2024-04-20 nightly release (70baafe)

.ci/docker/ci_commit_pins/pytorch.txt

@@ -1 +1 @@
-0a038cf0cff2d071b7359ac0491fd2ba7798a438
+19f50333e91e9e8b20a78517becd74bca70c7d46

.ci/scripts/build_llama_android.sh

@@ -26,6 +26,7 @@ install_executorch_and_backend_lib() {
     -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
     -DEXECUTORCH_BUILD_XNNPACK=ON \
     -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+    -DEXECUTORCH_BUILD_QUANTIZED=ON \
     -DXNNPACK_ENABLE_ARM_BF16=OFF \
     -Bcmake-android-out .
 

.ci/scripts/test_llama.sh

@@ -12,7 +12,7 @@ source "$(dirname "${BASH_SOURCE[0]}")/utils.sh"
 MODEL_NAME=$1 # stories110M.pt
 BUILD_TOOL=$2 # buck2 or cmake
 DTYPE=$3 # fp16 or fp32
-MODE=${4:-"xnnpack"} # portable or xnnpack
+MODE=${4:-"xnnpack+custom"} # portable or xnnpack+custom or xnnpack+custom+qe
 if [[ $# -lt 4 ]]; then # Assuming 4 mandatory args
     echo "Expecting atleast 4 positional arguments"
     echo "Usage: [...]"
@@ -37,6 +37,24 @@ if [[ -z "${MODE:-}" ]]; then
   exit 1
 fi
 
+if [[ "${MODE}" =~ .*xnnpack.* ]]; then
+  XNNPACK=ON
+else
+  XNNPACK=OFF
+fi
+
+if [[ "${MODE}" =~ .*custom.* ]]; then
+  CUSTOM=ON
+else
+  CUSTOM=OFF
+fi
+
+if [[ "${MODE}" =~ .*qe.* ]]; then
+  QE=ON
+else
+  QE=OFF
+fi
+
 if [[ -z "${BUCK:-}" ]]; then
   BUCK=buck2
 fi
@@ -47,38 +65,36 @@ fi
 
 which "${PYTHON_EXECUTABLE}"
 
-
 cmake_install_executorch_libraries() {
     echo "Installing libexecutorch.a, libextension_module.so, libportable_ops_lib.a"
     rm -rf cmake-out
-    if [[ "${MODE}" == "xnnpack" ]]; then
-      XNNPACK=ON
-    else
-      XNNPACK=OFF
-    fi
     retry cmake -DBUCK2="$BUCK" \
         -DCMAKE_INSTALL_PREFIX=cmake-out \
-        -DCMAKE_BUILD_TYPE=Release \
+        -DCMAKE_BUILD_TYPE=Debug \
         -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
         -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
+        -DEXECUTORCH_BUILD_CUSTOM="$CUSTOM" \
         -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+        -DEXECUTORCH_BUILD_QUANTIZED=ON \
         -DEXECUTORCH_BUILD_XNNPACK="$XNNPACK" \
         -DPYTHON_EXECUTABLE="$PYTHON_EXECUTABLE" \
         -Bcmake-out .
-    cmake --build cmake-out -j9 --target install --config Release
+    cmake --build cmake-out -j9 --target install --config Debug
 }
 
 cmake_build_llama_runner() {
     echo "Building llama runner"
     dir="examples/models/llama2"
     retry cmake -DBUCK2="$BUCK" \
         -DCMAKE_INSTALL_PREFIX=cmake-out \
-        -DCMAKE_BUILD_TYPE=Release \
+        -DCMAKE_BUILD_TYPE=Debug \
+        -DEXECUTORCH_BUILD_CUSTOM="$CUSTOM" \
         -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+        -DEXECUTORCH_BUILD_XNNPACK="$XNNPACK" \
         -DPYTHON_EXECUTABLE="$PYTHON_EXECUTABLE" \
         -Bcmake-out/${dir} \
         ${dir}
-    cmake --build cmake-out/${dir} -j9 --config Release
+    cmake --build cmake-out/${dir} -j9 --config Debug
 
 }
 
@@ -116,10 +132,17 @@ fi
 # Export model.
 EXPORTED_MODEL_NAME="${EXPORTED_MODEL_NAME}.pte"
 echo "Exporting ${EXPORTED_MODEL_NAME}"
-EXPORT_ARGS="-c stories110M.pt -p ${PARAMS} -d ${DTYPE} -n ${EXPORTED_MODEL_NAME}"
-if [[ "${MODE}" == "xnnpack" ]]; then
-  EXPORT_ARGS="${EXPORT_ARGS} -kv --use_sdpa_with_kv_cache -X -qmode 8da4w -G 128"
+EXPORT_ARGS="-c stories110M.pt -p ${PARAMS} -d ${DTYPE} -n ${EXPORTED_MODEL_NAME} -kv"
+if [[ "${XNNPACK}" == "ON" ]]; then
+  EXPORT_ARGS="${EXPORT_ARGS} -X -qmode 8da4w -G 128"
+fi
+if [[ "${CUSTOM}" == "ON" ]]; then
+  EXPORT_ARGS="${EXPORT_ARGS} --use_sdpa_with_kv_cache"
+fi
+if [[ "${QE}" == "ON" ]]; then
+  EXPORT_ARGS="${EXPORT_ARGS} --embedding-quantize 8,1024"
 fi
+# Add dynamically linked library location
 $PYTHON_EXECUTABLE -m examples.models.llama2.export_llama ${EXPORT_ARGS}
 
 # Create tokenizer.bin.

.ci/scripts/test_quantized_aot_lib.sh

@@ -24,7 +24,7 @@ build_cmake_quantized_aot_lib() {
     && retry cmake -DBUCK2=buck2 \
       -DCMAKE_BUILD_TYPE=Release \
       -DCMAKE_PREFIX_PATH="$CMAKE_PREFIX_PATH" \
-      -DEXECUTORCH_BUILD_QUANTIZED=ON \
+      -DEXECUTORCH_BUILD_QUANTIZED_OPS_AOT=ON \
       -DPYTHON_EXECUTABLE="$PYTHON_EXECUTABLE" ..)
 
   cmake --build ${CMAKE_OUTPUT_DIR} -j4

.clang-format

@@ -2,22 +2,55 @@
 Language:        Cpp
 AccessModifierOffset: -1
 AlignAfterOpenBracket: AlwaysBreak
-AlignConsecutiveMacros: None
-AlignConsecutiveAssignments: None
-AlignConsecutiveBitFields: None
-AlignConsecutiveDeclarations: None
+AlignArrayOfStructures: None
+AlignConsecutiveAssignments:
+  Enabled:         false
+  AcrossEmptyLines: false
+  AcrossComments:  false
+  AlignCompound:   false
+  AlignFunctionPointers: false
+  PadOperators:    true
+AlignConsecutiveBitFields:
+  Enabled:         false
+  AcrossEmptyLines: false
+  AcrossComments:  false
+  AlignCompound:   false
+  AlignFunctionPointers: false
+  PadOperators:    true
+AlignConsecutiveDeclarations:
+  Enabled:         false
+  AcrossEmptyLines: false
+  AcrossComments:  false
+  AlignCompound:   false
+  AlignFunctionPointers: false
+  PadOperators:    true
+AlignConsecutiveMacros:
+  Enabled:         false
+  AcrossEmptyLines: false
+  AcrossComments:  false
+  AlignCompound:   false
+  AlignFunctionPointers: false
+  PadOperators:    true
+AlignConsecutiveShortCaseStatements:
+  Enabled:         false
+  AcrossEmptyLines: false
+  AcrossComments:  false
+  AlignCaseColons: false
 AlignEscapedNewlines: Left
 AlignOperands:   DontAlign
-AlignTrailingComments: false
+AlignTrailingComments:
+  Kind:            Never
+  OverEmptyLines:  0
 AllowAllArgumentsOnNextLine: true
-AllowAllConstructorInitializersOnNextLine: true
 AllowAllParametersOfDeclarationOnNextLine: false
-AllowShortEnumsOnASingleLine: true
+AllowBreakBeforeNoexceptSpecifier: Never
 AllowShortBlocksOnASingleLine: Never
 AllowShortCaseLabelsOnASingleLine: false
+AllowShortCompoundRequirementOnASingleLine: true
+AllowShortEnumsOnASingleLine: true
 AllowShortFunctionsOnASingleLine: Empty
-AllowShortLambdasOnASingleLine: All
 AllowShortIfStatementsOnASingleLine: Never
+AllowShortLambdasOnASingleLine: All
 AllowShortLoopsOnASingleLine: false
 AlwaysBreakAfterDefinitionReturnType: None
 AlwaysBreakAfterReturnType: None
@@ -27,17 +60,18 @@ AttributeMacros:
   - __capability
 BinPackArguments: false
 BinPackParameters: false
+BitFieldColonSpacing: Both
 BraceWrapping:
   AfterCaseLabel:  false
   AfterClass:      false
   AfterControlStatement: Never
   AfterEnum:       false
+  AfterExternBlock: false
   AfterFunction:   false
   AfterNamespace:  false
   AfterObjCDeclaration: false
   AfterStruct:     false
   AfterUnion:      false
-  AfterExternBlock: false
   BeforeCatch:     false
   BeforeElse:      false
   BeforeLambdaBody: false
@@ -46,35 +80,36 @@ BraceWrapping:
   SplitEmptyFunction: true
   SplitEmptyRecord: true
   SplitEmptyNamespace: true
+BreakAdjacentStringLiterals: true
+BreakAfterAttributes: Leave
+BreakAfterJavaFieldAnnotations: false
+BreakArrays:     true
 BreakBeforeBinaryOperators: None
-BreakBeforeConceptDeclarations: true
+BreakBeforeConceptDeclarations: Always
 BreakBeforeBraces: Attach
-BreakBeforeInheritanceComma: false
-BreakInheritanceList: BeforeColon
+BreakBeforeInlineASMColon: OnlyMultiline
 BreakBeforeTernaryOperators: true
-BreakConstructorInitializersBeforeComma: false
 BreakConstructorInitializers: BeforeColon
-BreakAfterJavaFieldAnnotations: false
+BreakInheritanceList: BeforeColon
 BreakStringLiterals: false
 ColumnLimit:     80
 CommentPragmas:  '^ IWYU pragma:'
 CompactNamespaces: false
-ConstructorInitializerAllOnOneLineOrOnePerLine: true
 ConstructorInitializerIndentWidth: 4
 ContinuationIndentWidth: 4
 Cpp11BracedListStyle: true
-DeriveLineEnding: true
 DerivePointerAlignment: false
 DisableFormat:   false
+EmptyLineAfterAccessModifier: Never
 EmptyLineBeforeAccessModifier: LogicalBlock
 ExperimentalAutoDetectBinPacking: false
 FixNamespaceComments: true
 ForEachMacros:
   - FOR_EACH
   - FOR_EACH_R
   - FOR_EACH_RANGE
-StatementAttributeLikeMacros:
-  - Q_EMIT
+IfMacros:
+  - KJ_IF_MAYBE
 IncludeBlocks:   Preserve
 IncludeCategories:
   - Regex:           '^<.*\.h(pp)?>'
@@ -91,18 +126,31 @@ IncludeCategories:
     CaseSensitive:   false
 IncludeIsMainRegex: '(Test)?$'
 IncludeIsMainSourceRegex: ''
-IndentCaseLabels: true
+IndentAccessModifiers: false
 IndentCaseBlocks: false
+IndentCaseLabels: true
+IndentExternBlock: AfterExternBlock
 IndentGotoLabels: true
 IndentPPDirectives: None
-IndentExternBlock: AfterExternBlock
-IndentRequires:  false
+IndentRequiresClause: true
 IndentWidth:     2
 IndentWrappedFunctionNames: false
+InsertBraces:    false
+InsertNewlineAtEOF: false
 InsertTrailingCommas: None
+IntegerLiteralSeparator:
+  Binary:          0
+  BinaryMinDigits: 0
+  Decimal:         0
+  DecimalMinDigits: 0
+  Hex:             0
+  HexMinDigits:    0
 JavaScriptQuotes: Leave
 JavaScriptWrapImports: true
 KeepEmptyLinesAtTheStartOfBlocks: false
+KeepEmptyLinesAtEOF: false
+LambdaBodyIndentation: Signature
+LineEnding:      DeriveLF
 MacroBlockBegin: ''
 MacroBlockEnd:   ''
 MaxEmptyLinesToKeep: 1
@@ -112,53 +160,85 @@ ObjCBlockIndentWidth: 2
 ObjCBreakBeforeNestedBlockParam: true
 ObjCSpaceAfterProperty: false
 ObjCSpaceBeforeProtocolList: false
+PackConstructorInitializers: NextLine
 PenaltyBreakAssignment: 2
 PenaltyBreakBeforeFirstCallParameter: 1
 PenaltyBreakComment: 300
 PenaltyBreakFirstLessLess: 120
+PenaltyBreakOpenParenthesis: 0
+PenaltyBreakScopeResolution: 500
 PenaltyBreakString: 1000
 PenaltyBreakTemplateDeclaration: 10
 PenaltyExcessCharacter: 1000000
-PenaltyReturnTypeOnItsOwnLine: 200
 PenaltyIndentedWhitespace: 0
+PenaltyReturnTypeOnItsOwnLine: 200
 PointerAlignment: Left
+PPIndentWidth:   -1
+QualifierAlignment: Leave
+ReferenceAlignment: Pointer
 ReflowComments:  true
-SortIncludes:    true
+RemoveBracesLLVM: false
+RemoveParentheses: Leave
+RemoveSemicolon: false
+RequiresClausePosition: OwnLine
+RequiresExpressionIndentation: OuterScope
+SeparateDefinitionBlocks: Leave
+ShortNamespaceLines: 1
+SkipMacroDefinitionBody: false
+SortIncludes:    CaseSensitive
 SortJavaStaticImport: Before
-SortUsingDeclarations: true
+SortUsingDeclarations: LexicographicNumeric
 SpaceAfterCStyleCast: false
 SpaceAfterLogicalNot: false
 SpaceAfterTemplateKeyword: true
+SpaceAroundPointerQualifiers: Default
 SpaceBeforeAssignmentOperators: true
 SpaceBeforeCaseColon: false
 SpaceBeforeCpp11BracedList: false
 SpaceBeforeCtorInitializerColon: true
 SpaceBeforeInheritanceColon: true
+SpaceBeforeJsonColon: false
 SpaceBeforeParens: ControlStatements
-SpaceAroundPointerQualifiers: Default
+SpaceBeforeParensOptions:
+  AfterControlStatements: true
+  AfterForeachMacros: true
+  AfterFunctionDefinitionName: false
+  AfterFunctionDeclarationName: false
+  AfterIfMacros:   true
+  AfterOverloadedOperator: false
+  AfterPlacementOperator: true
+  AfterRequiresInClause: false
+  AfterRequiresInExpression: false
+  BeforeNonEmptyParentheses: false
 SpaceBeforeRangeBasedForLoopColon: true
+SpaceBeforeSquareBrackets: false
 SpaceInEmptyBlock: false
-SpaceInEmptyParentheses: false
 SpacesBeforeTrailingComments: 1
-SpacesInAngles:  false
-SpacesInConditionalStatement: false
+SpacesInAngles:  Never
 SpacesInContainerLiterals: true
-SpacesInCStyleCastParentheses: false
-SpacesInParentheses: false
+SpacesInLineCommentPrefix:
+  Minimum:         1
+  Maximum:         -1
+SpacesInParens:  Never
+SpacesInParensOptions:
+  InCStyleCasts:   false
+  InConditionalStatements: false
+  InEmptyParentheses: false
+  Other:           false
 SpacesInSquareBrackets: false
-SpaceBeforeSquareBrackets: false
-BitFieldColonSpacing: Both
 Standard:        Latest
+StatementAttributeLikeMacros:
+  - Q_EMIT
 StatementMacros:
   - Q_UNUSED
   - QT_REQUIRE_VERSION
 TabWidth:        8
-UseCRLF:         false
 UseTab:          Never
+VerilogBreakBetweenInstancePorts: true
 WhitespaceSensitiveMacros:
-  - STRINGIZE
-  - PP_STRINGIZE
   - BOOST_PP_STRINGIZE
-  - NS_SWIFT_NAME
   - CF_SWIFT_NAME
+  - NS_SWIFT_NAME
+  - PP_STRINGIZE
+  - STRINGIZE
 ...