elementwise

elementwise[func: def[width: Int, rank: Int, alignment: Int = 1](IndexList[rank]) capturing -> None, simd_width: Int, *, use_blocking_impl: Bool = False, target: StringSlice[StaticConstantOrigin] = StringSlice("cpu"), _trace_description: StringSlice[StaticConstantOrigin] = StringSlice("elementwise")](shape: Int, ctx: Optional[DeviceContext] = None)

Executes func[width, rank](indices), possibly as sub-tasks, for a suitable combination of width and indices so as to cover shape. Returns when all sub-tasks have completed.

Parameters:

• ​func (def[width: Int, rank: Int, alignment: Int = 1](IndexList[rank]) capturing -> None): The body function.
• ​simd_width (Int): The SIMD vector width to use.
• ​use_blocking_impl (Bool): Do not invoke the function using asynchronous calls.
• ​target (StringSlice): The target to run on.
• ​_trace_description (StringSlice): Description of the trace.

Args:

• ​shape (Int): The shape of the buffer.
• ​ctx (Optional): The context to execute the work on.

Raises:

If the operation fails.

elementwise[rank: Int, //, func: def[width: Int, rank: Int, alignment: Int = 1](IndexList[rank]) capturing -> None, simd_width: Int, *, use_blocking_impl: Bool = False, target: StringSlice[StaticConstantOrigin] = StringSlice("cpu"), _trace_description: StringSlice[StaticConstantOrigin] = StringSlice("elementwise")](shape: IndexList[rank, element_type=shape.element_type], ctx: Optional[DeviceContext] = None)

Executes func[width, rank](indices), possibly as sub-tasks, for a suitable combination of width and indices so as to cover shape. Returns when all sub-tasks have completed.

Parameters:

• ​rank (Int): The rank of the buffer.
• ​func (def[width: Int, rank: Int, alignment: Int = 1](IndexList[rank]) capturing -> None): The body function.
• ​simd_width (Int): The SIMD vector width to use.
• ​use_blocking_impl (Bool): Do not invoke the function using asynchronous calls.
• ​target (StringSlice): The target to run on.
• ​_trace_description (StringSlice): Description of the trace.

Args:

• ​shape (IndexList): The shape of the buffer.
• ​ctx (Optional): The context to execute the work on.

Raises:

If the operation fails.

elementwise[func: def[width: Int, rank: Int, alignment: Int = 1](IndexList[rank]) capturing -> None, simd_width: Int, *, use_blocking_impl: Bool = False, target: StringSlice[StaticConstantOrigin] = StringSlice("cpu"), pdl_level: PDLLevel = PDLLevel(1), _trace_description: StringSlice[StaticConstantOrigin] = StringSlice("elementwise")](shape: Int, context: DeviceContext)

Executes func[width, rank](indices), possibly as sub-tasks, for a suitable combination of width and indices so as to cover shape. Returns when all sub-tasks have completed.

Parameters:

• ​func (def[width: Int, rank: Int, alignment: Int = 1](IndexList[rank]) capturing -> None): The body function.
• ​simd_width (Int): The SIMD vector width to use.
• ​use_blocking_impl (Bool): Do not invoke the function using asynchronous calls.
• ​target (StringSlice): The target to run on.
• ​pdl_level (PDLLevel): The PDL level controlling GPU kernel overlap behavior.
• ​_trace_description (StringSlice): Description of the trace.

Args:

• ​shape (Int): The shape of the buffer.
• ​context (DeviceContext): The device context to use.

Raises:

If the operation fails.

elementwise[rank: Int, //, func: def[width: Int, rank: Int, alignment: Int = 1](IndexList[rank]) capturing -> None, simd_width: Int, *, use_blocking_impl: Bool = False, target: StringSlice[StaticConstantOrigin] = StringSlice("cpu"), pdl_level: PDLLevel = PDLLevel(1), _trace_description: StringSlice[StaticConstantOrigin] = StringSlice("elementwise")](shape: IndexList[rank, element_type=shape.element_type], context: DeviceContext)

Executes func[width, rank](indices), possibly as sub-tasks, for a suitable combination of width and indices so as to cover shape. Returns when all sub-tasks have completed.

Parameters:

• ​rank (Int): The rank of the buffer.
• ​func (def[width: Int, rank: Int, alignment: Int = 1](IndexList[rank]) capturing -> None): The body function.
• ​simd_width (Int): The SIMD vector width to use.
• ​use_blocking_impl (Bool): Do not invoke the function using asynchronous calls.
• ​target (StringSlice): The target to run on.
• ​pdl_level (PDLLevel): The PDL level controlling GPU kernel overlap behavior.
• ​_trace_description (StringSlice): Description of the trace.

Args:

• ​shape (IndexList): The shape of the buffer.
• ​context (DeviceContext): The device context to use.

Raises:

If the operation fails.

elementwise[rank: Int, //, func: def[width: Int, rank: Int, alignment: Int = 1](IndexList[rank]) capturing -> None, simd_width: Int, *, use_blocking_impl: Bool = False, target: StringSlice[StaticConstantOrigin] = StringSlice("cpu"), pdl_level: PDLLevel = PDLLevel(1), _trace_description: StringSlice[StaticConstantOrigin] = StringSlice("elementwise")](shape: IndexList[rank, element_type=shape.element_type], context: DeviceContextPtr)

Executes func[width, rank](indices), possibly as sub-tasks, for a suitable combination of width and indices so as to cover shape. Returns when all sub-tasks have completed.

Parameters:

• ​rank (Int): The rank of the buffer.
• ​func (def[width: Int, rank: Int, alignment: Int = 1](IndexList[rank]) capturing -> None): The body function.
• ​simd_width (Int): The SIMD vector width to use.
• ​use_blocking_impl (Bool): Do not invoke the function using asynchronous calls.
• ​target (StringSlice): The target to run on.
• ​pdl_level (PDLLevel): The PDL level controlling GPU kernel overlap behavior.
• ​_trace_description (StringSlice): Description of the trace.

Args:

• ​shape (IndexList): The shape of the buffer.
• ​context (DeviceContextPtr): The device context to use.

Raises:

If the operation fails.