Struct StreamMemoryPool 

pub struct StreamMemoryPool { /* private fields */ }

A memory pool for stream-ordered allocations.

On platforms with a real CUDA driver (Linux, Windows), creating a pool calls cuMemPoolCreate under the hood. On macOS (where there is no NVIDIA driver), pool metadata is tracked locally but any operation that would require the driver returns Err(CudaError::NotSupported).

Allocation tracking

The pool tracks allocation counts and byte totals locally for diagnostics. These statistics are maintained even on macOS so that the API surface can be exercised in tests.

Implementations

impl StreamMemoryPool

pub fn new(config: StreamOrderedAllocConfig) -> CudaResult<Self>

Create a new memory pool for the given device.

The configuration is validated before the pool is created. On platforms with a real CUDA driver, cuMemPoolCreate is invoked. On macOS, a local-only pool is created for testing purposes.

Errors

• CudaError::InvalidValue if the config fails validation.
• CudaError::NotSupported on macOS (pool metadata is still created so that tests can exercise the API).

pub fn alloc_async( &mut self, size: usize, stream: u64, ) -> CudaResult<StreamAllocation>

Allocate memory on a stream (stream-ordered).

The allocation becomes available when all prior work on the stream has completed. The returned StreamAllocation tracks the pointer, size, and ownership.

Errors

• CudaError::InvalidValue if size is zero.
• CudaError::OutOfMemory if max_pool_size would be exceeded.
• CudaError::NotSupported on macOS.

pub fn free_async(&mut self, alloc: &mut StreamAllocation) -> CudaResult<()>

Free memory on a stream (stream-ordered).

The memory is returned to the pool when all prior work on the stream has completed. The allocation is marked as freed and cannot be freed again.

Errors

• CudaError::InvalidValue if the allocation is already freed.
• CudaError::NotSupported on macOS.

pub fn trim(&mut self, min_bytes_to_keep: usize) -> CudaResult<()>

Trim the pool, releasing unused memory back to the OS.

At least min_bytes_to_keep bytes of reserved memory will remain in the pool for future allocations.

Errors

• CudaError::NotSupported on macOS.

pub fn stats(&self) -> PoolUsageStats

Get pool usage statistics.

The returned PoolUsageStats combines locally tracked allocation counts with byte-level information. On macOS, the reserved/used byte fields mirror the local bookkeeping since no driver is available.

pub fn set_attribute(&mut self, attr: PoolAttribute) -> CudaResult<()>

Set a pool attribute.

Only attributes that carry a value (e.g. PoolAttribute::ReleaseThreshold) modify pool state. Read-only attributes (e.g. ReservedMemCurrent) return CudaError::InvalidValue.

Errors

• CudaError::InvalidValue for read-only attributes.
• CudaError::NotSupported on macOS.

pub fn enable_peer_access(&self, peer_device: i32) -> CudaResult<()>

Enable peer access from another device to allocations in this pool.

After this call, kernels running on peer_device can access memory allocated from this pool.

Errors

• CudaError::InvalidDevice if peer_device equals this pool’s device.
• CudaError::NotSupported on macOS.

pub fn disable_peer_access(&self, peer_device: i32) -> CudaResult<()>

Disable peer access from another device to allocations in this pool.

Errors

• CudaError::InvalidDevice if peer_device equals this pool’s device.
• CudaError::NotSupported on macOS.

pub fn reset_peak_stats(&mut self)

Reset peak statistics (peak allocated bytes and peak allocation count).

pub fn default_pool(device: i32) -> CudaResult<Self>

Get the default memory pool for a device.

CUDA provides a default pool per device. On macOS, this returns a local-only pool with default configuration.

Errors

• CudaError::InvalidValue if device is negative.

pub fn handle(&self) -> u64

Returns the raw pool handle.

pub fn device(&self) -> i32

Returns the device ordinal.

pub fn config(&self) -> &StreamOrderedAllocConfig

Returns the pool configuration.

Trait Implementations

impl Debug for StreamMemoryPool

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

Formats the value using the given formatter.