triton.language.tensor

class triton.language.tensor(self, handle, type: dtype)

Represents an N-dimensional array of values or pointers.

tensor is the fundamental data structure in Triton programs. Most functions in triton.language operate on and return tensors.

Most of the named member functions here are duplicates of the free functions in triton.language. For example, triton.language.sqrt(x) is equivalent to x.sqrt().

tensor also defines most of the magic/dunder methods, so you can write x+y, x << 2, etc.

Constructors

__init__(self, handle, type: dtype)

Not called by user code.

Methods

__init__(self, handle, type)	Not called by user code.
abs(self[, _semantic])	Forwards to abs() free function
advance(self, offsets[, _semantic])	Forwards to advance() free function
argmax(input, axis[, tie_break_left, keep_dims])	Returns the maximum index of all elements in the input tensor along the provided axis
argmin(input, axis[, tie_break_left, keep_dims])	Returns the minimum index of all elements in the input tensor along the provided axis
associative_scan(self, axis, combine_fn[, ...])	Forwards to associative_scan() free function
atomic_add(self, val[, mask, sem, scope, ...])	Forwards to atomic_add() free function
atomic_and(self, val[, mask, sem, scope, ...])	Forwards to atomic_and() free function
atomic_cas(self, cmp, val[, sem, scope, ...])	Forwards to atomic_cas() free function
atomic_max(self, val[, mask, sem, scope, ...])	Forwards to atomic_max() free function
atomic_min(self, val[, mask, sem, scope, ...])	Forwards to atomic_min() free function
atomic_or(self, val[, mask, sem, scope, ...])	Forwards to atomic_or() free function
atomic_xchg(self, val[, mask, sem, scope, ...])	Forwards to atomic_xchg() free function
atomic_xor(self, val[, mask, sem, scope, ...])	Forwards to atomic_xor() free function
broadcast_to(self, *shape[, _semantic])	Forwards to broadcast_to() free function
cast(self, dtype[, fp_downcast_rounding, ...])	Forwards to cast() free function
cdiv(x, div)	Computes the ceiling division of x by div
ceil(self[, _semantic])	Forwards to ceil() free function
cos(self[, _semantic])	Forwards to cos() free function
cumprod(input[, axis, reverse])	Returns the cumprod of all elements in the input tensor along the provided axis
cumsum(input[, axis, reverse, dtype])	Returns the cumsum of all elements in the input tensor along the provided axis
erf(self[, _semantic])	Forwards to erf() free function
exp(self[, _semantic])	Forwards to exp() free function
exp2(self[, _semantic])	Forwards to exp2() free function
expand_dims(self, axis[, _semantic])	Forwards to expand_dims() free function
flip(x[, dim])	Flips a tensor x along the dimension dim.
floor(self[, _semantic])	Forwards to floor() free function
gather(self, index, axis[, _semantic])	Forwards to gather() free function
histogram(self, num_bins[, mask, _semantic, ...])	Forwards to histogram() free function
item(self[, _semantic, _generator])	Forwards to item() free function
log(self[, _semantic])	Forwards to log() free function
log2(self[, _semantic])	Forwards to log2() free function
logical_and(self, other[, _semantic])
logical_or(self, other[, _semantic])
max(input[, axis, return_indices, ...])	Returns the maximum of all elements in the input tensor along the provided axis
min(input[, axis, return_indices, ...])	Returns the minimum of all elements in the input tensor along the provided axis
permute(self, *dims[, _semantic])	Forwards to permute() free function
ravel(x[, can_reorder])	Returns a contiguous flattened view of x.
reduce(self, axis, combine_fn[, keep_dims, ...])	Forwards to reduce() free function
reduce_or(input, axis[, keep_dims])	Returns the reduce_of of all elements in the input tensor along the provided axis
reshape(self, *shape[, can_reorder, ...])	Forwards to reshape() free function
rsqrt(self[, _semantic])	Forwards to rsqrt() free function
sigmoid(x)	Computes the element-wise sigmoid of x.
sin(self[, _semantic])	Forwards to sin() free function
softmax(x[, dim, keep_dims, ieee_rounding])	Computes the element-wise softmax of x.
sort(self[, dim, descending])
split(self[, _semantic, _generator])	Forwards to split() free function
sqrt(self[, _semantic])	Forwards to sqrt() free function
sqrt_rn(self[, _semantic])	Forwards to sqrt_rn() free function
store(self, value[, mask, boundary_check, ...])	Forwards to store() free function
sum(input[, axis, keep_dims, dtype])	Returns the sum of all elements in the input tensor along the provided axis
to(self, dtype[, fp_downcast_rounding, ...])	Alias for tensor.cast().
trans(self, *dims[, _semantic])	Forwards to trans() free function
view(self, *shape[, _semantic])	Forwards to view() free function
xor_sum(input[, axis, keep_dims])	Returns the xor sum of all elements in the input tensor along the provided axis

Attributes

T	Transposes a 2D tensor.
type