1 <!DOCTYPE HTML PUBLIC
"-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
4 <meta http-equiv=
"content-type" content=
"text/html; charset=utf-8">
5 <title>Gallium LLVMpipe Driver
</title>
6 <link rel=
"stylesheet" type=
"text/css" href=
"mesa.css">
11 The Mesa
3D Graphics Library
14 <iframe src=
"contents.html"></iframe>
17 <h1>Gallium LLVMpipe Driver
</h1>
22 The Gallium llvmpipe driver is a software rasterizer that uses LLVM to
23 do runtime code generation.
24 Shaders, point/line/triangle rasterization and vertex processing are
25 implemented with LLVM IR which is translated to x86, x86-
64, or ppc64le machine
27 Also, the driver is multithreaded to take advantage of multiple CPU cores
28 (up to
8 at this time).
29 It's the fastest software rasterizer for Mesa.
38 For x86 or amd64 processors,
64-bit mode is recommended.
39 Support for SSE2 is strongly encouraged. Support for SSE3 and SSE4.1 will
40 yield the most efficient code. The fewer features the CPU has the more
41 likely it is that you will run into underperforming, buggy, or incomplete code.
44 For ppc64le processors, use of the Altivec feature (the Vector
45 Facility) is recommended if supported; use of the VSX feature (the
46 Vector-Scalar Facility) is recommended if supported AND Mesa is
47 built with LLVM version
4.0 or later.
50 See /proc/cpuinfo to know what your CPU supports.
54 <p>Unless otherwise stated, LLVM version
3.4 is recommended;
3.3 or later is required.
</p>
56 For Linux, on a recent Debian based distribution do:
59 aptitude install llvm-dev
62 If you want development snapshot builds of LLVM for Debian and derived
63 distributions like Ubuntu, you can use the APT repository at
<a
64 href=
"https://apt.llvm.org/" title=
"Debian Development packages for LLVM"
65 >apt.llvm.org
</a>, which are maintained by Debian's LLVM maintainer.
68 For a RPM-based distribution do:
71 yum install llvm-devel
75 For Windows you will need to build LLVM from source with MSVC or MINGW
76 (either natively or through cross compilers) and CMake, and set the LLVM
77 environment variable to the directory you installed it to.
79 LLVM will be statically linked, so when building on MSVC it needs to be
80 built with a matching CRT as Mesa, and you'll need to pass
81 <code>-DLLVM_USE_CRT_xxx=yyy
</code> as described below.
86 <th rowspan=
"2">LLVM build-type
</th>
87 <th colspan=
"2" align=
"center">Mesa build-type
</th>
90 <th>debug,checked
</th>
91 <th>release,profile
</th>
95 <td><code>-DLLVM_USE_CRT_DEBUG=MTd
</code></td>
96 <td><code>-DLLVM_USE_CRT_DEBUG=MT
</code></td>
100 <td><code>-DLLVM_USE_CRT_RELEASE=MTd
</code></td>
101 <td><code>-DLLVM_USE_CRT_RELEASE=MT
</code></td>
106 You can build only the x86 target by passing -DLLVM_TARGETS_TO_BUILD=X86
112 <p>scons (optional)
</p>
119 To build everything on Linux invoke scons as:
122 scons build=debug libgl-xlib
125 Alternatively, you can build it with meson with:
129 meson -D glx=gallium-xlib -D gallium-drivers=swrast
133 but the rest of these instructions assume that scons is used.
135 For Windows the procedure is similar except the target:
138 scons platform=windows build=debug libgl-gdi
146 <p>On Linux, building will create a drop-in alternative for libGL.so into
</p>
149 build/foo/gallium/targets/libgl-xlib/libGL.so
156 <p>To use it set the LD_LIBRARY_PATH environment variable accordingly.
</p>
158 <p>For performance evaluation pass build=release to scons, and use the corresponding
159 lib directory without the
"-debug" suffix.
</p>
165 On Windows, building will create
166 <code>build/windows-x86-debug/gallium/targets/libgl-gdi/opengl32.dll
</code>
167 which is a drop-in alternative for system's
<code>opengl32.dll
</code>. To use
168 it put it in the same directory as your application. It can also be used by
169 replacing the native ICD driver, but it's quite an advanced usage, so if you
170 need to ask, don't even try it.
174 There is however an easy way to replace the OpenGL software renderer that comes
175 with Microsoft Windows
7 (or later) with llvmpipe (that is, on systems without
180 <li><p>copy build/windows-x86-debug/gallium/targets/libgl-gdi/opengl32.dll to C:\Windows\SysWOW64\mesadrv.dll
</p></li>
181 <li><p>load this registry settings:
</p>
184 ; https://technet.microsoft.com/en-us/library/cc749368.aspx
185 ; https://www.msfn.org/board/topic/
143241-portable-windows-
7-build-from-winpe-
30/page-
5#entry942596
186 [HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\Microsoft\Windows NT\CurrentVersion\OpenGLDrivers\MSOGL]
188 "DriverVersion"=dword:
00000001
189 "Flags"=dword:
00000001
190 "Version"=dword:
00000002
193 <li>Ditto for
64 bits drivers if you need them.
</li>
200 To profile llvmpipe you should build as
203 scons build=profile
<same-as-before
>
207 This will ensure that frame pointers are used both in C and JIT functions, and
208 that no tail call optimizations are done by gcc.
211 <h3>Linux perf integration
</h3>
214 On Linux, it is possible to have symbol resolution of JIT code with
<a href=
"https://perf.wiki.kernel.org/">Linux perf
</a>:
218 perf record -g /my/application
223 When run inside Linux perf, llvmpipe will create a /tmp/perf-XXXXX.map file with
224 symbol address table. It also dumps assembly code to /tmp/perf-XXXXX.map.asm,
225 which can be used by the bin/perf-annotate-jit.py script to produce disassembly of
226 the generated code annotated with the samples.
229 <p>You can obtain a call graph via
230 <a href=
"https://github.com/jrfonseca/gprof2dot#linux-perf">Gprof2Dot
</a>.
</p>
233 <h2>Unit testing
</h2>
236 Building will also create several unit tests in
237 build/linux-???-debug/gallium/drivers/llvmpipe:
241 <li> lp_test_blend: blending
242 <li> lp_test_conv: SIMD vector conversion
243 <li> lp_test_format: pixel unpacking/packing
247 Some of these tests can output results and benchmarks to a tab-separated file
248 for later analysis, e.g.:
251 build/linux-x86_64-debug/gallium/drivers/llvmpipe/lp_test_blend -o blend.tsv
255 <h2>Development Notes
</h2>
259 When looking at this code for the first time, start in lp_state_fs.c, and
260 then skim through the lp_bld_* functions called there, and the comments
261 at the top of the lp_bld_*.c functions.
264 The driver-independent parts of the LLVM / Gallium code are found in
265 src/gallium/auxiliary/gallivm/. The filenames and function prefixes
266 need to be renamed from
"lp_bld_" to something else though.
269 We use LLVM-C bindings for now. They are not documented, but follow the C++
270 interfaces very closely, and appear to be complete enough for code
272 <a href=
"https://npcontemplation.blogspot.com/2008/06/secret-of-llvm-c-bindings.html">
273 this stand-alone example
</a>. See the llvm-c/Core.h file for reference.
277 <h2 id=
"recommended_reading">Recommended Reading
</h2>
283 <li><a href=
"https://www.cs.unc.edu/~olano/papers/2dh-tri/">Triangle Scan Conversion using
2D Homogeneous Coordinates
</a></li>
284 <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>
285 <li><a href=
"http://devmaster.net/posts/6133/rasterization-using-half-space-functions">Rasterization using half-space functions
</a></li>
286 <li><a href=
"http://devmaster.net/posts/6145/advanced-rasterization">Advanced Rasterization
</a></li>
287 <li><a href=
"https://fgiesen.wordpress.com/2013/02/17/optimizing-sw-occlusion-culling-index/">Optimizing Software Occlusion Culling
</a></li>
291 <p>Texture sampling
</p>
293 <li><a href=
"http://chrishecker.com/Miscellaneous_Technical_Articles#Perspective_Texture_Mapping">Perspective Texture Mapping
</a></li>
294 <li><a href=
"https://www.flipcode.com/archives/Texturing_As_In_Unreal.shtml">Texturing As In Unreal
</a></li>
295 <li><a href=
"http://www.gamasutra.com/view/feature/3301/runtime_mipmap_filtering.php">Run-Time MIP-Map Filtering
</a></li>
296 <li><a href=
"http://alt.3dcenter.org/artikel/2003/10-26_a_english.php">Will
"brilinear" filtering persist?
</a></li>
297 <li><a href=
"http://ixbtlabs.com/articles2/gffx/nv40-rx800-3.html">Trilinear filtering
</a></li>
298 <li><a href=
"http://devmaster.net/posts/12785/texture-swizzling">Texture Swizzling
</a></li>
304 <li><a href=
"http://www.cdl.uni-saarland.de/projects/wfv/#header4">Whole-Function Vectorization
</a></li>
310 <li><a href=
"http://www.drdobbs.com/optimizing-pixomatic-for-modern-x86-proc/184405807">Optimizing Pixomatic For Modern x86 Processors
</a></li>
311 <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>
312 <li><a href=
"http://www.agner.org/optimize/">Software optimization resources
</a></li>
313 <li><a href=
"https://software.intel.com/en-us/articles/intel-intrinsics-guide">Intel Intrinsics Guide
</a></li>
319 <li><a href=
"http://llvm.org/docs/LangRef.html">LLVM Language Reference Manual
</a></li>
320 <li><a href=
"https://npcontemplation.blogspot.co.uk/2008/06/secret-of-llvm-c-bindings.html">The secret of LLVM C bindings
</a></li>
326 <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>
327 <li><a href=
"https://msdn.microsoft.com/en-us/library/gg615082.aspx#architecture">WARP Architecture and Performance
</a></li>