Skip to content
Snippets Groups Projects
Commit db4d905d authored by Spencer Sherwin's avatar Spencer Sherwin
Browse files

Use BOOST_PP to generate switch statement in MatrixFreeOps to allow ability to secify range

parent d4a32a6a
No related branches found
No related tags found
No related merge requests found
Expand all Hide whitespace changes
Inline Side-by-side
Showing
with 394 additions and 1259 deletions
CHANGELOG.md
+ 2
− 1
View file @ db4d905d
...@@ -2,8 +2,9 @@ Changelog ...@@ -2,8 +2,9 @@ Changelog
========= =========
v5.7.0 v5.7.0
------ -----
**Library** **Library**
- Modified MatrixFreeOp library switch initialisation to use BOOST_PP (!1794)
- Fix memory-leak with LowEnergyBlock preconditioner for time-updated matrices (!1627) - Fix memory-leak with LowEnergyBlock preconditioner for time-updated matrices (!1627)
- Fix Fourier expansion integration weights are related test (!1803) - Fix Fourier expansion integration weights are related test (!1803)
......
CMakeLists.txt
+ 7
− 0
View file @ db4d905d
...@@ -220,6 +220,13 @@ IF (${CMAKE_COMPILER_IS_GNUCXX}) ...@@ -220,6 +220,13 @@ IF (${CMAKE_COMPILER_IS_GNUCXX})
MARK_AS_ADVANCED(NEKTAR_ENABLE_PROFILE) MARK_AS_ADVANCED(NEKTAR_ENABLE_PROFILE)
ENDIF (${CMAKE_COMPILER_IS_GNUCXX}) ENDIF (${CMAKE_COMPILER_IS_GNUCXX})
# Automatic Switch generation range
SET(NEKTAR_SWITCH_MIN 2 CACHE STRING "Set min range of auto generated switch")
MARK_AS_ADVANCED(NEKTAR_SWITCH_MIN)
SET(NEKTAR_SWITCH_MAX 8 CACHE STRING "Set max range of auto generated switch")
MARK_AS_ADVANCED(NEKTAR_SWITCH_MAX)
# Memory pools # Memory pools
OPTION(NEKTAR_USE_MEMORY_POOLS OPTION(NEKTAR_USE_MEMORY_POOLS
"Use memory pools to accelerate memory allocation." ON) "Use memory pools to accelerate memory allocation." ON)
......
library/MatrixFreeOps/CMakeLists.txt
+ 5
− 0
View file @ db4d905d
...@@ -6,6 +6,11 @@ SET(MATRIXFREEOPS_SOURCES ...@@ -6,6 +6,11 @@ SET(MATRIXFREEOPS_SOURCES
Operator.cpp Operator.cpp
) )
CONFIGURE_FILE(${CMAKE_CURRENT_SOURCE_DIR}/SwitchLimits.h.in
${CMAKE_CURRENT_SOURCE_DIR}/SwitchLimits.h
@ONLY)
# Boilerplate cpp implementation files for registering an operator # Boilerplate cpp implementation files for registering an operator
# with the factory. There are individual files for each operator shape # with the factory. There are individual files for each operator shape
# so to reduce compile time timeout failures due to the large number # so to reduce compile time timeout failures due to the large number
......
library/MatrixFreeOps/SwitchLimits.h.in 0 → 100644
+ 42
− 0
View file @ db4d905d
///////////////////////////////////////////////////////////////////////////////
//
// File: SwitchLimits.h
//
// For more information, please see: http://www.nektar.info
//
// The MIT License
//
// Copyright (c) 2006 Division of Applied Mathematics, Brown University (USA),
// Department of Aeronautics, Imperial College London (UK), and Scientific
// Computing and Imaging Institute, University of Utah (USA).
//
// Permission is hereby granted, free of charge, to any person obtaining a
// copy of this software and associated documentation files (the "Software"),
// to deal in the Software without restriction, including without limitation
// the rights to use, copy, modify, merge, publish, distribute, sublicense,
// and/or sell copies of the Software, and to permit persons to whom the
// Software is furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included
// in all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
// OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
// THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
// FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
// DEALINGS IN THE SOFTWARE.
//
// Description: Set the wwitch limits
//
///////////////////////////////////////////////////////////////////////////////
#define MIN1D @NEKTAR_SWITCH_MIN@
#define MAX1D @NEKTAR_SWITCH_MAX@
#define MIN2D @NEKTAR_SWITCH_MIN@
#define MAX2D @NEKTAR_SWITCH_MAX@
#define MIN3D @NEKTAR_SWITCH_MIN@
#define MAX3D @NEKTAR_SWITCH_MAX@
library/MatrixFreeOps/SwitchNodesPoints.h
+ 209
− 1040
View file @ db4d905d
This diff is collapsed.
library/MatrixFreeOps/SwitchPoints.h
+ 126
− 215
View file @ db4d905d
...@@ -41,52 +41,77 @@ ...@@ -41,52 +41,77 @@
// but becasue the operator{1,23}D is both a function and template that // but becasue the operator{1,23}D is both a function and template that
// need to be in the inherited class it is not possible. // need to be in the inherited class it is not possible.
#include <boost/preprocessor/arithmetic/inc.hpp>
#include <boost/preprocessor/comparison/not_equal.hpp>
#include <boost/preprocessor/repetition/for.hpp>
#include <boost/preprocessor/tuple/elem.hpp>
#include "SwitchLimits.h"
/* Macro compares the values of the tuple 'state' to see if the first
element, given by BOOST_PP_TUPLE_ELEM(2, 0, state), is not equal to
the second element plus one, given by
BOOST_PP_INC(BOOST_PP_TUPLE_ELEM(2, 1, state)) and returns 1 if not
equal otherwise zero */
#define TEST(r, state) \
BOOST_PP_NOT_EQUAL(BOOST_PP_TUPLE_ELEM(2, 0, state), \
BOOST_PP_INC(BOOST_PP_TUPLE_ELEM(2, 1, state)))
/* Macro returns a tuple where the first element, given by
BOOST_PP_TUPLE_ELEM(2, 0, state), is incremented by one and the
second element, given by BOOST_PP_TUPLE_ELEM(2, 1, state) remains
the same*/
#define UPDATE(r, state) \
(BOOST_PP_INC(BOOST_PP_TUPLE_ELEM(2, 0, state)), \
BOOST_PP_TUPLE_ELEM(2, 1, state))
#if defined(SHAPE_DIMENSION_1D) #if defined(SHAPE_DIMENSION_1D)
/* Define the switch statement for 1D shape operators and min, max range */
#define OPERATOR1D(r, state) \
case BOOST_PP_TUPLE_ELEM(2, 0, state): \
operator1D<BOOST_PP_TUPLE_ELEM(2, 0, state)>(input, output); \
break;
/* start of switch include for operator1D */
{ {
const int nq0 = m_basis[0]->GetNumPoints(); const int nq0 = m_basis[0]->GetNumPoints();
#if defined(SHAPE_TYPE_SEG) #if defined(SHAPE_TYPE_SEG)
switch (nq0) switch (nq0)
{ {
case 2: /*
operator1D<2>(input, output); expand OPERATOR1D case from 2 to 10, i.e.
break; ..
case 3: case 4:
operator1D<3>(input, output); operator1D<4>(input,output);
break; break;
case 4: ...
operator1D<4>(input, output); */
break; BOOST_PP_FOR((MIN1D, MAX1D), TEST, UPDATE, OPERATOR1D)
case 5:
operator1D<5>(input, output);
break;
case 6:
operator1D<6>(input, output);
break;
case 7:
operator1D<7>(input, output);
break;
case 8:
operator1D<8>(input, output);
break;
case 9:
operator1D<9>(input, output);
break;
case 10:
operator1D<10>(input, output);
break;
default: default:
operator1D(input, output); operator1D(input, output);
break; break;
} }
#endif #endif
} }
#elif defined(SHAPE_DIMENSION_2D) #elif defined(SHAPE_DIMENSION_2D)
/* Define the switch statement for 2D shape operators and min, max range */
#define OPERATOR2D_TRI(r, state) \
case BOOST_PP_TUPLE_ELEM(2, 0, state): \
operator2D<BOOST_PP_TUPLE_ELEM(2, 0, state), \
BOOST_PP_DEC(BOOST_PP_TUPLE_ELEM(2, 0, state))>(input, \
output); \
break;
#define OPERATOR2D_QUAD(r, state) \
case BOOST_PP_TUPLE_ELEM(2, 0, state): \
operator2D<BOOST_PP_TUPLE_ELEM(2, 0, state), \
BOOST_PP_TUPLE_ELEM(2, 0, state)>(input, output); \
break;
/* start of switch include for operator2D */
{ {
const int nq0 = m_basis[0]->GetNumPoints(); const int nq0 = m_basis[0]->GetNumPoints();
const int nq1 = m_basis[1]->GetNumPoints(); const int nq1 = m_basis[1]->GetNumPoints();
...@@ -97,33 +122,16 @@ ...@@ -97,33 +122,16 @@
{ {
switch (m_basis[0]->GetNumPoints()) switch (m_basis[0]->GetNumPoints())
{ {
case 2: /*
operator2D<2, 1>(input, output); expand OPERATOR2D case from 2 to 10, i.e.
break; ..
case 3: case 4:
operator2D<3, 2>(input, output); operator2D<4,3>(input,output);
break; break;
case 4: ...
operator2D<4, 3>(input, output); */
break; BOOST_PP_FOR((MIN2D, MAX2D), TEST, UPDATE, OPERATOR2D_TRI);
case 5:
operator2D<5, 4>(input, output);
break;
case 6:
operator2D<6, 5>(input, output);
break;
case 7:
operator2D<7, 6>(input, output);
break;
case 8:
operator2D<8, 7>(input, output);
break;
case 9:
operator2D<9, 8>(input, output);
break;
case 10:
operator2D<10, 9>(input, output);
break;
default: default:
operator2D(input, output); operator2D(input, output);
break; break;
...@@ -136,33 +144,15 @@ ...@@ -136,33 +144,15 @@
{ {
switch (m_basis[0]->GetNumPoints()) switch (m_basis[0]->GetNumPoints())
{ {
case 2: /*
operator2D<2, 2>(input, output); expand OPERATOR2D case from 2 to 10, i.e.
break; ..
case 3: case 4:
operator2D<3, 3>(input, output); operator2D<4,4>(input,output);
break; break;
case 4: ...
operator2D<4, 4>(input, output); */
break; BOOST_PP_FOR((MIN2D, MAX2D), TEST, UPDATE, OPERATOR2D_QUAD)
case 5:
operator2D<5, 5>(input, output);
break;
case 6:
operator2D<6, 6>(input, output);
break;
case 7:
operator2D<7, 7>(input, output);
break;
case 8:
operator2D<8, 8>(input, output);
break;
case 9:
operator2D<9, 9>(input, output);
break;
case 10:
operator2D<10, 10>(input, output);
break;
default: default:
operator2D(input, output); operator2D(input, output);
break; break;
...@@ -179,6 +169,31 @@ ...@@ -179,6 +169,31 @@
#elif defined(SHAPE_DIMENSION_3D) #elif defined(SHAPE_DIMENSION_3D)
/* Define the switch statement for 3D shape operators and min, max range */
#define OPERATOR3D_HEX(r, state) \
case BOOST_PP_TUPLE_ELEM(2, 0, state): \
operator3D<BOOST_PP_TUPLE_ELEM(2, 0, state), \
BOOST_PP_TUPLE_ELEM(2, 0, state), \
BOOST_PP_TUPLE_ELEM(2, 0, state)>(input, output); \
break;
#define OPERATOR3D_TET(r, state) \
case BOOST_PP_TUPLE_ELEM(2, 0, state): \
operator3D<BOOST_PP_TUPLE_ELEM(2, 0, state), \
BOOST_PP_DEC(BOOST_PP_TUPLE_ELEM(2, 0, state)), \
BOOST_PP_DEC(BOOST_PP_TUPLE_ELEM(2, 0, state))>(input, \
output); \
break;
#define OPERATOR3D_PYR_PRISM(r, state) \
case BOOST_PP_TUPLE_ELEM(2, 0, state): \
operator3D<BOOST_PP_TUPLE_ELEM(2, 0, state), \
BOOST_PP_TUPLE_ELEM(2, 0, state), \
BOOST_PP_DEC(BOOST_PP_TUPLE_ELEM(2, 0, state))>(input, \
output); \
break;
/* start of include switch details for operator3D */
{ {
const int nq0 = m_basis[0]->GetNumPoints(); const int nq0 = m_basis[0]->GetNumPoints();
const int nq1 = m_basis[1]->GetNumPoints(); const int nq1 = m_basis[1]->GetNumPoints();
...@@ -190,51 +205,16 @@ ...@@ -190,51 +205,16 @@
{ {
switch (m_basis[0]->GetNumPoints()) switch (m_basis[0]->GetNumPoints())
{ {
case 2: /*
operator3D<2, 2, 2>(input, output); expand OPERATOR3D case from 2 to 16, i.e.
break; ..
case 3: case 4:
operator3D<3, 3, 3>(input, output); operator3D<4,4,4>(input,output);
break; break;
case 4: ...
operator3D<4, 4, 4>(input, output); */
break; BOOST_PP_FOR((MIN3D, MAX3D), TEST, UPDATE, OPERATOR3D_HEX)
case 5:
operator3D<5, 5, 5>(input, output);
break;
case 6:
operator3D<6, 6, 6>(input, output);
break;
case 7:
operator3D<7, 7, 7>(input, output);
break;
case 8:
operator3D<8, 8, 8>(input, output);
break;
case 9:
operator3D<9, 9, 9>(input, output);
break;
case 10:
operator3D<10, 10, 10>(input, output);
break;
case 11:
operator3D<11, 11, 11>(input, output);
break;
case 12:
operator3D<12, 12, 12>(input, output);
break;
case 13:
operator3D<13, 13, 13>(input, output);
break;
case 14:
operator3D<14, 14, 14>(input, output);
break;
case 15:
operator3D<15, 15, 15>(input, output);
break;
case 16:
operator3D<16, 16, 16>(input, output);
break;
default: default:
operator3D(input, output); operator3D(input, output);
break; break;
...@@ -247,48 +227,15 @@ ...@@ -247,48 +227,15 @@
{ {
switch (nq0) switch (nq0)
{ {
case 3: /*
operator3D<3, 2, 2>(input, output); expand OPERATOR3D case from 2 to 16, i.e.
break; ..
case 4: case 4:
operator3D<4, 3, 3>(input, output); operator3D<4,3,3>(input,output);
break; break;
case 5: ...
operator3D<5, 4, 4>(input, output); */
break; BOOST_PP_FOR((MIN3D, MAX3D), TEST, UPDATE, OPERATOR3D_TET)
case 6:
operator3D<6, 5, 5>(input, output);
break;
case 7:
operator3D<7, 6, 6>(input, output);
break;
case 8:
operator3D<8, 7, 7>(input, output);
break;
case 9:
operator3D<9, 8, 8>(input, output);
break;
case 10:
operator3D<10, 9, 9>(input, output);
break;
case 11:
operator3D<11, 10, 10>(input, output);
break;
case 12:
operator3D<12, 11, 11>(input, output);
break;
case 13:
operator3D<13, 12, 12>(input, output);
break;
case 14:
operator3D<14, 13, 13>(input, output);
break;
case 15:
operator3D<15, 14, 14>(input, output);
break;
case 16:
operator3D<16, 15, 15>(input, output);
break;
default: default:
operator3D(input, output); operator3D(input, output);
break; break;
...@@ -301,51 +248,15 @@ ...@@ -301,51 +248,15 @@
{ {
switch (m_basis[0]->GetNumPoints()) switch (m_basis[0]->GetNumPoints())
{ {
case 2: /*
operator3D<2, 2, 1>(input, output); expand OPERATOR3D case from 3 to 16, i.e.
break; ..
case 3: case 4:
operator3D<3, 3, 2>(input, output); operator3D<4,3,3>(input,output);
break; break;
case 4: ...
operator3D<4, 4, 3>(input, output); */
break; BOOST_PP_FOR((MIN3D, MAX3D), TEST, UPDATE, OPERATOR3D_PYR_PRISM)
case 5:
operator3D<5, 5, 4>(input, output);
break;
case 6:
operator3D<6, 6, 5>(input, output);
break;
case 7:
operator3D<7, 7, 6>(input, output);
break;
case 8:
operator3D<8, 8, 7>(input, output);
break;
case 9:
operator3D<9, 9, 8>(input, output);
break;
case 10:
operator3D<10, 10, 9>(input, output);
break;
case 11:
operator3D<11, 11, 10>(input, output);
break;
case 12:
operator3D<12, 12, 11>(input, output);
break;
case 13:
operator3D<13, 13, 12>(input, output);
break;
case 14:
operator3D<14, 14, 13>(input, output);
break;
case 15:
operator3D<15, 15, 14>(input, output);
break;
case 16:
operator3D<16, 16, 15>(input, output);
break;
default: default:
operator3D(input, output); operator3D(input, output);
break; break;
......
solvers/CompressibleFlowSolver/Tests/Perf_IsentropicVortex16Implicit_P8.tst
+ 1
− 1
View file @ db4d905d
...@@ -20,7 +20,7 @@ ...@@ -20,7 +20,7 @@
<value variable="E" tolerance="1e-12">3.03058e-05</value> <value variable="E" tolerance="1e-12">3.03058e-05</value>
</metric> </metric>
<metric type="ExecutionTime" id="3"> <metric type="ExecutionTime" id="3">
<value tolerance="1e0" hostname="42.debian-bullseye-performance-build-and-test">36.0</value> <value tolerance="1e0" hostname="42.debian-bullseye-performance-build-and-test">34.0</value>
</metric> </metric>
</metrics> </metrics>
</test> </test>
solvers/IncNavierStokesSolver/Tests/Perf_ChanFlow_3DH1D_pRef.tst
+ 1
− 1
View file @ db4d905d
...@@ -20,7 +20,7 @@ ...@@ -20,7 +20,7 @@
<value variable="p" tolerance="1e-7">0.0671074</value> <value variable="p" tolerance="1e-7">0.0671074</value>
</metric> </metric>
<metric type="ExecutionTime" id="3"> <metric type="ExecutionTime" id="3">
<value tolerance="8e-1" hostname="42.debian-bullseye-performance-build-and-test">13.758</value> <value tolerance="8e-1" hostname="42.debian-bullseye-performance-build-and-test">13.3</value>
</metric> </metric>
</metrics> </metrics>
</test> </test>
solvers/PulseWaveSolver/Tests/Perf_TwoBifurcations.tst
+ 1
− 1
View file @ db4d905d
...@@ -16,7 +16,7 @@ ...@@ -16,7 +16,7 @@
<value variable="u" tolerance="1e-12">6.05588</value> <value variable="u" tolerance="1e-12">6.05588</value>
</metric> </metric>
<metric type="ExecutionTime" id="3"> <metric type="ExecutionTime" id="3">
<value tolerance="3e-1" hostname="42.debian-bullseye-performance-build-and-test">6.05743</value> <value tolerance="6e-1" hostname="42.debian-bullseye-performance-build-and-test">6.05743</value>
</metric> </metric>
</metrics> </metrics>
</test> </test>
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment