Shortcuts

Struct Topology

Struct Documentation

struct ripple::Topology

The Topology struct stores information about the topology of the system.

It holds information about the caches, processors, processor mapping, gpu devies, etc, which can be used to optimally partiion parallel algorithms.

Public Functions

Topology() noexcept

Constructor to create the topology.

auto num_gpus() const noexcept -> size_t

Gets the number of gpus in the system.

Return

The total number of gpus in the system.

auto num_cores() const noexcept -> size_t

Gets the number of cpu cores for the system.

Return

The number of independent cpu cores in the system.

auto combined_gpu_memory() const noexcept -> uint64_t

Calculates the total number of bytes of memory available from all gpus.

Return

The total number of bytes of gpu memory in the system.

auto combined_gpu_memory_gb() const noexcept -> float

Calculates the total number of gigabytes of memory available from all gpus.

Return

The total number of gigabytes of gpu memory in the system.

auto device_to_device_available(size_t gpu_id1, size_t gpu_id2) const noexcept -> bool

Determines if peer to peer access is possible between two gpus.

Return

true if peer to peer is possible between the gpus.

Parameters
  • size_t: gpu_id1 The index of the first gpu.

  • size_t: gpu_id2 The index of the second gpu.

Public Members

std::vector<GpuInfo> gpus

Gpus for the system.

CpuInfo cpu_info

Cpu info for the system.

Docs

Access comprehensive developer documentation for Ripple

View Docs

Tutorials

Get tutorials to help with understand all features

View Tutorials

Examples

Find examples to help get started

View Examples