Cleanup. Remove copyright year. Reformat.

.clang-format

@@ -0,0 +1,176 @@
+---
+Language:        Cpp
+# BasedOnStyle:  WebKit
+AccessModifierOffset: -4
+AlignAfterOpenBracket: DontAlign
+AlignArrayOfStructures: None
+AlignConsecutiveAssignments: Consecutive
+AlignConsecutiveBitFields: None
+AlignConsecutiveDeclarations: None
+AlignConsecutiveMacros: Consecutive
+AlignEscapedNewlines: Right
+AlignOperands: Align
+AlignTrailingComments: true
+AllowAllArgumentsOnNextLine: false
+AllowAllParametersOfDeclarationOnNextLine: true
+AllowShortEnumsOnASingleLine: true
+AllowShortBlocksOnASingleLine: Never
+AllowShortCaseLabelsOnASingleLine: false
+AllowShortFunctionsOnASingleLine: All
+AllowShortLambdasOnASingleLine: All
+AllowShortIfStatementsOnASingleLine: OnlyFirstIf
+AllowShortLoopsOnASingleLine: false
+AlwaysBreakAfterDefinitionReturnType: None
+AlwaysBreakAfterReturnType: None
+AlwaysBreakBeforeMultilineStrings: false
+AlwaysBreakTemplateDeclarations: MultiLine
+AttributeMacros:
+  - __capability
+BinPackArguments: false
+BinPackParameters: false
+BraceWrapping:
+  AfterCaseLabel:  false
+  AfterClass:      false
+  AfterControlStatement: Never
+  AfterEnum:       false
+  AfterFunction:   true
+  AfterNamespace:  false
+  AfterObjCDeclaration: false
+  AfterStruct:     false
+  AfterUnion:      false
+  AfterExternBlock: false
+  BeforeCatch:     false
+  BeforeElse:      false
+  BeforeLambdaBody: false
+  BeforeWhile:     false
+  IndentBraces:    false
+  SplitEmptyFunction: true
+  SplitEmptyRecord: true
+  SplitEmptyNamespace: true
+BreakBeforeBinaryOperators: None
+BreakBeforeBraces: WebKit
+BreakBeforeInheritanceComma: false
+BreakInheritanceList: BeforeColon
+BreakBeforeTernaryOperators: true
+BreakConstructorInitializersBeforeComma: false
+BreakConstructorInitializers: BeforeComma
+BreakAfterJavaFieldAnnotations: false
+BreakStringLiterals: true
+ColumnLimit:     90
+CommentPragmas:  '^ IWYU pragma:'
+CompactNamespaces: false
+ConstructorInitializerIndentWidth: 4
+ContinuationIndentWidth: 4
+Cpp11BracedListStyle: false
+DeriveLineEnding: true
+DerivePointerAlignment: false
+DisableFormat:   false
+EmptyLineAfterAccessModifier: Never
+EmptyLineBeforeAccessModifier: LogicalBlock
+ExperimentalAutoDetectBinPacking: false
+BasedOnStyle:    ''
+ConstructorInitializerAllOnOneLineOrOnePerLine: false
+AllowAllConstructorInitializersOnNextLine: true
+FixNamespaceComments: false
+ForEachMacros:
+  - foreach
+  - Q_FOREACH
+  - BOOST_FOREACH
+IfMacros:
+  - KJ_IF_MAYBE
+IncludeBlocks:   Preserve
+IncludeCategories:
+  - Regex:           '^"(llvm|llvm-c|clang|clang-c)/'
+    Priority:        2
+    SortPriority:    0
+    CaseSensitive:   false
+  - Regex:           '^(<|"(gtest|gmock|isl|json)/)'
+    Priority:        3
+    SortPriority:    0
+    CaseSensitive:   false
+  - Regex:           '.*'
+    Priority:        1
+    SortPriority:    0
+    CaseSensitive:   false
+IncludeIsMainRegex: '(Test)?$'
+IncludeIsMainSourceRegex: ''
+IndentAccessModifiers: false
+IndentCaseLabels: false
+IndentCaseBlocks: false
+IndentGotoLabels: true
+IndentPPDirectives: None
+IndentExternBlock: AfterExternBlock
+IndentWidth:     4
+IndentWrappedFunctionNames: false
+InsertTrailingCommas: None
+JavaScriptQuotes: Leave
+JavaScriptWrapImports: true
+KeepEmptyLinesAtTheStartOfBlocks: true
+LambdaBodyIndentation: Signature
+MacroBlockBegin: ''
+MacroBlockEnd:   ''
+MaxEmptyLinesToKeep: 1
+NamespaceIndentation: Inner
+ObjCBinPackProtocolList: Auto
+ObjCBlockIndentWidth: 4
+ObjCBreakBeforeNestedBlockParam: true
+ObjCSpaceAfterProperty: true
+ObjCSpaceBeforeProtocolList: true
+PenaltyBreakAssignment: 200
+PenaltyBreakBeforeFirstCallParameter: 19
+PenaltyBreakComment: 300
+PenaltyBreakFirstLessLess: 120
+PenaltyBreakString: 1000
+PenaltyBreakTemplateDeclaration: 10
+PenaltyExcessCharacter: 1000000
+PenaltyReturnTypeOnItsOwnLine: 60
+PenaltyIndentedWhitespace: 0
+PointerAlignment: Left
+PPIndentWidth:   -1
+ReferenceAlignment: Pointer
+ReflowComments:  true
+ShortNamespaceLines: 1
+SortIncludes:    CaseSensitive
+SortJavaStaticImport: Before
+SortUsingDeclarations: true
+SpaceAfterCStyleCast: false
+SpaceAfterLogicalNot: false
+SpaceAfterTemplateKeyword: true
+SpaceBeforeAssignmentOperators: true
+SpaceBeforeCaseColon: false
+SpaceBeforeCpp11BracedList: true
+SpaceBeforeCtorInitializerColon: true
+SpaceBeforeInheritanceColon: true
+SpaceBeforeParens: ControlStatements
+SpaceAroundPointerQualifiers: Default
+SpaceBeforeRangeBasedForLoopColon: true
+SpaceInEmptyBlock: false
+SpaceInEmptyParentheses: false
+SpacesBeforeTrailingComments: 1
+SpacesInAngles:  Never
+SpacesInConditionalStatement: false
+SpacesInContainerLiterals: true
+SpacesInCStyleCastParentheses: false
+SpacesInLineCommentPrefix:
+  Minimum:         1
+  Maximum:         -1
+SpacesInParentheses: false
+SpacesInSquareBrackets: false
+SpaceBeforeSquareBrackets: false
+BitFieldColonSpacing: Both
+Standard:        Latest
+StatementAttributeLikeMacros:
+  - Q_EMIT
+StatementMacros:
+  - Q_UNUSED
+  - QT_REQUIRE_VERSION
+TabWidth:        8
+UseCRLF:         false
+UseTab:          Never
+WhitespaceSensitiveMacros:
+  - STRINGIZE
+  - PP_STRINGIZE
+  - BOOST_PP_STRINGIZE
+  - NS_SWIFT_NAME
+  - CF_SWIFT_NAME
+...

.clang-tidy

@@ -0,0 +1,14 @@
+---
+Checks:          'clang-diagnostic-*,clang-analyzer-*,clang-bugprone-*,readability-identifier-naming'
+WarningsAsErrors: true
+HeaderFilterRegex: '.*'
+AnalyzeTemporaryDtors: false
+CheckOptions:
+  - key:             readability-identifier-naming.StructCase
+    value:           'CamelCase'
+  - key:             readability-identifier-naming.FunctionCase
+    value:           'camelBack'
+  - key:             readability-identifier-naming.VariableCase
+    value:           'camelBack'
+  - key:             readability-identifier-naming.GlobalConstantCase
+    value:           'UPPER_CASE'

.clangd

@@ -0,0 +1,3 @@
+CompileFlags:
+  Add: [-I/Users/jan/prg/MD-Bench/verletlist/includes, -I/Users/jan/prg/MD-Bench/common/includes, -DALIGNMENT=64]
+  Compiler: clang

Makefile

@@ -98,10 +98,6 @@ ifeq ($(strip $(ENABLE_OMP_SIMD)),true)
     DEFINES += -DENABLE_OMP_SIMD
 endif
 
-ifeq ($(strip $(USE_SIMD_KERNEL)),true)
-    DEFINES += -DUSE_SIMD_KERNEL
-endif
-
 VPATH     = $(SRC_DIR) $(ASM_DIR) $(CUDA_DIR)
 ASM       = $(patsubst $(SRC_DIR)/%.c, $(BUILD_DIR)/%.s,$(wildcard $(SRC_DIR)/*.c))
 OVERWRITE:= $(patsubst $(ASM_DIR)/%-new.s, $(BUILD_DIR)/%.o,$(wildcard $(ASM_DIR)/*-new.s))

clusterpair/atom.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/cuda/force_lj.cu

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/force_eam.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/force_lj.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/includes/atom.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/includes/integrate.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/includes/neighbor.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/includes/pbc.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/includes/stats.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/includes/tracing.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/includes/vtk.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/includes/xtc.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/main-stub.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/main.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/neighbor.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/pbc.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/stats.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/tracing.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/vtk.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

clusterpair/xtc.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/allocate.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/device.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/eam_utils.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/allocate.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/device.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/eam.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/likwid-marker.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/parameter.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/simd.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/simd/avx2_double.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/simd/avx2_float.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/simd/avx512_double.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/simd/avx512_float.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/simd/avx_double.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/simd/avx_float.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/thermo.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/timers.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/timing.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/includes/util.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.
@@ -7,6 +7,7 @@
 #ifndef __UTIL_H_
 #define __UTIL_H_
 
+#include <stdio.h>
 #ifndef MIN
 #define MIN(x, y) ((x) < (y) ? (x) : (y))
 #endif
@@ -41,6 +42,6 @@ extern int str2ff(const char *string);
 extern const char *ff2str(int ff);
 extern void readline(char *line, FILE *fp);
 extern void debug_printf(const char *format, ...);
-extern int get_cuda_num_threads();
+extern int get_cuda_num_threads(void);
 
 #endif

common/parameter.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/thermo.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/timing.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

common/util.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

config.mk

@@ -1,17 +1,17 @@
 # Compiler tag (GCC/CLANG/ICC/ICX/ONEAPI/NVCC)
-TAG ?= ICC
+TAG ?= CLANG
 # Instruction set (SSE/AVX/AVX_FMA/AVX2/AVX512)
-ISA ?= AVX512
+ISA ?= SSE
-# Optimization scheme (lammps/gromacs/clusters_per_bin)
+# Optimization scheme (verletlist/clusterpair/clusters_per_bin)
-OPT_SCHEME ?= lammps
+OPT_SCHEME ?= verletlist
 # Enable likwid (true or false)
-ENABLE_LIKWID ?= true
+ENABLE_LIKWID ?= false
 # SP or DP
 DATA_TYPE ?= DP
 # AOS or SOA
 DATA_LAYOUT ?= AOS
 # Assembly syntax to generate (ATT/INTEL)
-ASM_SYNTAX ?= ATT
+ASM_SYNTAX ?= INTEL
 # Debug
 DEBUG ?= false
 
@@ -28,7 +28,7 @@ COMPUTE_STATS ?= true
 
 # Configurations for lammps optimization scheme
 # Use omp simd pragma when running with half neighbor-lists
-ENABLE_OMP_SIMD ?= true
+ENABLE_OMP_SIMD ?= false
 # Use kernel with explicit SIMD intrinsics
 USE_SIMD_KERNEL ?= false
 

lammps/force_lj.c

@@ -1,278 +0,0 @@
-/*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
- * All rights reserved. This file is part of MD-Bench.
- * Use of this source code is governed by a LGPL-3.0
- * license that can be found in the LICENSE file.
- */
-#include <stdio.h>
-#include <stdlib.h>
-//---
-#include <atom.h>
-#include <likwid-marker.h>
-#include <neighbor.h>
-#include <parameter.h>
-#include <stats.h>
-#include <timing.h>
-
-#ifdef __SIMD_KERNEL__
-#include <simd.h>
-#endif
-
-double computeForceLJFullNeigh_plain_c(Parameter *param, Atom *atom, Neighbor *neighbor, Stats *stats) {
-    int Nlocal = atom->Nlocal;
-    int* neighs;
-    #ifndef EXPLICIT_TYPES
-    MD_FLOAT cutforcesq = param->cutforce * param->cutforce;
-    MD_FLOAT sigma6 = param->sigma6;
-    MD_FLOAT epsilon = param->epsilon;
-    #endif
-    const MD_FLOAT num1 = 1.0;
-    const MD_FLOAT num48 = 48.0;
-    const MD_FLOAT num05 = 0.5;
-
-    for(int i = 0; i < Nlocal; i++) {
-        atom_fx(i) = 0.0;
-        atom_fy(i) = 0.0;
-        atom_fz(i) = 0.0;
-    }
-    double S = getTimeStamp();
-
-    #pragma omp parallel
-    {
-    LIKWID_MARKER_START("force");
-
-    #pragma omp for schedule(runtime)
-    for(int i = 0; i < Nlocal; i++) {
-        neighs = &neighbor->neighbors[i * neighbor->maxneighs];
-        int numneighs = neighbor->numneigh[i];
-        MD_FLOAT xtmp = atom_x(i);
-        MD_FLOAT ytmp = atom_y(i);
-        MD_FLOAT ztmp = atom_z(i);
-        MD_FLOAT fix = 0;
-        MD_FLOAT fiy = 0;
-        MD_FLOAT fiz = 0;
-
-        #ifdef EXPLICIT_TYPES
-        const int type_i = atom->type[i];
-        #endif
-
-        for(int k = 0; k < numneighs; k++) {
-            int j = neighs[k];
-            MD_FLOAT delx = xtmp - atom_x(j);
-            MD_FLOAT dely = ytmp - atom_y(j);
-            MD_FLOAT delz = ztmp - atom_z(j);
-            MD_FLOAT rsq = delx * delx + dely * dely + delz * delz;
-
-            #ifdef EXPLICIT_TYPES
-            const int type_j = atom->type[j];
-            const int type_ij = type_i * atom->ntypes + type_j;
-            const MD_FLOAT cutforcesq = atom->cutforcesq[type_ij];
-            const MD_FLOAT sigma6 = atom->sigma6[type_ij];
-            const MD_FLOAT epsilon = atom->epsilon[type_ij];
-            #endif
-
-            if(rsq < cutforcesq) {
-                MD_FLOAT sr2 = num1 / rsq;
-                MD_FLOAT sr6 = sr2 * sr2 * sr2 * sigma6;
-                MD_FLOAT force = num48 * sr6 * (sr6 - num05) * sr2 * epsilon;
-                fix += delx * force;
-                fiy += dely * force;
-                fiz += delz * force;
-            #ifdef USE_REFERENCE_VERSION
-                addStat(stats->atoms_within_cutoff, 1);
-            } else {
-                addStat(stats->atoms_outside_cutoff, 1);
-            #endif
-            }
-        }
-
-        atom_fx(i) += fix;
-        atom_fy(i) += fiy;
-        atom_fz(i) += fiz;
-
-        #ifdef USE_REFERENCE_VERSION
-        if(numneighs % VECTOR_WIDTH > 0) {
-            addStat(stats->atoms_outside_cutoff, VECTOR_WIDTH - (numneighs % VECTOR_WIDTH));
-        }
-        #endif
-
-        addStat(stats->total_force_neighs, numneighs);
-        addStat(stats->total_force_iters, (numneighs + VECTOR_WIDTH - 1) / VECTOR_WIDTH);
-    }
-
-    LIKWID_MARKER_STOP("force");
-    }
-
-    double E = getTimeStamp();
-    return E-S;
-}
-
-double computeForceLJHalfNeigh(Parameter *param, Atom *atom, Neighbor *neighbor, Stats *stats) {
-    int Nlocal = atom->Nlocal;
-    int* neighs;
-    #ifndef EXPLICIT_TYPES
-    MD_FLOAT cutforcesq = param->cutforce * param->cutforce;
-    MD_FLOAT sigma6 = param->sigma6;
-    MD_FLOAT epsilon = param->epsilon;
-    #endif
-    const MD_FLOAT num1 = 1.0;
-    const MD_FLOAT num48 = 48.0;
-    const MD_FLOAT num05 = 0.5;
-
-    for(int i = 0; i < Nlocal; i++) {
-        atom_fx(i) = 0.0;
-        atom_fy(i) = 0.0;
-        atom_fz(i) = 0.0;
-    }
-
-    double S = getTimeStamp();
-
-    #pragma omp parallel
-    {
-    LIKWID_MARKER_START("forceLJ-halfneigh");
-
-    #pragma omp for schedule(runtime)
-    for(int i = 0; i < Nlocal; i++) {
-        neighs = &neighbor->neighbors[i * neighbor->maxneighs];
-        int numneighs = neighbor->numneigh[i];
-        MD_FLOAT xtmp = atom_x(i);
-        MD_FLOAT ytmp = atom_y(i);
-        MD_FLOAT ztmp = atom_z(i);
-        MD_FLOAT fix = 0;
-        MD_FLOAT fiy = 0;
-        MD_FLOAT fiz = 0;
-
-        #ifdef EXPLICIT_TYPES
-        const int type_i = atom->type[i];
-        #endif
-
-        // Pragma required to vectorize the inner loop
-        #ifdef ENABLE_OMP_SIMD
-        #pragma omp simd reduction(+: fix,fiy,fiz)
-        #endif
-        for(int k = 0; k < numneighs; k++) {
-            int j = neighs[k];
-            MD_FLOAT delx = xtmp - atom_x(j);
-            MD_FLOAT dely = ytmp - atom_y(j);
-            MD_FLOAT delz = ztmp - atom_z(j);
-            MD_FLOAT rsq = delx * delx + dely * dely + delz * delz;
-
-            #ifdef EXPLICIT_TYPES
-            const int type_j = atom->type[j];
-            const int type_ij = type_i * atom->ntypes + type_j;
-            const MD_FLOAT cutforcesq = atom->cutforcesq[type_ij];
-            const MD_FLOAT sigma6 = atom->sigma6[type_ij];
-            const MD_FLOAT epsilon = atom->epsilon[type_ij];
-            #endif
-
-            if(rsq < cutforcesq) {
-                MD_FLOAT sr2 = num1 / rsq;
-                MD_FLOAT sr6 = sr2 * sr2 * sr2 * sigma6;
-                MD_FLOAT force = num48 * sr6 * (sr6 - num05) * sr2 * epsilon;
-                fix += delx * force;
-                fiy += dely * force;
-                fiz += delz * force;
-
-                // We do not need to update forces for ghost atoms
-                if(j < Nlocal) {
-                    atom_fx(j) -= delx * force;
-                    atom_fy(j) -= dely * force;
-                    atom_fz(j) -= delz * force;
-                }
-            }
-        }
-
-        atom_fx(i) += fix;
-        atom_fy(i) += fiy;
-        atom_fz(i) += fiz;
-
-        addStat(stats->total_force_neighs, numneighs);
-        addStat(stats->total_force_iters, (numneighs + VECTOR_WIDTH - 1) / VECTOR_WIDTH);
-    }
-
-    LIKWID_MARKER_STOP("forceLJ-halfneigh");
-    }
-
-    double E = getTimeStamp();
-    return E-S;
-}
-
-double computeForceLJFullNeigh_simd(Parameter *param, Atom *atom, Neighbor *neighbor, Stats *stats) {
-    int Nlocal = atom->Nlocal;
-    int* neighs;
-    MD_FLOAT cutforcesq = param->cutforce * param->cutforce;
-    MD_FLOAT sigma6 = param->sigma6;
-    MD_FLOAT epsilon = param->epsilon;
-
-    for(int i = 0; i < Nlocal; i++) {
-        atom_fx(i) = 0.0;
-        atom_fy(i) = 0.0;
-        atom_fz(i) = 0.0;
-    }
-
-    double S = getTimeStamp();
-
-    #ifndef __SIMD_KERNEL__
-    fprintf(stderr, "Error: SIMD kernel not implemented for specified instruction set!");
-    exit(-1);
-    #else
-    MD_SIMD_FLOAT cutforcesq_vec = simd_broadcast(cutforcesq);
-    MD_SIMD_FLOAT sigma6_vec = simd_broadcast(sigma6);
-    MD_SIMD_FLOAT eps_vec = simd_broadcast(epsilon);
-    MD_SIMD_FLOAT c48_vec = simd_broadcast(48.0);
-    MD_SIMD_FLOAT c05_vec = simd_broadcast(0.5);
-
-
-    #pragma omp parallel
-    {
-    LIKWID_MARKER_START("force");
-
-    #pragma omp for schedule(runtime)
-    for(int i = 0; i < Nlocal; i++) {
-        neighs = &neighbor->neighbors[i * neighbor->maxneighs];
-        int numneighs = neighbor->numneigh[i];
-        MD_SIMD_INT numneighs_vec = simd_int_broadcast(numneighs);
-        MD_SIMD_FLOAT xtmp = simd_broadcast(atom_x(i));
-        MD_SIMD_FLOAT ytmp = simd_broadcast(atom_y(i));
-        MD_SIMD_FLOAT ztmp = simd_broadcast(atom_z(i));
-        MD_SIMD_FLOAT fix = simd_zero();
-        MD_SIMD_FLOAT fiy = simd_zero();
-        MD_SIMD_FLOAT fiz = simd_zero();
-
-        for(int k = 0; k < numneighs; k += VECTOR_WIDTH) {
-            // If the last iteration of this loop is separated from the rest, this mask can be set only there
-            MD_SIMD_MASK mask_numneighs = simd_mask_int_cond_lt(simd_int_add(simd_int_broadcast(k), simd_int_seq()), numneighs_vec);
-            MD_SIMD_INT j = simd_int_mask_load(&neighs[k], mask_numneighs);
-            #ifdef AOS
-            MD_SIMD_INT j3 = simd_int_add(simd_int_add(j, j), j); // j * 3
-            MD_SIMD_FLOAT delx = xtmp - simd_gather(j3, &(atom->x[0]), sizeof(MD_FLOAT));
-            MD_SIMD_FLOAT dely = ytmp - simd_gather(j3, &(atom->x[1]), sizeof(MD_FLOAT));
-            MD_SIMD_FLOAT delz = ztmp - simd_gather(j3, &(atom->x[2]), sizeof(MD_FLOAT));
-            #else
-            MD_SIMD_FLOAT delx = xtmp - simd_gather(j, atom->x, sizeof(MD_FLOAT));
-            MD_SIMD_FLOAT dely = ytmp - simd_gather(j, atom->y, sizeof(MD_FLOAT));
-            MD_SIMD_FLOAT delz = ztmp - simd_gather(j, atom->z, sizeof(MD_FLOAT));
-            #endif
-            MD_SIMD_FLOAT rsq = simd_fma(delx, delx, simd_fma(dely, dely, simd_mul(delz, delz)));
-            MD_SIMD_MASK cutoff_mask = simd_mask_and(mask_numneighs, simd_mask_cond_lt(rsq, cutforcesq_vec));
-            MD_SIMD_FLOAT sr2 = simd_reciprocal(rsq);
-            MD_SIMD_FLOAT sr6 = simd_mul(sr2, simd_mul(sr2, simd_mul(sr2, sigma6_vec)));
-            MD_SIMD_FLOAT force = simd_mul(c48_vec, simd_mul(sr6, simd_mul(simd_sub(sr6, c05_vec), simd_mul(sr2, eps_vec))));
-
-            fix = simd_masked_add(fix, simd_mul(delx, force), cutoff_mask);
-            fiy = simd_masked_add(fiy, simd_mul(dely, force), cutoff_mask);
-            fiz = simd_masked_add(fiz, simd_mul(delz, force), cutoff_mask);
-        }
-
-        atom_fx(i) += simd_h_reduce_sum(fix);
-        atom_fy(i) += simd_h_reduce_sum(fiy);
-        atom_fz(i) += simd_h_reduce_sum(fiz);
-    }
-
-    LIKWID_MARKER_STOP("force");
-    }
-    #endif
-
-    double E = getTimeStamp();
-    return E-S;
-}

lammps/pbc.c

@@ -1,171 +0,0 @@
-/*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
- * All rights reserved. This file is part of MD-Bench.
- * Use of this source code is governed by a LGPL-3.0
- * license that can be found in the LICENSE file.
- */
-#include <stdbool.h>
-#include <stdlib.h>
-#include <stdio.h>
-//---
-#include <pbc.h>
-#include <atom.h>
-#include <allocate.h>
-
-#define DELTA 20000
-
-int NmaxGhost;
-int *PBCx, *PBCy, *PBCz;
-
-static void growPbc(Atom*);
-
-/* exported subroutines */
-void initPbc(Atom* atom) {
-    NmaxGhost = 0;
-    atom->border_map = NULL;
-    PBCx = NULL; PBCy = NULL; PBCz = NULL;
-}
-
-/* update coordinates of ghost atoms */
-/* uses mapping created in setupPbc */
-void updatePbc_cpu(Atom *atom, Parameter *param, bool doReneighbor) {
-    int *border_map = atom->border_map;
-    int nlocal = atom->Nlocal;
-    MD_FLOAT xprd = param->xprd;
-    MD_FLOAT yprd = param->yprd;
-    MD_FLOAT zprd = param->zprd;
-
-    for(int i = 0; i < atom->Nghost; i++) {
-        atom_x(nlocal + i) = atom_x(border_map[i]) + PBCx[i] * xprd;
-        atom_y(nlocal + i) = atom_y(border_map[i]) + PBCy[i] * yprd;
-        atom_z(nlocal + i) = atom_z(border_map[i]) + PBCz[i] * zprd;
-    }
-}
-
-/* relocate atoms that have left domain according
- * to periodic boundary conditions */
-void updateAtomsPbc_cpu(Atom *atom, Parameter *param) {
-    MD_FLOAT xprd = param->xprd;
-    MD_FLOAT yprd = param->yprd;
-    MD_FLOAT zprd = param->zprd;
-
-    for(int i = 0; i < atom->Nlocal; i++) {
-        if(atom_x(i) < 0.0) {
-            atom_x(i) += xprd;
-        } else if(atom_x(i) >= xprd) {
-            atom_x(i) -= xprd;
-        }
-
-        if(atom_y(i) < 0.0) {
-            atom_y(i) += yprd;
-        } else if(atom_y(i) >= yprd) {
-            atom_y(i) -= yprd;
-        }
-
-        if(atom_z(i) < 0.0) {
-            atom_z(i) += zprd;
-        } else if(atom_z(i) >= zprd) {
-            atom_z(i) -= zprd;
-        }
-    }
-}
-
-/* setup periodic boundary conditions by
- * defining ghost atoms around domain
- * only creates mapping and coordinate corrections
- * that are then enforced in updatePbc */
-#define ADDGHOST(dx,dy,dz)                              \
-    Nghost++;                                           \
-    border_map[Nghost] = i;                             \
-    PBCx[Nghost] = dx;                                  \
-    PBCy[Nghost] = dy;                                  \
-    PBCz[Nghost] = dz;                                  \
-    atom->type[atom->Nlocal + Nghost] = atom->type[i]
-
-void setupPbc(Atom *atom, Parameter *param) {
-    int *border_map = atom->border_map;
-    MD_FLOAT xprd = param->xprd;
-    MD_FLOAT yprd = param->yprd;
-    MD_FLOAT zprd = param->zprd;
-    MD_FLOAT Cutneigh = param->cutneigh;
-    int Nghost = -1;
-
-    for(int i = 0; i < atom->Nlocal; i++) {
-        if (atom->Nlocal + Nghost + 7 >= atom->Nmax) {
-            growAtom(atom);
-        }
-
-        if (Nghost + 7 >= NmaxGhost) {
-            growPbc(atom);
-            border_map = atom->border_map;
-        }
-
-        MD_FLOAT x = atom_x(i);
-        MD_FLOAT y = atom_y(i);
-        MD_FLOAT z = atom_z(i);
-
-        /* Setup ghost atoms */
-        /* 6 planes */
-        if(param->pbc_x != 0) {
-            if (x < Cutneigh)         { ADDGHOST(+1,0,0); }
-            if (x >= (xprd-Cutneigh)) { ADDGHOST(-1,0,0); }
-        }
-
-        if(param->pbc_y != 0) {
-            if (y < Cutneigh)         { ADDGHOST(0,+1,0); }
-            if (y >= (yprd-Cutneigh)) { ADDGHOST(0,-1,0); }
-        }
-
-        if(param->pbc_z != 0) {
-            if (z < Cutneigh)         { ADDGHOST(0,0,+1); }
-            if (z >= (zprd-Cutneigh)) { ADDGHOST(0,0,-1); }
-        }
-
-        /* 8 corners */
-        if(param->pbc_x != 0 && param->pbc_y != 0 && param->pbc_z != 0) {
-            if (x < Cutneigh         && y < Cutneigh         && z < Cutneigh)         { ADDGHOST(+1,+1,+1); }
-            if (x < Cutneigh         && y >= (yprd-Cutneigh) && z < Cutneigh)         { ADDGHOST(+1,-1,+1); }
-            if (x < Cutneigh         && y < Cutneigh        && z >= (zprd-Cutneigh))  { ADDGHOST(+1,+1,-1); }
-            if (x < Cutneigh         && y >= (yprd-Cutneigh) && z >= (zprd-Cutneigh)) { ADDGHOST(+1,-1,-1); }
-            if (x >= (xprd-Cutneigh) && y < Cutneigh         && z < Cutneigh)         { ADDGHOST(-1,+1,+1); }
-            if (x >= (xprd-Cutneigh) && y >= (yprd-Cutneigh) && z < Cutneigh)         { ADDGHOST(-1,-1,+1); }
-            if (x >= (xprd-Cutneigh) && y < Cutneigh         && z >= (zprd-Cutneigh)) { ADDGHOST(-1,+1,-1); }
-            if (x >= (xprd-Cutneigh) && y >= (yprd-Cutneigh) && z >= (zprd-Cutneigh)) { ADDGHOST(-1,-1,-1); }
-        }
-
-        /* 12 edges */
-        if(param->pbc_x != 0 && param->pbc_z != 0) {
-            if (x < Cutneigh         && z < Cutneigh)         { ADDGHOST(+1,0,+1); }
-            if (x < Cutneigh         && z >= (zprd-Cutneigh)) { ADDGHOST(+1,0,-1); }
-            if (x >= (xprd-Cutneigh) && z < Cutneigh)         { ADDGHOST(-1,0,+1); }
-            if (x >= (xprd-Cutneigh) && z >= (zprd-Cutneigh)) { ADDGHOST(-1,0,-1); }
-        }
-
-        if(param->pbc_y != 0 && param->pbc_z != 0) {
-            if (y < Cutneigh         && z < Cutneigh)         { ADDGHOST(0,+1,+1); }
-            if (y < Cutneigh         && z >= (zprd-Cutneigh)) { ADDGHOST(0,+1,-1); }
-            if (y >= (yprd-Cutneigh) && z < Cutneigh)         { ADDGHOST(0,-1,+1); }
-            if (y >= (yprd-Cutneigh) && z >= (zprd-Cutneigh)) { ADDGHOST(0,-1,-1); }
-        }
-
-        if(param->pbc_x != 0 && param->pbc_y != 0) {
-            if (y < Cutneigh         && x < Cutneigh)         { ADDGHOST(+1,+1,0); }
-            if (y < Cutneigh         && x >= (xprd-Cutneigh)) { ADDGHOST(-1,+1,0); }
-            if (y >= (yprd-Cutneigh) && x < Cutneigh)         { ADDGHOST(+1,-1,0); }
-            if (y >= (yprd-Cutneigh) && x >= (xprd-Cutneigh)) { ADDGHOST(-1,-1,0); }
-        }
-    }
-    // increase by one to make it the ghost atom count
-    atom->Nghost = Nghost + 1;
-}
-
-/* internal subroutines */
-void growPbc(Atom* atom) {
-    int nold = NmaxGhost;
-    NmaxGhost += DELTA;
-
-    atom->border_map = (int*) reallocate(atom->border_map, ALIGNMENT, NmaxGhost * sizeof(int), nold * sizeof(int));
-    PBCx = (int*) reallocate(PBCx, ALIGNMENT, NmaxGhost * sizeof(int), nold * sizeof(int));
-    PBCy = (int*) reallocate(PBCy, ALIGNMENT, NmaxGhost * sizeof(int), nold * sizeof(int));
-    PBCz = (int*) reallocate(PBCz, ALIGNMENT, NmaxGhost * sizeof(int), nold * sizeof(int));
-}

util/gather-bench/LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2021 RRZE-HPC
+Copyright (c)  RRZE-HPC
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

util/gather-bench/src/main-md-trace.c

@@ -2,7 +2,7 @@
  * =======================================================================================
  *
  *      Author:   Jan Eitzinger (je), jan.eitzinger@fau.de
- *      Copyright (c) 2021 RRZE, University Erlangen-Nuremberg
+ *      Copyright (c)  RRZE, University Erlangen-Nuremberg
  *
  *      Permission is hereby granted, free of charge, to any person obtaining a copy
  *      of this software and associated documentation files (the "Software"), to deal

util/gather-bench/src/main-md.c

@@ -2,7 +2,7 @@
  * =======================================================================================
  *
  *      Author:   Jan Eitzinger (je), jan.eitzinger@fau.de
- *      Copyright (c) 2021 RRZE, University Erlangen-Nuremberg
+ *      Copyright (c)  RRZE, University Erlangen-Nuremberg
  *
  *      Permission is hereby granted, free of charge, to any person obtaining a copy
  *      of this software and associated documentation files (the "Software"), to deal

util/gather-bench/src/main.c

@@ -2,7 +2,7 @@
  * =======================================================================================
  *
  *      Author:   Jan Eitzinger (je), jan.eitzinger@fau.de
- *      Copyright (c) 2021 RRZE, University Erlangen-Nuremberg
+ *      Copyright (c)  RRZE, University Erlangen-Nuremberg
  *
  *      Permission is hereby granted, free of charge, to any person obtaining a copy
  *      of this software and associated documentation files (the "Software"), to deal

util/mdBench.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/atom.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/cuda/force.cu

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/cuda/neighbor.cu

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/cuda/pbc.cu

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/device_spec.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/force_dem.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/force_eam.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/force_lj.c

@@ -0,0 +1,298 @@
+/*
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
+ * All rights reserved. This file is part of MD-Bench.
+ * Use of this source code is governed by a LGPL-3.0
+ * license that can be found in the LICENSE file.
+ */
+#include <stdio.h>
+#include <stdlib.h>
+//---
+#include <atom.h>
+#include <likwid-marker.h>
+#include <neighbor.h>
+#include <parameter.h>
+#include <stats.h>
+#include <timing.h>
+
+#ifdef __SIMD_KERNEL__
+#include <simd.h>
+#endif
+
+double computeForceLJFullNeigh(
+    Parameter* param, Atom* atom, Neighbor* neighbor, Stats* stats)
+{
+    int nLocal = atom->Nlocal;
+    int* neighs;
+#ifndef EXPLICIT_TYPES
+    MD_FLOAT cutforcesq = param->cutforce * param->cutforce;
+    MD_FLOAT sigma6     = param->sigma6;
+    MD_FLOAT epsilon    = param->epsilon;
+#endif
+    const MD_FLOAT num1  = 1.0;
+    const MD_FLOAT num48 = 48.0;
+    const MD_FLOAT num05 = 0.5;
+
+    for (int i = 0; i < nLocal; i++) {
+        atom_fx(i) = 0.0;
+        atom_fy(i) = 0.0;
+        atom_fz(i) = 0.0;
+    }
+    double timeStart = getTimeStamp();
+
+#pragma omp parallel
+    {
+        LIKWID_MARKER_START("force");
+
+#pragma omp for schedule(runtime)
+        for (int i = 0; i < nLocal; i++) {
+            neighs        = &neighbor->neighbors[i * neighbor->maxneighs];
+            int numneighs = neighbor->numneigh[i];
+            MD_FLOAT xtmp = atom_x(i);
+            MD_FLOAT ytmp = atom_y(i);
+            MD_FLOAT ztmp = atom_z(i);
+            MD_FLOAT fix  = 0;
+            MD_FLOAT fiy  = 0;
+            MD_FLOAT fiz  = 0;
+
+#ifdef EXPLICIT_TYPES
+            const int type_i = atom->type[i];
+#endif
+
+            for (int k = 0; k < numneighs; k++) {
+                int j         = neighs[k];
+                MD_FLOAT delx = xtmp - atom_x(j);
+                MD_FLOAT dely = ytmp - atom_y(j);
+                MD_FLOAT delz = ztmp - atom_z(j);
+                MD_FLOAT rsq  = delx * delx + dely * dely + delz * delz;
+
+#ifdef EXPLICIT_TYPES
+                const int type_j          = atom->type[j];
+                const int type_ij         = type_i * atom->ntypes + type_j;
+                const MD_FLOAT cutforcesq = atom->cutforcesq[type_ij];
+                const MD_FLOAT sigma6     = atom->sigma6[type_ij];
+                const MD_FLOAT epsilon    = atom->epsilon[type_ij];
+#endif
+
+                if (rsq < cutforcesq) {
+                    MD_FLOAT sr2   = num1 / rsq;
+                    MD_FLOAT sr6   = sr2 * sr2 * sr2 * sigma6;
+                    MD_FLOAT force = num48 * sr6 * (sr6 - num05) * sr2 * epsilon;
+                    fix += delx * force;
+                    fiy += dely * force;
+                    fiz += delz * force;
+#ifdef USE_REFERENCE_VERSION
+                    addStat(stats->atoms_within_cutoff, 1);
+                } else {
+                    addStat(stats->atoms_outside_cutoff, 1);
+#endif
+                }
+            }
+
+            atom_fx(i) += fix;
+            atom_fy(i) += fiy;
+            atom_fz(i) += fiz;
+
+#ifdef USE_REFERENCE_VERSION
+            if (numneighs % VECTOR_WIDTH > 0) {
+                addStat(stats->atoms_outside_cutoff,
+                    VECTOR_WIDTH - (numneighs % VECTOR_WIDTH));
+            }
+#endif
+
+            addStat(stats->total_force_neighs, numneighs);
+            addStat(stats->total_force_iters,
+                (numneighs + VECTOR_WIDTH - 1) / VECTOR_WIDTH);
+        }
+
+        LIKWID_MARKER_STOP("force");
+    }
+
+    double timeStop = getTimeStamp();
+    return timeStop - timeStart;
+}
+
+double computeForceLJHalfNeigh(
+    Parameter* param, Atom* atom, Neighbor* neighbor, Stats* stats)
+{
+    int nlocal = atom->Nlocal;
+    int* neighs;
+#ifndef EXPLICIT_TYPES
+    MD_FLOAT cutforcesq = param->cutforce * param->cutforce;
+    MD_FLOAT sigma6     = param->sigma6;
+    MD_FLOAT epsilon    = param->epsilon;
+#endif
+    const MD_FLOAT num1  = 1.0;
+    const MD_FLOAT num48 = 48.0;
+    const MD_FLOAT num05 = 0.5;
+
+    for (int i = 0; i < nlocal; i++) {
+        atom_fx(i) = 0.0;
+        atom_fy(i) = 0.0;
+        atom_fz(i) = 0.0;
+    }
+
+    double timeStart = getTimeStamp();
+
+#pragma omp parallel
+    {
+        LIKWID_MARKER_START("forceLJ-halfneigh");
+
+#pragma omp for schedule(runtime)
+        for (int i = 0; i < nlocal; i++) {
+            neighs        = &neighbor->neighbors[i * neighbor->maxneighs];
+            int numneighs = neighbor->numneigh[i];
+            MD_FLOAT xtmp = atom_x(i);
+            MD_FLOAT ytmp = atom_y(i);
+            MD_FLOAT ztmp = atom_z(i);
+            MD_FLOAT fix  = 0;
+            MD_FLOAT fiy  = 0;
+            MD_FLOAT fiz  = 0;
+
+#ifdef EXPLICIT_TYPES
+            const int type_i = atom->type[i];
+#endif
+
+// Pragma required to vectorize the inner loop
+#ifdef ENABLE_OMP_SIMD
+#pragma omp simd reduction(+ : fix, fiy, fiz)
+#endif
+            for (int k = 0; k < numneighs; k++) {
+                int j         = neighs[k];
+                MD_FLOAT delx = xtmp - atom_x(j);
+                MD_FLOAT dely = ytmp - atom_y(j);
+                MD_FLOAT delz = ztmp - atom_z(j);
+                MD_FLOAT rsq  = delx * delx + dely * dely + delz * delz;
+
+#ifdef EXPLICIT_TYPES
+                const int type_j          = atom->type[j];
+                const int type_ij         = type_i * atom->ntypes + type_j;
+                const MD_FLOAT cutforcesq = atom->cutforcesq[type_ij];
+                const MD_FLOAT sigma6     = atom->sigma6[type_ij];
+                const MD_FLOAT epsilon    = atom->epsilon[type_ij];
+#endif
+
+                if (rsq < cutforcesq) {
+                    MD_FLOAT sr2   = num1 / rsq;
+                    MD_FLOAT sr6   = sr2 * sr2 * sr2 * sigma6;
+                    MD_FLOAT force = num48 * sr6 * (sr6 - num05) * sr2 * epsilon;
+                    fix += delx * force;
+                    fiy += dely * force;
+                    fiz += delz * force;
+
+                    // We do not need to update forces for ghost atoms
+                    if (j < nlocal) {
+                        atom_fx(j) -= delx * force;
+                        atom_fy(j) -= dely * force;
+                        atom_fz(j) -= delz * force;
+                    }
+                }
+            }
+
+            atom_fx(i) += fix;
+            atom_fy(i) += fiy;
+            atom_fz(i) += fiz;
+
+            addStat(stats->total_force_neighs, numneighs);
+            addStat(stats->total_force_iters,
+                (numneighs + VECTOR_WIDTH - 1) / VECTOR_WIDTH);
+        }
+
+        LIKWID_MARKER_STOP("forceLJ-halfneigh");
+    }
+
+    double timeStop = getTimeStamp();
+    return timeStop - timeStart;
+}
+
+double computeForceLJFullNeigh_simd(
+    Parameter* param, Atom* atom, Neighbor* neighbor, Stats* stats)
+{
+    int Nlocal = atom->Nlocal;
+    int* neighs;
+    MD_FLOAT cutforcesq = param->cutforce * param->cutforce;
+    MD_FLOAT sigma6     = param->sigma6;
+    MD_FLOAT epsilon    = param->epsilon;
+
+    for (int i = 0; i < Nlocal; i++) {
+        atom_fx(i) = 0.0;
+        atom_fy(i) = 0.0;
+        atom_fz(i) = 0.0;
+    }
+
+    double S = getTimeStamp();
+
+#ifndef __SIMD_KERNEL__
+    fprintf(stderr, "Error: SIMD kernel not implemented for specified instruction set!");
+    exit(-1);
+#else
+    MD_SIMD_FLOAT cutforcesq_vec = simd_broadcast(cutforcesq);
+    MD_SIMD_FLOAT sigma6_vec     = simd_broadcast(sigma6);
+    MD_SIMD_FLOAT eps_vec        = simd_broadcast(epsilon);
+    MD_SIMD_FLOAT c48_vec        = simd_broadcast(48.0);
+    MD_SIMD_FLOAT c05_vec        = simd_broadcast(0.5);
+
+#pragma omp parallel
+    {
+        LIKWID_MARKER_START("force");
+
+#pragma omp for schedule(runtime)
+        for (int i = 0; i < Nlocal; i++) {
+            neighs                    = &neighbor->neighbors[i * neighbor->maxneighs];
+            int numneighs             = neighbor->numneigh[i];
+            MD_SIMD_INT numneighs_vec = simd_int_broadcast(numneighs);
+            MD_SIMD_FLOAT xtmp        = simd_broadcast(atom_x(i));
+            MD_SIMD_FLOAT ytmp        = simd_broadcast(atom_y(i));
+            MD_SIMD_FLOAT ztmp        = simd_broadcast(atom_z(i));
+            MD_SIMD_FLOAT fix         = simd_zero();
+            MD_SIMD_FLOAT fiy         = simd_zero();
+            MD_SIMD_FLOAT fiz         = simd_zero();
+
+            for (int k = 0; k < numneighs; k += VECTOR_WIDTH) {
+                // If the last iteration of this loop is separated from the rest, this
+                // mask can be set only there
+                MD_SIMD_MASK mask_numneighs = simd_mask_int_cond_lt(
+                    simd_int_add(simd_int_broadcast(k), simd_int_seq()),
+                    numneighs_vec);
+                MD_SIMD_INT j = simd_int_mask_load(&neighs[k], mask_numneighs);
+#ifdef AOS
+                MD_SIMD_INT j3     = simd_int_add(simd_int_add(j, j), j); // j * 3
+                MD_SIMD_FLOAT delx = xtmp -
+                                     simd_gather(j3, &(atom->x[0]), sizeof(MD_FLOAT));
+                MD_SIMD_FLOAT dely = ytmp -
+                                     simd_gather(j3, &(atom->x[1]), sizeof(MD_FLOAT));
+                MD_SIMD_FLOAT delz = ztmp -
+                                     simd_gather(j3, &(atom->x[2]), sizeof(MD_FLOAT));
+#else
+                MD_SIMD_FLOAT delx = xtmp - simd_gather(j, atom->x, sizeof(MD_FLOAT));
+                MD_SIMD_FLOAT dely = ytmp - simd_gather(j, atom->y, sizeof(MD_FLOAT));
+                MD_SIMD_FLOAT delz = ztmp - simd_gather(j, atom->z, sizeof(MD_FLOAT));
+#endif
+                MD_SIMD_FLOAT rsq        = simd_fma(delx,
+                    delx,
+                    simd_fma(dely, dely, simd_mul(delz, delz)));
+                MD_SIMD_MASK cutoff_mask = simd_mask_and(mask_numneighs,
+                    simd_mask_cond_lt(rsq, cutforcesq_vec));
+                MD_SIMD_FLOAT sr2        = simd_reciprocal(rsq);
+                MD_SIMD_FLOAT sr6        = simd_mul(sr2,
+                    simd_mul(sr2, simd_mul(sr2, sigma6_vec)));
+                MD_SIMD_FLOAT force      = simd_mul(c48_vec,
+                    simd_mul(sr6,
+                        simd_mul(simd_sub(sr6, c05_vec), simd_mul(sr2, eps_vec))));
+
+                fix = simd_masked_add(fix, simd_mul(delx, force), cutoff_mask);
+                fiy = simd_masked_add(fiy, simd_mul(dely, force), cutoff_mask);
+                fiz = simd_masked_add(fiz, simd_mul(delz, force), cutoff_mask);
+            }
+
+            atom_fx(i) += simd_h_reduce_sum(fix);
+            atom_fy(i) += simd_h_reduce_sum(fiy);
+            atom_fz(i) += simd_h_reduce_sum(fiz);
+        }
+
+        LIKWID_MARKER_STOP("force");
+    }
+#endif
+
+    double E = getTimeStamp();
+    return E - S;
+}

verletlist/includes/atom.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/includes/integrate.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/includes/neighbor.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/includes/pbc.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.
@@ -11,7 +11,7 @@
 
 #ifndef __PBC_H_
 #define __PBC_H_
-extern void initPbc();
+extern void initPbc(Atom*);
 extern void updatePbc_cpu(Atom*, Parameter*, bool);
 extern void updateAtomsPbc_cpu(Atom*, Parameter*);
 extern void setupPbc(Atom*, Parameter*);

verletlist/includes/stats.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/includes/tracing.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/includes/vtk.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/main-stub.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/main.c

@@ -1,18 +1,16 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.
  */
-#include <stdlib.h>
+#include <math.h>
 #include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
-#include <limits.h>
-#include <math.h>
-#include <float.h>
-#include <omp.h>
 
+// #include <omp.h>
 #include <likwid-marker.h>
 
 #include <allocate.h>
@@ -20,20 +18,19 @@
 #include <device.h>
 #include <eam.h>
 #include <integrate.h>
-#include <thermo.h>
-#include <timing.h>
 #include <neighbor.h>
 #include <parameter.h>
 #include <pbc.h>
 #include <stats.h>
+#include <thermo.h>
 #include <timers.h>
+#include <timing.h>
 #include <util.h>
 #include <vtk.h>
 
-#define HLINE "----------------------------------------------------------------------------\n"
+#define HLINE "------------------------------------------------------------------\n"
 
 extern double computeForceLJFullNeigh_plain_c(Parameter*, Atom*, Neighbor*, Stats*);
-extern double computeForceLJFullNeigh_simd(Parameter*, Atom*, Neighbor*, Stats*);
 extern double computeForceLJHalfNeigh(Parameter*, Atom*, Neighbor*, Stats*);
 extern double computeForceEam(Eam*, Parameter*, Atom*, Neighbor*, Stats*);
 extern double computeForceDemFullNeigh(Parameter*, Atom*, Neighbor*, Stats*);
@@ -42,15 +39,18 @@ extern double computeForceDemFullNeigh(Parameter*, Atom*, Neighbor*, Stats*);
 extern double computeForceLJFullNeigh_cuda(Parameter*, Atom*, Neighbor*);
 #endif
 
-double setup(Parameter *param, Eam *eam, Atom *atom, Neighbor *neighbor, Stats *stats) {
+double setup(Parameter* param, Eam* eam, Atom* atom, Neighbor* neighbor, Stats* stats)
-    if(param->force_field == FF_EAM) { initEam(eam, param); }
+{
-    double S, E;
+    if (param->force_field == FF_EAM) {
+        initEam(eam, param);
+    }
+    double timeStart, timeStop;
     param->lattice = pow((4.0 / param->rho), (1.0 / 3.0));
     param->xprd    = param->nx * param->lattice;
     param->yprd    = param->ny * param->lattice;
     param->zprd    = param->nz * param->lattice;
 
-    S = getTimeStamp();
+    timeStart = getTimeStamp();
     initAtom(atom);
     initPbc(atom);
     initStats(stats);
@@ -63,7 +63,9 @@ double setup(Parameter *param, Eam *eam, Atom *atom, Neighbor *neighbor, Stats *
 
     setupNeighbor(param);
     setupThermo(param, atom->Natoms);
-    if(param->input_file == NULL) { adjustThermo(param, atom); }
+    if (param->input_file == NULL) {
+        adjustThermo(param, atom);
+    }
 #ifdef SORT_ATOMS
     atom->Nghost = 0;
     sortAtom(atom);
@@ -72,13 +74,14 @@ double setup(Parameter *param, Eam *eam, Atom *atom, Neighbor *neighbor, Stats *
     initDevice(atom, neighbor);
     updatePbc(atom, param, true);
     buildNeighbor(atom, neighbor);
-    E = getTimeStamp();
+    timeStop = getTimeStamp();
-    return E-S;
+    return timeStop - timeStart;
 }
 
-double reneighbour(Parameter *param, Atom *atom, Neighbor *neighbor) {
+double reneighbour(Parameter* param, Atom* atom, Neighbor* neighbor)
-    double S, E;
+{
-    S = getTimeStamp();
+    double timeStart, timeStop;
+    timeStart = getTimeStamp();
     LIKWID_MARKER_START("reneighbour");
     updateAtomsPbc(atom, param);
 #ifdef SORT_ATOMS
@@ -89,19 +92,26 @@ double reneighbour(Parameter *param, Atom *atom, Neighbor *neighbor) {
     updatePbc(atom, param, true);
     buildNeighbor(atom, neighbor);
     LIKWID_MARKER_STOP("reneighbour");
-    E = getTimeStamp();
+    timeStop = getTimeStamp();
-    return E-S;
+    return timeStop - timeStart;
 }
 
-void printAtomState(Atom *atom) {
+void printAtomState(Atom* atom)
-    printf("Atom counts: Natoms=%d Nlocal=%d Nghost=%d Nmax=%d\n", atom->Natoms, atom->Nlocal, atom->Nghost, atom->Nmax);
+{
+    printf("Atom counts: Natoms=%d Nlocal=%d Nghost=%d Nmax=%d\n",
+        atom->Natoms,
+        atom->Nlocal,
+        atom->Nghost,
+        atom->Nmax);
     // int nall = atom->Nlocal + atom->Nghost;
     // for (int i=0; i<nall; i++) {
     //     printf("%d  %f %f %f\n", i, atom->x[i], atom->y[i], atom->z[i]);
     // }
 }
 
-double computeForce(Eam *eam, Parameter *param, Atom *atom, Neighbor *neighbor, Stats *stats) {
+double computeForce(
+    Eam* eam, Parameter* param, Atom* atom, Neighbor* neighbor, Stats* stats)
+{
     if (param->force_field == FF_EAM) {
         return computeForceEam(eam, param, atom, neighbor, stats);
     } else if (param->force_field == FF_DEM) {
@@ -124,18 +134,27 @@ double computeForce(Eam *eam, Parameter *param, Atom *atom, Neighbor *neighbor,
 #endif
 }
 
-void writeInput(Parameter *param, Atom *atom) {
+void writeInput(Parameter* param, Atom* atom)
+{
     FILE* fpin = fopen("input.in", "w");
     fprintf(fpin, "0,%f,0,%f,0,%f\n", param->xprd, param->yprd, param->zprd);
 
     for (int i = 0; i < atom->Nlocal; i++) {
-        fprintf(fpin, "1,%f,%f,%f,%f,%f,%f\n", atom_x(i), atom_y(i), atom_z(i), atom_vx(i), atom_vy(i), atom_vz(i));
+        fprintf(fpin,
+            "1,%f,%f,%f,%f,%f,%f\n",
+            atom_x(i),
+            atom_y(i),
+            atom_z(i),
+            atom_vx(i),
+            atom_vy(i),
+            atom_vz(i));
     }
 
     fclose(fpin);
 }
 
-int main(int argc, char** argv) {
+int main(int argc, char** argv)
+{
     double timer[NUMTIMER];
     Eam eam;
     Atom atom;
@@ -215,13 +234,19 @@ int main(int argc, char** argv) {
         if ((strcmp(argv[i], "-h") == 0) || (strcmp(argv[i], "--help") == 0)) {
             printf("MD Bench: A minimalistic re-implementation of miniMD\n");
             printf(HLINE);
-            printf("-p / --params <string>:     file to read parameters from (can be specified more than once)\n");
+            printf("-p / --params <string>:     file to read parameters from (can be "
-            printf("-f <string>:                force field (lj, eam or dem), default lj\n");
+                   "specified more than once)\n");
-            printf("-i <string>:                input file with atom positions (dump)\n");
+            printf("-f <string>:                force field (lj, eam or dem), "
+                   "default lj\n");
+            printf("-i <string>:                input file with atom positions "
+                   "(dump)\n");
             printf("-e <string>:                input file for EAM\n");
-            printf("-n / --nsteps <int>:        set number of timesteps for simulation\n");
+            printf("-n / --nsteps <int>:        set number of timesteps for "
-            printf("-nx/-ny/-nz <int>:          set linear dimension of systembox in x/y/z direction\n");
+                   "simulation\n");
-            printf("-half <int>:                use half (1) or full (0) neighbor lists\n");
+            printf("-nx/-ny/-nz <int>:          set linear dimension of systembox in "
+                   "x/y/z direction\n");
+            printf("-half <int>:                use half (1) or full (0) neighbor "
+                   "lists\n");
             printf("-r / --radius <real>:       set cutoff radius\n");
             printf("-s / --skin <real>:         set skin (verlet buffer)\n");
             printf("-w <file>:                  write input atoms to file\n");
@@ -289,33 +314,46 @@ int main(int argc, char** argv) {
     computeThermo(-1, &param, &atom);
 
     printf(HLINE);
-    printf("System: %d atoms %d ghost atoms, Steps: %d\n", atom.Natoms, atom.Nghost, param.ntimes);
+    printf("System: %d atoms %d ghost atoms, Steps: %d\n",
+        atom.Natoms,
+        atom.Nghost,
+        param.ntimes);
     printf("TOTAL %.2fs FORCE %.2fs NEIGH %.2fs REST %.2fs\n",
-            timer[TOTAL], timer[FORCE], timer[NEIGH], timer[TOTAL]-timer[FORCE]-timer[NEIGH]);
+        timer[TOTAL],
+        timer[FORCE],
+        timer[NEIGH],
+        timer[TOTAL] - timer[FORCE] - timer[NEIGH]);
     printf(HLINE);
 
-    int nthreads = 0;
+    //   int nthreads = 0;
-    int chunkSize = 0;
+    //   int chunkSize = 0;
-    omp_sched_t schedKind;
+    //   omp_sched_t schedKind;
-    char schedType[10];
+    //   char schedType[10];
-#pragma omp parallel
+    // #pragma omp parallel
-#pragma omp master
+    // #pragma omp master
-    {
+    //   {
-    	omp_get_schedule(&schedKind, &chunkSize);
+    //     omp_get_schedule(&schedKind, &chunkSize);
-
+    //
-    	switch (schedKind)
+    //     switch (schedKind) {
-    	{
+    //     case omp_sched_static:
-        	case omp_sched_static:  strcpy(schedType, "static"); break;
+    //       strcpy(schedType, "static");
-        	case omp_sched_dynamic: strcpy(schedType, "dynamic"); break;
+    //       break;
-        	case omp_sched_guided:  strcpy(schedType, "guided"); break;
+    //     case omp_sched_dynamic:
-        	case omp_sched_auto:    strcpy(schedType, "auto"); break;
+    //       strcpy(schedType, "dynamic");
-    	}
+    //       break;
-	
+    //     case omp_sched_guided:
-	nthreads = omp_get_max_threads();
+    //       strcpy(schedType, "guided");
-    }
+    //       break;
-
+    //     case omp_sched_auto:
-    printf("Num threads: %d\n", nthreads);
+    //       strcpy(schedType, "auto");
-    printf("Schedule: (%s,%d)\n", schedType, chunkSize);
+    //       break;
+    //     }
+    //
+    //     nthreads = omp_get_max_threads();
+    //   }
+    //
+    //   printf("Num threads: %d\n", nthreads);
+    //   printf("Schedule: (%s,%d)\n", schedType, chunkSize);
 
     printf("Performance: %.2f million atom updates per second\n",
         1e-6 * (double)atom.Natoms * param.ntimes / timer[TOTAL]);

verletlist/neighbor.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/pbc.c

@@ -0,0 +1,234 @@
+/*
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
+ * All rights reserved. This file is part of MD-Bench.
+ * Use of this source code is governed by a LGPL-3.0
+ * license that can be found in the LICENSE file.
+ */
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+//---
+#include <allocate.h>
+#include <atom.h>
+#include <pbc.h>
+
+#define DELTA 20000
+
+int nmaxGhost;
+int *PBCx, *PBCy, *PBCz;
+
+static void growPbc(Atom*);
+
+/* exported subroutines */
+void initPbc(Atom* atom)
+{
+    nmaxGhost        = 0;
+    atom->border_map = NULL;
+    PBCx             = NULL;
+    PBCy             = NULL;
+    PBCz             = NULL;
+}
+
+/* update coordinates of ghost atoms */
+/* uses mapping created in setupPbc */
+void updatePbc_cpu(Atom* atom, Parameter* param, bool doReneighbor)
+{
+    int* borderMap = atom->border_map;
+    int nlocal     = atom->Nlocal;
+    MD_FLOAT xprd  = param->xprd;
+    MD_FLOAT yprd  = param->yprd;
+    MD_FLOAT zprd  = param->zprd;
+
+    for (int i = 0; i < atom->Nghost; i++) {
+        atom_x(nlocal + i) = atom_x(borderMap[i]) + PBCx[i] * xprd;
+        atom_y(nlocal + i) = atom_y(borderMap[i]) + PBCy[i] * yprd;
+        atom_z(nlocal + i) = atom_z(borderMap[i]) + PBCz[i] * zprd;
+    }
+}
+
+/* relocate atoms that have left domain according
+ * to periodic boundary conditions */
+void updateAtomsPbc_cpu(Atom* atom, Parameter* param)
+{
+    MD_FLOAT xprd = param->xprd;
+    MD_FLOAT yprd = param->yprd;
+    MD_FLOAT zprd = param->zprd;
+
+    for (int i = 0; i < atom->Nlocal; i++) {
+        if (atom_x(i) < 0.0) {
+            atom_x(i) += xprd;
+        } else if (atom_x(i) >= xprd) {
+            atom_x(i) -= xprd;
+        }
+
+        if (atom_y(i) < 0.0) {
+            atom_y(i) += yprd;
+        } else if (atom_y(i) >= yprd) {
+            atom_y(i) -= yprd;
+        }
+
+        if (atom_z(i) < 0.0) {
+            atom_z(i) += zprd;
+        } else if (atom_z(i) >= zprd) {
+            atom_z(i) -= zprd;
+        }
+    }
+}
+
+/* setup periodic boundary conditions by
+ * defining ghost atoms around domain
+ * only creates mapping and coordinate corrections
+ * that are then enforced in updatePbc */
+#define ADDGHOST(dx, dy, dz)                                                             \
+    Nghost++;                                                                            \
+    border_map[Nghost]                = i;                                               \
+    PBCx[Nghost]                      = dx;                                              \
+    PBCy[Nghost]                      = dy;                                              \
+    PBCz[Nghost]                      = dz;                                              \
+    atom->type[atom->Nlocal + Nghost] = atom->type[i]
+
+void setupPbc(Atom* atom, Parameter* param)
+{
+    int* border_map   = atom->border_map;
+    MD_FLOAT xprd     = param->xprd;
+    MD_FLOAT yprd     = param->yprd;
+    MD_FLOAT zprd     = param->zprd;
+    MD_FLOAT cutneigh = param->cutneigh;
+    int Nghost        = -1;
+
+    for (int i = 0; i < atom->Nlocal; i++) {
+        if (atom->Nlocal + Nghost + 7 >= atom->Nmax) {
+            growAtom(atom);
+        }
+
+        if (Nghost + 7 >= nmaxGhost) {
+            growPbc(atom);
+            border_map = atom->border_map;
+        }
+
+        MD_FLOAT x = atom_x(i);
+        MD_FLOAT y = atom_y(i);
+        MD_FLOAT z = atom_z(i);
+
+        /* Setup ghost atoms */
+        /* 6 planes */
+        if (param->pbc_x != 0) {
+            if (x < cutneigh) {
+                ADDGHOST(+1, 0, 0);
+            }
+            if (x >= (xprd - cutneigh)) {
+                ADDGHOST(-1, 0, 0);
+            }
+        }
+
+        if (param->pbc_y != 0) {
+            if (y < cutneigh) {
+                ADDGHOST(0, +1, 0);
+            }
+            if (y >= (yprd - cutneigh)) {
+                ADDGHOST(0, -1, 0);
+            }
+        }
+
+        if (param->pbc_z != 0) {
+            if (z < cutneigh) {
+                ADDGHOST(0, 0, +1);
+            }
+            if (z >= (zprd - cutneigh)) {
+                ADDGHOST(0, 0, -1);
+            }
+        }
+
+        /* 8 corners */
+        if (param->pbc_x != 0 && param->pbc_y != 0 && param->pbc_z != 0) {
+            if (x < cutneigh && y < cutneigh && z < cutneigh) {
+                ADDGHOST(+1, +1, +1);
+            }
+            if (x < cutneigh && y >= (yprd - cutneigh) && z < cutneigh) {
+                ADDGHOST(+1, -1, +1);
+            }
+            if (x < cutneigh && y < cutneigh && z >= (zprd - cutneigh)) {
+                ADDGHOST(+1, +1, -1);
+            }
+            if (x < cutneigh && y >= (yprd - cutneigh) && z >= (zprd - cutneigh)) {
+                ADDGHOST(+1, -1, -1);
+            }
+            if (x >= (xprd - cutneigh) && y < cutneigh && z < cutneigh) {
+                ADDGHOST(-1, +1, +1);
+            }
+            if (x >= (xprd - cutneigh) && y >= (yprd - cutneigh) && z < cutneigh) {
+                ADDGHOST(-1, -1, +1);
+            }
+            if (x >= (xprd - cutneigh) && y < cutneigh && z >= (zprd - cutneigh)) {
+                ADDGHOST(-1, +1, -1);
+            }
+            if (x >= (xprd - cutneigh) && y >= (yprd - cutneigh) &&
+                z >= (zprd - cutneigh)) {
+                ADDGHOST(-1, -1, -1);
+            }
+        }
+
+        /* 12 edges */
+        if (param->pbc_x != 0 && param->pbc_z != 0) {
+            if (x < cutneigh && z < cutneigh) {
+                ADDGHOST(+1, 0, +1);
+            }
+            if (x < cutneigh && z >= (zprd - cutneigh)) {
+                ADDGHOST(+1, 0, -1);
+            }
+            if (x >= (xprd - cutneigh) && z < cutneigh) {
+                ADDGHOST(-1, 0, +1);
+            }
+            if (x >= (xprd - cutneigh) && z >= (zprd - cutneigh)) {
+                ADDGHOST(-1, 0, -1);
+            }
+        }
+
+        if (param->pbc_y != 0 && param->pbc_z != 0) {
+            if (y < cutneigh && z < cutneigh) {
+                ADDGHOST(0, +1, +1);
+            }
+            if (y < cutneigh && z >= (zprd - cutneigh)) {
+                ADDGHOST(0, +1, -1);
+            }
+            if (y >= (yprd - cutneigh) && z < cutneigh) {
+                ADDGHOST(0, -1, +1);
+            }
+            if (y >= (yprd - cutneigh) && z >= (zprd - cutneigh)) {
+                ADDGHOST(0, -1, -1);
+            }
+        }
+
+        if (param->pbc_x != 0 && param->pbc_y != 0) {
+            if (y < cutneigh && x < cutneigh) {
+                ADDGHOST(+1, +1, 0);
+            }
+            if (y < cutneigh && x >= (xprd - cutneigh)) {
+                ADDGHOST(-1, +1, 0);
+            }
+            if (y >= (yprd - cutneigh) && x < cutneigh) {
+                ADDGHOST(+1, -1, 0);
+            }
+            if (y >= (yprd - cutneigh) && x >= (xprd - cutneigh)) {
+                ADDGHOST(-1, -1, 0);
+            }
+        }
+    }
+    // increase by one to make it the ghost atom count
+    atom->Nghost = Nghost + 1;
+}
+
+/* internal subroutines */
+void growPbc(Atom* atom)
+{
+    int nold = nmaxGhost;
+    nmaxGhost += DELTA;
+
+    atom->border_map = (int*)reallocate(atom->border_map,
+        ALIGNMENT,
+        nmaxGhost * sizeof(int),
+        nold * sizeof(int));
+    PBCx = (int*)reallocate(PBCx, ALIGNMENT, nmaxGhost * sizeof(int), nold * sizeof(int));
+    PBCy = (int*)reallocate(PBCy, ALIGNMENT, nmaxGhost * sizeof(int), nold * sizeof(int));
+    PBCz = (int*)reallocate(PBCz, ALIGNMENT, nmaxGhost * sizeof(int), nold * sizeof(int));
+}

verletlist/stats.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/tracing.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.

verletlist/vtk.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2022 NHR@FAU, University Erlangen-Nuremberg.
+ * Copyright (C)  NHR@FAU, University Erlangen-Nuremberg.
  * All rights reserved. This file is part of MD-Bench.
  * Use of this source code is governed by a LGPL-3.0
  * license that can be found in the LICENSE file.