<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<meta http-equiv="content-type" content="text/html; charset=utf-8">
<title>llvmpipe</title>
<link rel="stylesheet" type="text/css" href="mesa.css">
</head>
<body>
<div class="header">
<h1>The Mesa 3D Graphics Library</h1>
</div>
<iframe src="contents.html"></iframe>
<div class="content">
<h1>Introduction</h1>
<p>
The Gallium llvmpipe driver is a software rasterizer that uses LLVM to
do runtime code generation.
Shaders, point/line/triangle rasterization and vertex processing are
implemented with LLVM IR which is translated to x86, x86-64, or ppc64le machine
code.
Also, the driver is multithreaded to take advantage of multiple CPU cores
(up to 8 at this time).
It's the fastest software rasterizer for Mesa.
</p>
<h1>Requirements</h1>
<ul>
<li>
<p>
For x86 or amd64 processors, 64-bit mode is recommended.
Support for SSE2 is strongly encouraged. Support for SSE3 and SSE4.1 will
yield the most efficient code. The fewer features the CPU has the more
likely it is that you will run into underperforming, buggy, or incomplete code.
</p>
<p>
For ppc64le processors, use of the Altivec feature (the Vector
Facility) is recommended if supported; use of the VSX feature (the
Vector-Scalar Facility) is recommended if supported AND Mesa is
built with LLVM version 4.0 or later.
</p>
<p>
See /proc/cpuinfo to know what your CPU supports.
</p>
</li>
<li>
<p>Unless otherwise stated, LLVM version 3.4 is recommended; 3.3 or later is required.</p>
<p>
For Linux, on a recent Debian based distribution do:
</p>
<pre>
aptitude install llvm-dev
</pre>
<p>
If you want development snapshot builds of LLVM for Debian and derived
distributions like Ubuntu, you can use the APT repository at <a
href="https://apt.llvm.org/" title="Debian Development packages for LLVM"
>apt.llvm.org</a>, which are maintained by Debian's LLVM maintainer.
</p>
<p>
For a RPM-based distribution do:
</p>
<pre>
yum install llvm-devel
</pre>
<p>
For Windows you will need to build LLVM from source with MSVC or MINGW
(either natively or through cross compilers) and CMake, and set the LLVM
environment variable to the directory you installed it to.
LLVM will be statically linked, so when building on MSVC it needs to be
built with a matching CRT as Mesa, and you'll need to pass
<code>-DLLVM_USE_CRT_xxx=yyy</code> as described below.
</p>
<table border="1">
<tr>
<th rowspan="2">LLVM build-type</th>
<th colspan="2" align="center">Mesa build-type</th>
</tr>
<tr>
<th>debug,checked</th>
<th>release,profile</th>
</tr>
<tr>
<th>Debug</th>
<td><code>-DLLVM_USE_CRT_DEBUG=MTd</code></td>
<td><code>-DLLVM_USE_CRT_DEBUG=MT</code></td>
</tr>
<tr>
<th>Release</th>
<td><code>-DLLVM_USE_CRT_RELEASE=MTd</code></td>
<td><code>-DLLVM_USE_CRT_RELEASE=MT</code></td>
</tr>
</table>
<p>
You can build only the x86 target by passing -DLLVM_TARGETS_TO_BUILD=X86
to cmake.
</p>
</li>
<li>
<p>scons (optional)</p>
</li>
</ul>
<h1>Building</h1>
To build everything on Linux invoke scons as:
<pre>
scons build=debug libgl-xlib
</pre>
Alternatively, you can build it with autoconf/make with:
<pre>
./configure --enable-glx=gallium-xlib --with-gallium-drivers=swrast --disable-dri --disable-gbm --disable-egl
make
</pre>
but the rest of these instructions assume that scons is used.
For Windows the procedure is similar except the target:
<pre>
scons platform=windows build=debug libgl-gdi
</pre>
<h1>Using</h1>
<h2>Linux</h2>
<p>On Linux, building will create a drop-in alternative for libGL.so into</p>
<pre>
build/foo/gallium/targets/libgl-xlib/libGL.so
</pre>
or
<pre>
lib/gallium/libGL.so
</pre>
<p>To use it set the LD_LIBRARY_PATH environment variable accordingly.</p>
<p>For performance evaluation pass build=release to scons, and use the corresponding
lib directory without the "-debug" suffix.</p>
<h2>Windows</h2>
<p>
On Windows, building will create
<code>build/windows-x86-debug/gallium/targets/libgl-gdi/opengl32.dll</code>
which is a drop-in alternative for system's <code>opengl32.dll</code>. To use
it put it in the same directory as your application. It can also be used by
replacing the native ICD driver, but it's quite an advanced usage, so if you
need to ask, don't even try it.
</p>
<p>
There is however an easy way to replace the OpenGL software renderer that comes
with Microsoft Windows 7 (or later) with llvmpipe (that is, on systems without
any OpenGL drivers):
</p>
<ul>
<li><p>copy build/windows-x86-debug/gallium/targets/libgl-gdi/opengl32.dll to C:\Windows\SysWOW64\mesadrv.dll</p></li>
<li><p>load this registry settings:</p>
<pre>REGEDIT4
; https://technet.microsoft.com/en-us/library/cc749368.aspx
; https://www.msfn.org/board/topic/143241-portable-windows-7-build-from-winpe-30/page-5#entry942596
[HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\Microsoft\Windows NT\CurrentVersion\OpenGLDrivers\MSOGL]
"DLL"="mesadrv.dll"
"DriverVersion"=dword:00000001
"Flags"=dword:00000001
"Version"=dword:00000002
</pre>
</li>
<li>Ditto for 64 bits drivers if you need them.</li>
</ul>
<h1>Profiling</h1>
<p>
To profile llvmpipe you should build as
</p>
<pre>
scons build=profile <same-as-before>
</pre>
<p>
This will ensure that frame pointers are used both in C and JIT functions, and
that no tail call optimizations are done by gcc.
</p>
<h2>Linux perf integration</h2>
<p>
On Linux, it is possible to have symbol resolution of JIT code with <a href="https://perf.wiki.kernel.org/">Linux perf</a>:
</p>
<pre>
perf record -g /my/application
perf report
</pre>
<p>
When run inside Linux perf, llvmpipe will create a /tmp/perf-XXXXX.map file with
symbol address table. It also dumps assembly code to /tmp/perf-XXXXX.map.asm,
which can be used by the bin/perf-annotate-jit.py script to produce disassembly of
the generated code annotated with the samples.
</p>
<p>You can obtain a call graph via
<a href="https://github.com/jrfonseca/gprof2dot#linux-perf">Gprof2Dot</a>.</p>
<h1>Unit testing</h1>
<p>
Building will also create several unit tests in
build/linux-???-debug/gallium/drivers/llvmpipe:
</p>
<ul>
<li> lp_test_blend: blending
<li> lp_test_conv: SIMD vector conversion
<li> lp_test_format: pixel unpacking/packing
</ul>
<p>
Some of these tests can output results and benchmarks to a tab-separated file
for later analysis, e.g.:
</p>
<pre>
build/linux-x86_64-debug/gallium/drivers/llvmpipe/lp_test_blend -o blend.tsv
</pre>
<h1>Development Notes</h1>
<ul>
<li>
When looking at this code for the first time, start in lp_state_fs.c, and
then skim through the lp_bld_* functions called there, and the comments
at the top of the lp_bld_*.c functions.
</li>
<li>
The driver-independent parts of the LLVM / Gallium code are found in
src/gallium/auxiliary/gallivm/. The filenames and function prefixes
need to be renamed from "lp_bld_" to something else though.
</li>
<li>
We use LLVM-C bindings for now. They are not documented, but follow the C++
interfaces very closely, and appear to be complete enough for code
generation. See
<a href="https://npcontemplation.blogspot.com/2008/06/secret-of-llvm-c-bindings.html">
this stand-alone example</a>. See the llvm-c/Core.h file for reference.
</li>
</ul>
<h1 id="recommended_reading">Recommended Reading</h1>
<ul>
<li>
<p>Rasterization</p>
<ul>
<li><a href="https://www.cs.unc.edu/~olano/papers/2dh-tri/">Triangle Scan Conversion using 2D Homogeneous Coordinates</a></li>
<li><a href="http://www.drdobbs.com/parallel/rasterization-on-larrabee/217200602">Rasterization on Larrabee</a> (<a href="http://devmaster.net/posts/2887/rasterization-on-larrabee">DevMaster copy</a>)</li>
<li><a href="http://devmaster.net/posts/6133/rasterization-using-half-space-functions">Rasterization using half-space functions</a></li>
<li><a href="http://devmaster.net/posts/6145/advanced-rasterization">Advanced Rasterization</a></li>
<li><a href="https://fgiesen.wordpress.com/2013/02/17/optimizing-sw-occlusion-culling-index/">Optimizing Software Occlusion Culling</a></li>
</ul>
</li>
<li>
<p>Texture sampling</p>
<ul>
<li><a href="http://chrishecker.com/Miscellaneous_Technical_Articles#Perspective_Texture_Mapping">Perspective Texture Mapping</a></li>
<li><a href="https://www.flipcode.com/archives/Texturing_As_In_Unreal.shtml">Texturing As In Unreal</a></li>
<li><a href="http://www.gamasutra.com/view/feature/3301/runtime_mipmap_filtering.php">Run-Time MIP-Map Filtering</a></li>
<li><a href="http://alt.3dcenter.org/artikel/2003/10-26_a_english.php">Will "brilinear" filtering persist?</a></li>
<li><a href="http://ixbtlabs.com/articles2/gffx/nv40-rx800-3.html">Trilinear filtering</a></li>
<li><a href="http://devmaster.net/posts/12785/texture-swizzling">Texture Swizzling</a></li>
</ul>
</li>
<li>
<p>SIMD</p>
<ul>
<li><a href="http://www.cdl.uni-saarland.de/projects/wfv/#header4">Whole-Function Vectorization</a></li>
</ul>
</li>
<li>
<p>Optimization</p>
<ul>
<li><a href="http://www.drdobbs.com/optimizing-pixomatic-for-modern-x86-proc/184405807">Optimizing Pixomatic For Modern x86 Processors</a></li>
<li><a href="http://www.intel.com/content/www/us/en/architecture-and-technology/64-ia-32-architectures-optimization-manual.html">Intel 64 and IA-32 Architectures Optimization Reference Manual</a></li>
<li><a href="http://www.agner.org/optimize/">Software optimization resources</a></li>
<li><a href="https://software.intel.com/en-us/articles/intel-intrinsics-guide">Intel Intrinsics Guide</a><li>
</ul>
</li>
<li>
<p>LLVM</p>
<ul>
<li><a href="http://llvm.org/docs/LangRef.html">LLVM Language Reference Manual</a></li>
<li><a href="https://npcontemplation.blogspot.co.uk/2008/06/secret-of-llvm-c-bindings.html">The secret of LLVM C bindings</a></li>
</ul>
</li>
<li>
<p>General</p>
<ul>
<li><a href="https://fgiesen.wordpress.com/2011/07/09/a-trip-through-the-graphics-pipeline-2011-index/">A trip through the Graphics Pipeline</a></li>
<li><a href="https://msdn.microsoft.com/en-us/library/gg615082.aspx#architecture">WARP Architecture and Performance</a></li>
</ul>
</li>
</ul>
</div>
</body>
</html>