fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-08 09:08:10 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

llvmpipe: use ppc64le/ppc64 Large code model for JIT-compiled shaders

Browse source 
Large programs, e.g. gnome-shell and firefox, may tax the
addressability of the Medium code model once a (potentially unbounded)
number of dynamically generated JIT-compiled shader programs are
linked in and relocated.  Yet the default code model as of LLVM 8 is
Medium or even Small.

The cost of changing from Medium to Large is negligible:
- an additional 8-byte pointer stored immediately before the shader entrypoint;
- change an add-immediate (addis) instruction to a load (ld).

Testing with WebGL Conformance
(https://www.khronos.org/registry/webgl/sdk/tests/webgl-conformance-tests.html)
yields clean runs with this change (and crashes without it).

Testing with glxgears shows no detectable performance difference.

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1753327, 1753789, 1543572, 1747110, and 1582226

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/223

Co-authored by: Nemanja Ivanovic <nemanjai@ca.ibm.com>, Tom Stellard <tstellar@redhat.com>

CC: mesa-stable@lists.freedesktop.org

Signed-off-by: Ben Crocker <bcrocker@redhat.com>
(cherry picked from commit 9c3be6d21f)
Conflicts resolved Dylan (PIPE_ARCH -> UTIL_ARCH rename)
This commit is contained in:
Ben Crocker 2019-11-13 20:27:24 +00:00 • committed by Dylan Baker
parent 35182247fc
commit 32aba91c07
1 changed files with 15 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
src/gallium/auxiliary/gallivm/lp_bld_misc.cpp
View file

@ -456,7 +456,20 @@ lp_build_create_jit_compiler_for_module(LLVMExecutionEngineRef *OutJIT,
* when not using MCJIT so no instructions are generated which the old JIT * when not using MCJIT so no instructions are generated which the old JIT
* can't handle. Not entirely sure if we really need to do anything yet. * can't handle. Not entirely sure if we really need to do anything yet.
*/ */
#if defined(PIPE_ARCH_LITTLE_ENDIAN) && defined(PIPE_ARCH_PPC_64)
#ifdef PIPE_ARCH_PPC_64
/*
* Large programs, e.g. gnome-shell and firefox, may tax the addressability
* of the Medium code model once dynamically generated JIT-compiled shader
* programs are linked in and relocated. Yet the default code model as of
* LLVM 8 is Medium or even Small.
* The cost of changing from Medium to Large is negligible:
* - an additional 8-byte pointer stored immediately before the shader entrypoint;
* - change an add-immediate (addis) instruction to a load (ld).
*/
builder.setCodeModel(CodeModel::Large);
#if PIPE_ARCH_LITTLE_ENDIAN
/* /*
* Versions of LLVM prior to 4.0 lacked a table entry for "POWER8NVL", * Versions of LLVM prior to 4.0 lacked a table entry for "POWER8NVL",
* resulting in (big-endian) "generic" being returned on * resulting in (big-endian) "generic" being returned on
@ -468,6 +481,7 @@ lp_build_create_jit_compiler_for_module(LLVMExecutionEngineRef *OutJIT,
*/ */
if (MCPU == "generic") if (MCPU == "generic")
MCPU = "pwr8"; MCPU = "pwr8";
#endif
#endif #endif
builder.setMCPU(MCPU); builder.setMCPU(MCPU);
if (gallivm_debug & (GALLIVM_DEBUG_IR | GALLIVM_DEBUG_ASM | GALLIVM_DEBUG_DUMP_BC)) { if (gallivm_debug & (GALLIVM_DEBUG_IR | GALLIVM_DEBUG_ASM | GALLIVM_DEBUG_DUMP_BC)) {