Mojo module

memory

This module includes NVIDIA GPUs memory operations.

Aliases

• ​AddressSpace = _GPUAddressSpace:

Structs

• ​CacheEviction:
• ​CacheOperation:
• ​Fill:

Functions

• ​async_copy: Asynchronously copy size amount of bytes from src global memory address to shared memory dst address.
• ​async_copy_commit_group: Commits all prior initiated but uncommitted cp.async instructions into a cp.async-group.
• ​async_copy_wait_all: Wait for the completion of all commited cp.async-groups.
• ​async_copy_wait_group: Wait for the completion of n or asynchronous copy operations.
• ​cp_async_bulk_tensor_global_shared_cta: Initiates an asynchronous copy operation on the tensor data from shared cta memory to global memory.
• ​cp_async_bulk_tensor_reduce: These instructions initiate an asynchronous reduction operation of tensor data in global memory with the tensor data in shared{::cta} memory, using tile mode.
• ​cp_async_bulk_tensor_shared_cluster_global: Initiates an asynchronous copy operation on the tensor data from global memory to shared memory.
• ​cp_async_bulk_tensor_shared_cluster_global_multicast: Initiates an asynchronous multicast load operation on the tensor data from global memory to shared memories of the cluster.
• ​external_memory: Gets a pointer to dynamic shared memory.
• ​fence_mbarrier_init: Fence that applies on the prior mbarrier.init.
• ​fence_proxy_tensormap_generic_sys_acquire: Acquires tensor map system's memory fence of particular size Args: ptr: Pointer to tensor map object in system's memory. size: The size of the object.
• ​fence_proxy_tensormap_generic_sys_release: Release tensor map system's memory fence.
• ​load:
• ​tma_store_fence: Fence for SMEM stores for subsequent TMA STORE.

Was this page helpful?

Thank you! We'll create more content like this.

Thank you for helping us improve!

😔 What went wrong?

Some code doesn’t work

It includes inaccurate information

It's missing information I need

It was difficult to understand

Other

Submit