Using the Vec class

This tutorial provided a basic overview of how to use the Vec class. For more advanced usage and additional features, please refer to the Vector class API in vec.hpp.

The Vec class enables efficient parallelization of computations on multiple data elements simultaneously using SIMD (Single Instruction, Multiple Data) instructions. This tutorial will guide you through the usage of the Vec class, including its initialization, arithmetic operations, comparison operators, and other functionalities.

Initializing Vectors

You can initialize vectors using various constructors provided by the Vec class. Here are some common initialization methods:

  • Zero-Initialized Vector:

    Vec<double> zeroVec = Vec<double>::Zero();

  • Initializing with a Scalar Value:

    Vec<double> scalarVec(5.0); // All elements initialized with 5.0

  • Initializing with Multiple Scalar Values:

    Vec<double> multiVec(1.0, 2.0, 3.0, 4.0); // Elements: {1.0, 2.0, 3.0, 4.0}

  • Loading from Memory:

    double data[4] = {1.0, 2.0, 3.0, 4.0};
Vec<double> loadedVec = Vec<double>::Load(data); // Load from unaligned memory

    double alignedData[4] __attribute__((aligned(64))) = {1.0, 2.0, 3.0, 4.0};
Vec<double> alignedVec = Vec<double>::LoadAligned(alignedData); // Load from aligned memory

Accessing Vector Elements

You can access individual elements of the vector using the subscript operator []:

Vec<double> vec(1.0, 2.0, 3.0);
double secondElement = vec[1]; // Accessing the second element (index 1)

Arithmetic Operations

The Vec class supports various arithmetic operations such as addition, subtraction, multiplication, and division:

Vec<double> a(1.0, 2.0, 3.0, 4.0);
Vec<double> b(5.0, 6.0, 7.0, 8.0);

Vec<double> sum = a + b; // Element-wise addition
Vec<double> difference = a - b; // Element-wise subtraction
Vec<double> product = a * b; // Element-wise multiplication
Vec<double> quotient = a / b; // Element-wise division

Comparison Operations

You can compare vectors using comparison operators, which return a mask indicating the comparison result. The mask can then be used in other operations.

Vec<double> a(1.0, 2.0, 3.0, 4.0);
Vec<double> b(5.0, 2.0, 7.0, -1.0);

auto lessThanMask = a < b; // Element-wise less-than comparison
auto greaterThanMask = a > b; // Element-wise greater-than comparison

Other Operations

  • Store to Memory:

    double result[4];
sum.Store(result); // Store vector data to unaligned memory

    double alignedResult[4] __attribute__((aligned(64)));
sum.StoreAligned(alignedResult); // Store vector data to aligned memory

  • Other Mathematical Functions:

    Additional mathematical functions such as square root (approx_sqrt), reciprocal square root (approx_rsqrt), sine and cosine (sincos), and exponential (exp) are provided.

  • Printing Vectors:

    Vectors can be printed using the operator<<:

    Vec<double,4> vec(1.0, 2.0, 3.0, 4.0);
std::cout << vec << std::endl; // Output: 1, 2, 3, 4