Struct cudarc::driver::safe::CudaSlice

pub struct CudaSlice<T> { /* private fields */ }

Contains a reference counted pointer to both device and host memory allocated for type T.

Host data

This owns the host data it is associated with. However it is possible to create device memory without having a corresponding host memory, so the host memory is actually Option.

Reclaiming host data

To reclaim the host data for this device data, use CudaDevice::sync_reclaim(). This will perform necessary synchronization to ensure that the device data finishes copying over.

Mutating device data

This can only be done by launching kernels via crate::driver::LaunchAsync which is implemented by CudaDevice. Pass &mut CudaSlice<T> if you want to mutate the rc, and &CudaSlice<T> otherwise.

Unfortunately, &CudaSlice<T> can still be mutated by the CudaFunction.

Implementations

impl<T> CudaSlice<T>

pub fn leak(self) -> CUdeviceptr

Takes ownership of the underlying sys::CUdeviceptr. It is up to the owner to free this value.

Drops the underlying host_buf if there is one.

impl<T> CudaSlice<T>

pub fn device(&self) -> Arc<CudaDevice>

Get a clone of the underlying CudaDevice.

impl<T: DeviceRepr> CudaSlice<T>

pub fn try_clone(&self) -> Result<Self, DriverError>

Allocates copy of self and schedules a device to device copy of memory.

impl<T> CudaSlice<T>

pub fn slice(&self, range: impl RangeBounds<usize>) -> CudaView<'_, T>

Creates a CudaView at the specified offset from the start of self.

Returns None if range.start >= self.len

pub fn try_slice( &self, range: impl RangeBounds<usize> ) -> Option<CudaView<'_, T>>

Fallible version of CudaSlice::slice

pub unsafe fn transmute<S>(&self, len: usize) -> Option<CudaView<'_, S>>

Reinterprets the slice of memory into a different type. len is the number of elements of the new type S that are expected. If not enough bytes are allocated in self for the view, then this returns None.

Safety

This is unsafe because not the memory for the view may not be a valid interpretation for the type S.

impl<T> CudaSlice<T>

pub fn slice_mut( &mut self, range: impl RangeBounds<usize> ) -> CudaViewMut<'_, T>

Creates a CudaViewMut at the specified offset from the start of self.

Returns None if offset >= self.len

pub fn try_slice_mut( &mut self, range: impl RangeBounds<usize> ) -> Option<CudaViewMut<'_, T>>

Fallible version of CudaSlice::slice_mut

pub unsafe fn transmute_mut<S>( &mut self, len: usize ) -> Option<CudaViewMut<'_, S>>

Reinterprets the slice of memory into a different type. len is the number of elements of the new type S that are expected. If not enough bytes are allocated in self for the view, then this returns None.

Safety

This is unsafe because not the memory for the view may not be a valid interpretation for the type S.

Trait Implementations

impl<T: DeviceRepr> Clone for CudaSlice<T>

fn clone(&self) -> Self

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<T: Debug> Debug for CudaSlice<T>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<T> DevicePtr<T> for CudaSlice<T>

fn device_ptr(&self) -> &CUdeviceptr

impl<T> DevicePtrMut<T> for CudaSlice<T>

fn device_ptr_mut(&mut self) -> &mut CUdeviceptr

impl<T: DeviceRepr> DeviceRepr for &CudaSlice<T>

fn as_kernel_param(&self) -> *mut c_void

impl<T: DeviceRepr> DeviceRepr for &mut CudaSlice<T>

fn as_kernel_param(&self) -> *mut c_void

impl<T> DeviceSlice<T> for CudaSlice<T>

fn len(&self) -> usize

fn num_bytes(&self) -> usize

fn is_empty(&self) -> bool

impl<T> Drop for CudaSlice<T>

fn drop(&mut self)

Executes the destructor for this type.

impl<T: Clone + Default + DeviceRepr + Unpin> TryFrom<CudaSlice<T>> for Vec<T>

type Error = DriverError

The type returned in the event of a conversion error.

fn try_from(value: CudaSlice<T>) -> Result<Self, Self::Error>

Performs the conversion.

impl<T: Send> Send for CudaSlice<T>

impl<T: Sync> Sync for CudaSlice<T>