warp.tid

warp.tid() → int

Kernel

Query the current thread index or indices.

The return type is determined by the unpacking syntax used:

• i = wp.tid() - Returns the first thread index as int

• i, j = wp.tid() - Returns the first two indices as tuple[int, int]

• i, j, k = wp.tid() - Returns the first three indices as tuple[int, int, int]

• i, j, k, l = wp.tid() - Returns all four indices as tuple[int, int, int, int]

The indices correspond to the thread’s position in the kernel launch grid. If fewer indices are requested than the launch dimensionality, only the leading indices are returned. For multi-dimensional launches, the linear thread order is unraveled in row-major order, with the last launch dimension varying fastest.

This function may not be called from user-defined Warp functions.

warp.tid() → tuple[int, int]

Kernel

Query the current thread index or indices.

The return type is determined by the unpacking syntax used:

• i = wp.tid() - Returns the first thread index as int

• i, j = wp.tid() - Returns the first two indices as tuple[int, int]

• i, j, k = wp.tid() - Returns the first three indices as tuple[int, int, int]

• i, j, k, l = wp.tid() - Returns all four indices as tuple[int, int, int, int]

The indices correspond to the thread’s position in the kernel launch grid. If fewer indices are requested than the launch dimensionality, only the leading indices are returned. For multi-dimensional launches, the linear thread order is unraveled in row-major order, with the last launch dimension varying fastest.

This function may not be called from user-defined Warp functions.

warp.tid() → tuple[int, int, int]

Kernel

Query the current thread index or indices.

The return type is determined by the unpacking syntax used:

• i = wp.tid() - Returns the first thread index as int

• i, j = wp.tid() - Returns the first two indices as tuple[int, int]

• i, j, k = wp.tid() - Returns the first three indices as tuple[int, int, int]

• i, j, k, l = wp.tid() - Returns all four indices as tuple[int, int, int, int]

The indices correspond to the thread’s position in the kernel launch grid. If fewer indices are requested than the launch dimensionality, only the leading indices are returned. For multi-dimensional launches, the linear thread order is unraveled in row-major order, with the last launch dimension varying fastest.

This function may not be called from user-defined Warp functions.

warp.tid() → tuple[int, int, int, int]

Kernel

Query the current thread index or indices.

The return type is determined by the unpacking syntax used:

• i = wp.tid() - Returns the first thread index as int

• i, j = wp.tid() - Returns the first two indices as tuple[int, int]

• i, j, k = wp.tid() - Returns the first three indices as tuple[int, int, int]

• i, j, k, l = wp.tid() - Returns all four indices as tuple[int, int, int, int]

The indices correspond to the thread’s position in the kernel launch grid. If fewer indices are requested than the launch dimensionality, only the leading indices are returned. For multi-dimensional launches, the linear thread order is unraveled in row-major order, with the last launch dimension varying fastest.

This function may not be called from user-defined Warp functions.