Creates a ClTensor from a provided buffer and shape. This should primarily be used by internal methods that need to create ClTensors or pass to lower level libraries

The generic type of the ClTensor must be specified, as LibCL::ClMem does not store this information

Arguments

buffer : LibCL::ClMem Memory buffer for a ClTensor shape : Array(Int) Size of the ClTensor in each dimension

Examples

t = ClTensor(Float32).new(buffer, [3, 3, 2])

Creates a ClTensor from a provided shape, initializing the ClTensor's buffer with a value specified by the user

The generic type of the ClTensor is inferred from the provided value

Arguments

shape : Array(Int) Size of ClTensor in each dimension value : T Value to fill the ClTensor

Examples

t = ClTensor(Float32).new([3, 3, 2], 3.5)

Creates a ClTensor from a provided shape, initializing the ClTensor's buffer without filling it with data

When creating a ClTensor from a shape alone, the generic type must be specified

Arguments

shape : Array(Int) Size of ClTensor in each dimension

Examples

t = ClTensor(Float32).new([3, 3, 2])

Creates a ClTensor of a provided shape, filled with a value. The generic type must be specified.

Arguments

shape Shape of returned ClTensor

Examples

t = ClTensor(Float32).full([3], 3.5)

Creates a ClTensor filled with a value, sharing the shape of another provided Tensor or ClTensor

Arguments

t Item to use for output shape

Examples

t = Tensor.new([3]) &.to_f
u = ClTensor(Float32).full_like(t, 3.5)

Creates a ClTensor of a provided shape, filled with 1. The generic type must be specified.

Arguments

shape Shape of returned ClTensor

Examples

t = ClTensor(Float32).ones([3])

Creates a ClTensor filled with 1, sharing the shape of another provided Tensor or ClTensor

Arguments

t Item to use for output shape

Examples

t = Tensor.new([3]) &.to_f
u = ClTensor(Float32).ones_like(t)

Creates a ClTensor of a provided shape, filled with 0. The generic type must be specified.

Arguments

shape Shape of returned ClTensor

Examples

t = ClTensor(Float32).zeros([3])

Creates a ClTensor filled with 0, sharing the shape of another provided Tensor or ClTensor

Arguments

t Item to use for output shape

Examples

t = Tensor.new([3]) &.to_f
u = ClTensor(Float32).zeros_like(t)

ASUM sums the absolute values of the elements of a vector

Arguments

Examples

a = ClTensor(Flaot32).new([3, 3, 2])
a.asum

AXPY constant times a vector plus a vector. uses unrolled loops for increments equal to one.

Arguments

a : Number Scalar alpha c : ClTensor Argument to add

Examples

a = ClTensor(Float32).ones([3])
b = ClTensor(Float32).ones([3])
c = 1.45
a.axpy!(c, b)

AXPY constant times a vector plus a vector. uses unrolled loops for increments equal to one.

Arguments

a : Number Scalar alpha c : ClTensor Argument to add

Examples

a = ClTensor(Float32).ones([3])
b = ClTensor(Float32).ones([3])
c = 1.45
a.axpy!(c, b)

COPY copies a vector, x, to a vector, y. uses unrolled loops for increments equal to 1.

Arguments

Examples

a = ClTensor(Float32).ones([3, 3, 2])
b = a.copy
b.shape # => [3, 3, 2]

Move a ClTensor from an OpenCL buffer to a CPU Tensor backed by a Pointer. This operation is slow, and should not be used frequently. Operations are most efficient when done exclusively on a single architecture

This method always blocks on write

Arguments

Examples

t = ClTensor(Float32).new([5], 2.2)
t.cpu # => [2.2, 2.2, 2.2, 2.2, 2.2]

DOT forms the dot product of two vectors. uses unrolled loops for increments equal to one.

This method treats all inputs as though they were one-dimensional

Arguments

c : ClTensor Argument to dot against self

Examples

a = ClTensor(Float32).ones([3])
b = ClTensor(Float32).ones([3])
c = a.dot(b)

GEMM performs one of the matrix-matrix operations

C := alpha*op( A )op( B ) + betaC,

where op( X ) is one of

op( X ) = X or op( X ) = X**T,

alpha and beta are scalars, and A, B and C are matrices, with op( A ) an m by k matrix, op( B ) a k by n matrix and C an m by n matrix.

Arguments

*b* : ClTensor
  Second operand for the matrix multiplication

Examples

a = ClTensor(Float32).ones([3, 3])
b = a.copy
puts a.gemm(b).cpu

# [[3, 3, 3],
#  [3, 3, 3],
#  [3, 3, 3]]

NRM2 returns the euclidean norm of a vector via the function name, so that

DNRM2 := sqrt( x'*x )

Treats ClTensor as though they were one-dimensional

Arguments

Examples

a = ClTensor(Float32).new([3])
a.nrm2

SCAL scales a vector by a constant. Uses unrolled loops for increment equal to 1.

Arguments

a : Number Scalar alpha

Examples

a = ClTensor(Float32).ones([3])
a.scal!(3)

SCAL scales a vector by a constant. Uses unrolled loops for increment equal to 1.

Arguments

a : Number Scalar alpha

Examples

a = ClTensor(Float32).ones([3])
a.scal!(3)