setTflCOperationEntryParameters - Set specified parameters for operator entry in code replacement
table - MATLAB ([original](https://www.mathworks.com/help/ecoder/ref/settflcoperationentryparameters.html)) ([raw](?raw))
Set specified parameters for operator entry in code replacement table
Syntax
Description
setTflCOperationEntryParameters([hEntry](#d126e47059),[varargin](#d126e47197))
sets specified parameters for an operator entry in a code replacement table.
Examples
This example shows how to use thesetTflCOperationEntryParameters
function to set parameters for a code replacement operator entry for uint8
addition that matches a cast-after-sum algorithm.
op_entry = RTW.TflCOperationEntry; op_entry.setTflCOperationEntryParameters( ... 'Key', 'RTW_OP_ADD', ... 'EntryInfoAlgorithm', 'RTW_CAST_AFTER_OP', ... 'Priority', 90, ... 'SaturationMode', 'RTW_SATURATE_UNSPECIFIED', ... 'RoundingModes', {'RTW_ROUND_UNSPECIFIED'}, ... 'ImplementationName', 'u8_add_u8_u8', ... 'ImplementationHeaderFile', 'u8_add_u8_u8.h', ... 'ImplementationSourceFile', 'u8_add_u8_u8.c');
This example shows how to use thesetTflCOperationEntryParameters
function to set parameters for a code replacement operator entry for fixed-pointint16
division. The table entry specifies a net scaling between the operator inputs and output to map a range of slope and bias values to a replacement operation.
op_entry = RTW.TflCOperationEntryGenerator_NetSlope; op_entry.setTflCOperationEntryParameters( ... 'Key', 'RTW_OP_DIV', ... 'Priority', 90, ... 'SaturationMode', 'RTW_WRAP_ON_OVERFLOW', ... 'RoundingModes', {'RTW_ROUND_CEILING'}, ... 'NetSlopeAdjustmentFactor', 1.0, ... 'NetFixedExponent', 0.0, ... 'ImplementationName', 's16_div_s16_s16', ... 'ImplementationHeaderFile', 's16_div_s16_s16.h', ... 'ImplementationSourceFile', 's16_div_s16_s16.c' );
This example shows how to use thesetTflCOperationEntryParameters
function to set parameters for a code replacement operator entry for fixed-pointuint16
addition that matches a cast-after-sum algorithm. The parameters 'SlopesMustBeTheSame' and 'MustHaveZeroNetBias' must be set totrue
to specify equal slope and zero net bias across operator inputs and output. This maps relative slope and bias values (rather than a specific slope and bias combination) to a replacement operation.
op_entry = RTW.TflCOperationEntryGenerator; op_entry.setTflCOperationEntryParameters( ... 'Key', 'RTW_OP_ADD', ... 'EntryInfoAlgorithm', 'RTW_CAST_AFTER_OP', ... 'Priority', 90, ... 'SaturationMode', 'RTW_WRAP_ON_OVERFLOW', ... 'RoundingModes', {'RTW_ROUND_UNSPECIFIED'}, ... 'SlopesMustBeTheSame', true, ... 'MustHaveZeroNetBias', true, ... 'ImplementationName', 'u16_add_SameSlopeZeroBias', ... 'ImplementationHeaderFile', 'u16_add_SameSlopeZeroBias.h', ... 'ImplementationSourceFile', 'u16_add_SameSlopeZeroBias.c');
Input Arguments
The hEntry
is a handle to a code replacement table entry previously returned by one of the class instantiations in the table.
Class Instantiation | Support |
---|---|
hEntry = RTW.TflCOperationEntry; | Supports operator replacement. |
hEntry = RTW.TflCOperationEntryGenerator; | Provides parameters for fixed-point addition and subtraction that are not available inRTW.TflCOperationEntry (SlopesMustBeTheSame andMustHaveZeroNetBias). |
hEntry = RTW.TflCOperationEntryGenerator_NetSlope; | Provides net slope parameters for fixed-point multiplication and division that are not available inRTW.TflCOperationEntry (NetSlopeAdjustmentFactor andNetFixedExponent). |
hEntry = RTW.TflBlasEntryGenerator; | Supports replacement of nonscalar operators with MathWorks BLAS functions. |
hEntry = RTW.TflCBlasEntryGenerator; | Supports replacement of nonscalar operators with ANSI®/ISO® C BLAS functions. |
hEntry =MyCustomOperationEntry; (where_MyCustomOperationEntry_ is a class derived fromRTW.TflCOperationEntry) | Supports operator replacement using custom code replacement table entries. |
If you want to specify SlopesMustBeTheSame
orMustHaveZeroNetBias
for your operator entry, instantiate your table entry using_`hEntry`_ = RTW.TflCOperationEntryGenerator
rather than_`hEntry`_ = RTW.TflCOperationEntry
. If you want to useNetSlopeAdjustmentFactor
andNetFixedExponent
, instantiate your table entry by using _`hEntry`_ = RTW.TflCOperationEntryGenerator_NetSlope
.
Example: op_entry
Example: 'Key','RTW_OP_ADD'
Name-Value Arguments
Specify optional pairs of arguments asName1=Value1,...,NameN=ValueN
, where Name
is the argument name and Value
is the corresponding value. Name-value arguments must appear after other arguments, but the order of the pairs does not matter.
Before R2021a, use commas to separate each name and value, and enclose Name
in quotes.
Example: 'Key','RTW_OP_ADD'
The AcceptExprInput
value flags the code generator that the implementation function described by this entry accepts expression inputs. The default value is true
if ImplType
equals FCN_IMPL_FUNCT
and false
if ImplType
equalsFCN_IMPL_MACRO
.
If the value is true
, expression inputs are integrated into the generated code in a form similar to this form:
rtY.Out1 = myAdd(rtU.In1, rtU.In2 * rtU.In3);
If the value is false
, a temporary variable is generated for the expression input:
real_T temp;
temp = rtU.In2 * rtU.In3; rtY.Out1 = myAdd(rtU.In1, temp);
Example: 'AcceptExprInput',true
The AdditionalHeaderFiles
value specifies additional header files for a code replacement table entry. The character vectors or string array can include tokens. For example, in the token $mytoken$
, mytoken
is a variable defined as a character vector or string scalar in the MATLAB® workspace or as a MATLAB function in the search path that returns a character vector.
Example: 'AdditionalHeaderFiles',{}
The AdditionalIncludePaths
value specifies the full path of additional include paths for a code replacement entry. The character vectors or string array can include tokens. For example, in the token $mytoken$
, mytoken
is a variable defined as a character vector or string scalar in the MATLAB workspace or as a MATLAB function in the search path that returns a character vector.
Example: 'AdditionalIncludePaths',{}
The AdditionalLinkObjs
value specifies additional link objects for a code replacement table entry. The character vectors or string array can include tokens. For example, in the token $mytoken$
, mytoken
is a variable defined as a character vector or string scalar in the MATLAB workspace or as a MATLAB function in the search path that returns a character vector.
Example: 'AdditionalLinkObjs',{}
The AdditionalLinkObjsPaths
value specifies the full path of additional link object paths for a code replacement entry. The character vectors or string array can include tokens. For example, in the token $mytoken$
,mytoken
is a variable defined as a character vector or string scalar in the MATLAB workspace or as a MATLAB function in the search path that returns a character vector.
Example: 'AdditionalLinkObjsPaths',{}
The AdditionalSourceFiles
value specifies additional source files for a code replacement table entry. The character vectors or string array can include tokens. For example, in the token $mytoken$
, mytoken
is a variable defined as a character vector or string scalar in the MATLAB workspace or as a MATLAB function in the search path that returns a character vector.
Example: 'AdditionalSourceFiles',{}
The AdditionalSourcePaths
value specifies the full path of additional source paths for a code replacement entry. The character vectors or string array can include tokens. For example, in the token $mytoken$
, mytoken
is a variable defined as a character vector or string scalar in the MATLAB workspace or as a MATLAB function in the search path that returns a character vector.
Example: 'AdditionalSourcePaths',{}
The AdditionalCompileFlags
value specifies additional flags required to compile the source files defined for a code replacement table entry.
Example: 'AdditionalCompileFlags',{}
The AdditionalLinkFlags
value specifies additional flags required to link the compiled files for a code replacement table entry.
Example: 'AdditionalLinkFlags',{}
The AllowShapeAgnosticMatch
value enables code replacement match based on total number of elements rather than specific matrix shape for matrices that are contiguously allocated in memory. For more information, see Allow Shape Agnostic Match.
Example: 'AllowShapeAgnosticMatch', false
The ArrayLayout
value specifies the order of array elements in memory supported by the replacement implementation. By default, the replacement implementation supports column-major data layout. For ROW-MAJOR
, the replacement implementation supports row-major data layout. For COLUMN_AND_ROW
, the replacement implementation supports column-major and row-major data layouts.
Example: 'ArrayLayout','ROW_MAJOR'
The EntryInfoAlgorithm
value specifies the algorithm for the specified math function that must be matched for operator replacement to occur. Code replacement libraries support replacement based on the algorithm for math operationsRTW_OP_ADD
and RTW_OP_MINUS
. Valid arguments for the supported operations are listed in the table. The arguments have the same meaning for both operations.
Argument | Meaning |
---|---|
RTW_CAST_BEFORE_OP | Before performing the operation, type cast input values to the output data type. If the output type cannot exactly represent the input values, losses can occur as a result of the cast to the output type. Additional loss can occur when the result of the operation is cast to the final output type. |
RTW_CAST_AFTER_OP | Compute the ideal result of the operation of inputs. Then, type cast the result to the output data type. Loss occurs during the type cast. This algorithm behaves similarly to the C language except when the signedness of the operands does not match. For example, when you add a signed long operation to an unsigned long operand, standard C language rules convert the signed long operand to an unsigned long operand. The result is a value that is not ideal. |
Example: 'EntryInfoAlgorithm','RTW_CAST_AFTER_OP'
The GenCallback
specifies a callback that follows code generation. If you specify'RTW.copyFileToBuildDir'
, and if this operation entry is matched and used, the code generator calls functionRTW.copyFileToBuildDir
after code generation. This callback function copies additional header, source, or object files that you have specified for this entry to the build folder.
Example: 'GenCallback','RTW.copyFileToBuildDir'
The ImplementationHeaderFile
value specifies the name of the header file that declares the implementation function. The character vector or string scalar can include tokens. For example, in the token $mytoken$
,mytoken
is a variable defined as a character vector or string scalar in the MATLAB workspace or as a MATLAB function in the search path that returns a character vector.
Example: 'ImplementationHeaderFile','s32_mul.h'
The ImplementationHeaderPath
value specifies the full path to the implementation header file. The character vector or string scalar can include tokens. For example, in the token$mytoken$
, mytoken
is a variable defined as a character vector or string scalar in the MATLAB workspace or as a MATLAB function in the search path that returns a character vector.
Example: 'ImplementationHeaderPath', fullfile('$MATLAB_ROOT', 'crl')
The ImplementationName
value specifies the name of the implementation function, which can match or differ from theKey
name.
Example: 'ImplementationName','s32_mul_s32_s32_sat'
The ImplementationSourceFile
value specifies the name of the implementation source file. The character vector or string scalar can include tokens. For example, in the token$mytoken$
, mytoken
is a variable defined as a character vector or string scalar in the MATLAB workspace or as a MATLAB function in the search path that returns a character vector.
Example: 'ImplementationSourceFile','s32_mul.c'
The ImplementationSourcePath
value specifies the full path to the implementation source file. The character vector or string scalar can include tokens. For example, in the token$mytoken$
, mytoken
is a variable defined as a character vector or string scalar in the MATLAB workspace or as a MATLAB function in the search path that returns a character vector.
Example: 'ImplementationSourcePath', fullfile('$MATLAB_ROOT', 'crl')
The ImplType
value specifies the type of table entry. Use FCN_IMPL_FUNCT
for function orFCN_IMPL_MACRO
for macro.
Example: 'ImplType','FCN_IMPL_FUNCT'
The MustHaveZeroNetBias
value specifies whether a replacement match requires that the net bias for conceptual arguments of Add/Minus entries is zero. For the Add/Minus of fixed-point operator inputs and output this parameter must be set totrue
. Instantiate the entry by using_`hEntry`_ = RTW.TflCOperationEntryGenerator
rather than_`hEntry`_ = RTW.TflCOperationEntry
.
For Mul/Div/MulDiv/Shift/Cast entries:
- The code generator ignores the value of this argument.
- The bias of the conceptual arguments for Mul/Div/MulDiv/Shift/Cast entries must be zero for a replacement match to occur.
Example: 'MustHaveZeroNetBias',true
The NetSlopeAdjustmentFactor
value specifies the fixed exponent (E) part of the net slope (F2E, for example, -3.0) for fixed-point conceptual arguments required for a replacement match to occur for Mul/Div/MulDiv/Shift/Cast entries. Instantiate an entry by using_`hEntry`_ = RTW.TflCOperationEntryGenerator_NetSlope
rather than_`hEntry`_ = RTW.TflCOperationEntry
.
For Add/Minus entries:
- The code generator ignores the value of this argument.
- The slope adjustment factor part of the net slope of the conceptual arguments for Add/Minus entries must be zero for a replacement match to occur.
Example: 'NetFixedExponent', -3.0
The NetSlopeAdjustmentFactor
value specifies the slope adjustment part of the net slope (F2E, for example, 1.0) for fixed-point conceptual arguments required for a replacement match to occur for Mul/Div/MulDiv/Shift/Cast entries. Instantiate an entry by using_`hEntry`_ = RTW.TflCOperationEntryGenerator_NetSlope
rather than_`hEntry`_ = RTW.TflCOperationEntry
.
For Add/Minus entries:
- The code generator ignores the value of this argument.
- The slope adjustment factor part of the net slope of the conceptual arguments for Add/Minus entries must be zero for a replacement match to occur.
Example: 'NetSlopeAdjustmentFactor',1.5
The Priority
value specifies the search priority of the operation entry, relative to other entries of the same operation name and conceptual argument list within this table. Highest priority is 0, and lowest priority is 100. If the table provides two implementations for an operation, the implementation with the higher priority shadows the one with the lower priority.
Example: 'Priority',100
The RoundingModes
value specifies one or more rounding modes supported by the implementation function.
Example: 'RoundingModes',{'RTW_ROUND_UNSPECIFIED'}
The SaturationMode
value specifies the saturation mode supported by the implementation function.
Example: 'SaturationMode','RTW_SATURATE_UNSPECIFIED'
The SideEffects
value flags the code generator not to optimize away the implementation function described by this entry. This parameter applies to implementation functions that return void
but are not to be optimized away, such as a memcpy
implementation or an implementation function that accesses global memory values. For those implementation functions only, you must include this parameter and specify the valuetrue
.
Example: 'SideEffects',false
The SlopesMustBeTheSame
value specifies whether a replacement match requires that the slope is the same for conceptual arguments of Mul/Div/MulDiv/Shift/Cast entries. Instantiate the entry by using _`hEntry`_ = RTW.TflCOperationEntryGenerator
rather than_`hEntry`_ = RTW.TflCOperationEntry
.
For Add/Minus entries:
- The code generator ignores the value of this argument.
- This parameter must be set to
true
. When set totrue
, the slopes of the conceptual arguments are equal for a replacement match to occur.
Example: 'SlopesMustBeTheSame',true
The StoreFcnReturnInLocalVar
value flags the code generator that the return value of the implementation function described by this entry must be stored in a local variable regardless of other expression folding settings. If the value isfalse
, other expression folding settings determine whether the return value is folded. Storing function returns in a local variable can increase the clarity of generated code. This example shows code generated with expression folding:
void sw_step(void)
{
if (ssub(sadd(sw_U.In1, sw_U.In2), sw_U.In3) <=
smul(ssub(sw_U.In4, sw_U.In5),sw_U.In6)) {
sw_Y.Out1 = sw_U.In7;
} else {
sw_Y.Out1 = sw_U.In8;
}
}
With StoreFcnReturnInLocalVar
set totrue
, the generated code is potentially easier to understand and debug:
void sw_step(void)
{
real32_T rtb_Switch;
real32_T hoistedExpr;
......
rtb_Switch = sadd(sw_U.In1, sw_U.In2);
rtb_Switch = ssub(rtb_Switch, sw_U.In3);
hoistedExpr = ssub(sw_U.In4, sw_U.In5);
hoistedExpr = smul(hoistedExpr, sw_U.In6);
if (rtb_Switch <= hoistedExpr) {
sw_Y.Out1 = sw_U.In7;
} else {
sw_Y.Out1 = sw_U.In8;
}
}
Example: 'StoreFcnReturnInLocalVar',false
Version History
Introduced in R2007b