feat: add C ndarray implementation for blas/base/zswap

PR-URL: #3080
Ref: #2039
Co-authored-by: Athan Reines <[email protected]>
Reviewed-by: Athan Reines <[email protected]> 
Signed-off-by: Athan Reines <[email protected]>

Author: aman-095

lib/node_modules/@stdlib/blas/base/zswap/README.md

@@ -265,6 +265,142 @@ console.log( y.get( y.length-1 ).toString() );
 
 <!-- /.examples -->
 
+<!-- C interface documentation. -->
+
+* * *
+
+<section class="c">
+
+## C APIs
+
+<!-- Section to include introductory text. Make sure to keep an empty line after the intro `section` element and another before the `/section` close. -->
+
+<section class="intro">
+
+</section>
+
+<!-- /.intro -->
+
+<!-- C usage documentation. -->
+
+<section class="usage">
+
+### Usage
+
+```c
+#include "stdlib/blas/base/zswap.h"
+```
+
+#### c_zswap( N, \*X, strideX, \*Y, strideY )
+
+Interchanges two complex double-precision floating-point vectors.
+
+```c
+double x[] = { 1.0, 2.0, 3.0, 4.0 }; // interleaved real and imaginary components
+double y[] = { 5.0, 6.0, 7.0, 8.0 };
+
+c_zswap( 2, (void *)x, 1, (void *)y, 1 );
+```
+
+The function accepts the following arguments:
+
+-   **N**: `[in] CBLAS_INT` number of indexed elements.
+-   **X**: `[inout] void*` first input array.
+-   **strideX**: `[in] CBLAS_INT` index increment for `X`.
+-   **Y**: `[inout] void*` second input array.
+-   **strideY**: `[in] CBLAS_INT` index increment for `Y`.
+
+```c
+void c_zswap( const CBLAS_INT N, void *X, const CBLAS_INT strideX, void *Y, const CBLAS_INT strideY );
+```
+
+#### c_zswap_ndarray( N, \*X, strideX, offsetX, \*Y, strideY, offsetY )
+
+Interchanges two complex double-precision floating-point vectors using alternative indexing semantics.
+
+```c
+double x[] = { 1.0, 2.0, 3.0, 4.0 }; // interleaved real and imaginary components
+double y[] = { 5.0, 6.0, 7.0, 8.0 };
+
+c_zswap_ndarray( 2, (void *)x, 1, 0, (void *)y, 1, 0 );
+```
+
+The function accepts the following arguments:
+
+-   **N**: `[in] CBLAS_INT` number of indexed elements.
+-   **X**: `[inout] void*` first input array.
+-   **strideX**: `[in] CBLAS_INT` index increment for `X`.
+-   **offsetX**: `[in] CBLAS_INT` starting index for `X`.
+-   **Y**: `[inout] void*` second input array.
+-   **strideY**: `[in] CBLAS_INT` index increment for `Y`.
+-   **offsetY**: `[in] CBLAS_INT` starting index for `Y`.
+
+```c
+void c_zswap_ndarray( const CBLAS_INT N, void *X, const CBLAS_INT strideX, const CBLAS_INT offsetX, void *Y, const CBLAS_INT strideY, const CBLAS_INT offsetY );
+```
+
+</section>
+
+<!-- /.usage -->
+
+<!-- C API usage notes. Make sure to keep an empty line after the `section` element and another before the `/section` close. -->
+
+<section class="notes">
+
+</section>
+
+<!-- /.notes -->
+
+<!-- C API usage examples. -->
+
+<section class="examples">
+
+### Examples
+
+```c
+#include "stdlib/blas/base/zswap.h"
+#include <stdio.h>
+
+int main( void ) {
+    // Create strided arrays:
+    double x[] = { 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0 };
+    double y[] = { 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 };
+
+    // Specify the number of elements:
+    const int N = 4;
+
+    // Specify stride lengths:
+    const int strideX = 1;
+    const int strideY = -1;
+
+    // Swap elements:
+    c_zswap( N, (void *)x, strideX, (void *)y, strideY );
+
+    // Print the result:
+    for ( int i = 0; i < N; i++ ) {
+        printf( "x[ %i ] = %lf + %lfj\n", i, x[ i*2 ], x[ (i*2)+1 ] );
+        printf( "y[ %i ] = %lf + %lfj\n", i, y[ i*2 ], y[ (i*2)+1 ] );
+    }
+
+    // Swap elements using alternative indexing semantics:
+    c_zswap_ndarray( N, (void *)x, -strideX, N-1, (void *)y, strideY, N-1 );
+
+    // Print the result:
+    for ( int i = 0; i < N; i++ ) {
+        printf( "x[ %i ] = %lf + %lfj\n", i, x[ i*2 ], x[ (i*2)+1 ] );
+        printf( "y[ %i ] = %lf + %lfj\n", i, y[ i*2 ], y[ (i*2)+1 ] );
+    }
+}
+```
+
+</section>
+
+<!-- /.examples -->
+
+</section>
+
+<!-- /.c -->
+
 <!-- Section for related `stdlib` packages. Do not manually edit this section, as it is automatically populated. -->
 
 <section class="related">

lib/node_modules/@stdlib/blas/base/zswap/benchmark/c/benchmark.length.c

@@ -94,7 +94,7 @@ static double rand_double( void ) {
 * @param len          array length
 * @return elapsed time in seconds
 */
-static double benchmark( int iterations, int len ) {
+static double benchmark1( int iterations, int len ) {
 	double elapsed;
 	double *x;
 	double *y;
@@ -127,6 +127,46 @@ static double benchmark( int iterations, int len ) {
 	return elapsed;
 }
 
+/**
+* Runs a benchmark.
+*
+* @param iterations   number of iterations
+* @param len          array length
+* @return elapsed time in seconds
+*/
+static double benchmark2( int iterations, int len ) {
+	double elapsed;
+	double *x;
+	double *y;
+	double t;
+	int i;
+
+	x = (double *) malloc( len*2 * sizeof( double ) );
+	y = (double *) malloc( len*2 * sizeof( double ) );
+	for ( i = 0; i < len; i++ ) {
+		x[ i ] = ( rand_double()*10000.0 ) - 5000.0;
+		x[ i+1 ] = ( rand_double()*10000.0 ) - 5000.0;
+		y[ i ] = 0.0;
+		y[ i+1 ] = 0.0;
+	}
+	t = tic();
+	for ( i = 0; i < iterations; i++ ) {
+		c_zswap_ndarray( len, (void *)x, 1, 0, (void *)y, 1, 0 );
+		if ( y[ 0 ] != y[ 0 ] ) {
+			printf( "should not return NaN\n" );
+			break;
+		}
+	}
+	elapsed = tic() - t;
+	if ( y[ 0 ] != y[ 0 ] ) {
+		printf( "should not return NaN\n" );
+	}
+	free( x );
+	free( y );
+
+	return elapsed;
+}
+
 /**
 * Main execution sequence.
 */
@@ -149,7 +189,14 @@ int main( void ) {
 		for ( j = 0; j < REPEATS; j++ ) {
 			count += 1;
 			printf( "# c::%s:len=%d\n", NAME, len );
-			elapsed = benchmark( iter, len );
+			elapsed = benchmark1( iter, len );
+			print_results( iter, elapsed );
+			printf( "ok %d benchmark finished\n", count );
+		}
+		for ( j = 0; j < REPEATS; j++ ) {
+			count += 1;
+			printf( "# c::%s:ndarray:len=%d\n", NAME, len );
+			elapsed = benchmark2( iter, len );
 			print_results( iter, elapsed );
 			printf( "ok %d benchmark finished\n", count );
 		}

lib/node_modules/@stdlib/blas/base/zswap/examples/c/example.c

@@ -32,4 +32,13 @@ int main( void ) {
 		printf( "x[ %i ] = %f + %fj\n", i, x[ i*2 ], x[ ( i*2 )+1 ] );
 		printf( "y[ %i ] = %f + %fj\n", i, y[ i*2 ], y[ ( i*2 )+1 ] );
 	}
+
+	// Swap elements using alternative indexing semantics:
+	c_zswap_ndarray( 4, (void *)x, -1, 3, (void *)y, -1, 3 );
+
+	// Print the result:
+	for ( int i = 0; i < 4; i++ ) {
+		printf( "x[ %i ] = %f + %fj\n", i, x[ i*2 ], x[ ( i*2 )+1 ] );
+		printf( "y[ %i ] = %f + %fj\n", i, y[ i*2 ], y[ ( i*2 )+1 ] );
+	}
 }

lib/node_modules/@stdlib/blas/base/zswap/include/stdlib/blas/base/zswap.h

@@ -22,6 +22,8 @@
 #ifndef ZSWAP_H
 #define ZSWAP_H
 
+#include "stdlib/blas/base/shared.h"
+
 /*
 * If C++, prevent name mangling so that the compiler emits a binary file having undecorated names, thus mirroring the behavior of a C compiler.
 */
@@ -32,7 +34,12 @@ extern "C" {
 /**
 * Interchanges two complex double-precision floating-point vectors.
 */
-void c_zswap( const int N, void *X, const int strideX, void *Y, const int strideY );
+void API_SUFFIX(c_zswap)( const CBLAS_INT N, void *X, const CBLAS_INT strideX, void *Y, const CBLAS_INT strideY );
+
+/**
+* Interchanges two complex double-precision floating-point vectors using alternative indexing semantics.
+*/
+void API_SUFFIX(c_zswap_ndarray)( const CBLAS_INT N, void *X, const CBLAS_INT strideX, const CBLAS_INT offsetX, void *Y, const CBLAS_INT strideY, const CBLAS_INT offsetY );
 
 #ifdef __cplusplus
 }

lib/node_modules/@stdlib/blas/base/zswap/include/stdlib/blas/base/zswap_cblas.h

@@ -22,6 +22,8 @@
 #ifndef ZSWAP_CBLAS_H
 #define ZSWAP_CBLAS_H
 
+#include "stdlib/blas/base/shared.h"
+
 /*
 * If C++, prevent name mangling so that the compiler emits a binary file having undecorated names, thus mirroring the behavior of a C compiler.
 */
@@ -32,7 +34,7 @@ extern "C" {
 /**
 * Interchanges two complex double-precision floating-point vectors.
 */
-void cblas_zswap( const int N, void *X, const int strideX, void *Y, const int strideY );
+void API_SUFFIX(cblas_zswap)( const CBLAS_INT N, void *X, const CBLAS_INT strideX, void *Y, const CBLAS_INT strideY );
 
 #ifdef __cplusplus
 }

lib/node_modules/@stdlib/blas/base/zswap/lib/ndarray.native.js

@@ -21,7 +21,6 @@
 // MODULES //
 
 var reinterpret = require( '@stdlib/strided/base/reinterpret-complex128' );
-var minViewBufferIndex = require( '@stdlib/strided/base/min-view-buffer-index' );
 var addon = require( './../src/addon.node' );
 
 
@@ -68,16 +67,10 @@ var addon = require( './../src/addon.node' );
 * // returns 8.0
 */
 function zswap( N, x, strideX, offsetX, y, strideY, offsetY ) {
-	var viewX;
-	var viewY;
+	var viewX = reinterpret( x, 0 );
+	var viewY = reinterpret( y, 0 );
 
-	offsetX = minViewBufferIndex( N, strideX, offsetX );
-	offsetY = minViewBufferIndex( N, strideY, offsetY );
-
-	viewX = reinterpret( x, offsetX );
-	viewY = reinterpret( y, offsetY );
-
-	addon( N, viewX, strideX, viewY, strideY );
+	addon.ndarray( N, viewX, strideX, offsetX, viewY, strideY, offsetY );
 	return y;
 }
 

lib/node_modules/@stdlib/blas/base/zswap/lib/zswap.js

@@ -20,7 +20,8 @@
 
 // MODULES //
 
-var reinterpret = require( '@stdlib/strided/base/reinterpret-complex128' );
+var stride2offset = require( '@stdlib/strided/base/stride2offset' );
+var ndarray = require( './ndarray.js' );
 
 
 // MAIN //
@@ -64,59 +65,9 @@ var reinterpret = require( '@stdlib/strided/base/reinterpret-complex128' );
 * // returns 8.0
 */
 function zswap( N, x, strideX, y, strideY ) {
-	var viewX;
-	var viewY;
-	var tmp;
-	var sx;
-	var sy;
-	var ix;
-	var iy;
-	var i;
-	var j;
-
-	if ( N <= 0 ) {
-		return y;
-	}
-	viewX = reinterpret( x, 0 );
-	viewY = reinterpret( y, 0 );
-	if ( strideX === 1 && strideY === 1 ) {
-		for ( i = 0; i < N*2; i += 2 ) {
-			tmp = viewX[ i ];
-			viewX[ i ] = viewY[ i ];
-			viewY[ i ] = tmp;
-
-			j = i + 1;
-			tmp = viewX[ j ];
-			viewX[ j ] = viewY[ j ];
-			viewY[ j ] = tmp;
-		}
-		return y;
-	}
-	if ( strideX < 0 ) {
-		ix = 2 * ( 1-N ) * strideX;
-	} else {
-		ix = 0;
-	}
-	if ( strideY < 0 ) {
-		iy = 2 * ( 1-N ) * strideY;
-	} else {
-		iy = 0;
-	}
-	sx = strideX * 2;
-	sy = strideY * 2;
-	for ( i = 0; i < N; i++ ) {
-		tmp = viewX[ ix ];
-		viewX[ ix ] = viewY[ iy ];
-		viewY[ iy ] = tmp;
-
-		tmp = viewX[ ix+1 ];
-		viewX[ ix+1 ] = viewY[ iy+1 ];
-		viewY[ iy+1 ] = tmp;
-
-		ix += sx;
-		iy += sy;
-	}
-	return y;
+	var ox = stride2offset( N, strideX );
+	var oy = stride2offset( N, strideY );
+	return ndarray( N, x, strideX, ox, y, strideY, oy );
 }