Converts an \c Eigen::Matrix4f to a \c float16_.

Returns \c true if the vector is empty.

Enqueues a command to copy data from \p src_image to \p dst_image. \see_opencl_ref{clEnqueueCopyImage}

Enqueues a command to map \p image into the host address space. Event associated with map operation is returned through \p map_image_event parameter.

Enqueues a command to map \p image into the host address space. Map operation is performed asynchronously. The pointer to the mapped region cannot be

Enqueues a marker in the queue and returns an event that can be used to track its progress.

Enqueues a function to execute on the host.

Enqueues a command to unmap \p image from the host memory space. \see_opencl_ref{clEnqueueUnmapMemObject}

Enqueues a command to write a rectangular region from host memory to \p buffer. \see_opencl_ref{clEnqueueWriteBufferRect} \opencl_version_warning{1,

Returns a string description of the error.

Returns a list of extensions supported by the platform.

Returns a list of extensions supported by the device.

Flushes the command queue. \see_opencl_ref{clFlush}

generate a matrix inside 'in' and do the tranposition inside 'out'

Returns the program object with \p key. Returns a null optional if no program with \p key exists in the cache.

Returns a pointer to the context object which generated the error notification.

Returns a vector of devices for the context.

Returns the address of the \p function_name extension function. Returns \c 0 if \p function_name is invalid.

\internal_ (deprecated)

Returns the flags for the memory object.

Returns a pointer to the private info memory block.

Returns the size of the private info memory block.

\internal_ (deprecated)

Returns the global memory size in bytes.

Inserts \p program into the cache with \p key.

the insert_function_call_args() method takes a tuple of argument values and inserts them into the source string with a comma in-between each. this is

returns true if the device is an nvidia gpu

returns true if the device is a POCL device

Returns \c true if the device is a sub-device.

Returns the linear size of the extents. This is equivalent to the product of each extent in each dimension.

generate a coalesced withtout bank conflicts kernel

generates a coalesced transpose kernel

generate a copy kernel program

Returns a new discard_iterator with \p index. \param index the index of the iterator \return a \c discard_iterator at \p index

generate a naive transpose kernel

warning precision is not precise due to the float error accumulation when size is large enough for more precision use double or a kahan sum else resul