitWidth, 0);