CpuTopology

The CpuTopology class provides system-level introspection for CPU layout and NUMA topology. It is used to understand the physical and logical arrangement of cores and to optimize thread and memory placement.

Purpose

Determine number of physical CPU cores
Identify isolated and non-isolated cores
Access detailed NUMA topology (if available)
Map CPU cores to their NUMA nodes

Key Structures

`NumaNode`

struct NumaNode {
  int nodeId;
  std::vector<int> cpuCores;
  size_t totalMemoryMB;
  size_t freeMemoryMB;
};
````

### `NumaTopology`

```cpp
struct NumaTopology {
  std::vector<NumaNode> nodes;
  int numNodes;
  bool numaAvailable;
};

Responsibilities

Method	Description
`getNumCores()`	Returns total number of logical cores
`getAllCores()`	Returns list of core indices from 0 to N-1
`getIsolatedCores()`	Returns list of isolated CPUs (from cmdline)
`getNonIsolatedCores()`	All cores minus isolated ones
`isNumaAvailable()`	Whether NUMA support is available on the system
`getNumaTopology()`	Returns detailed info about all NUMA nodes
`getNumaNodeForCore(coreId)`	Returns the NUMA node to which a specific core belongs
`getCoresForNumaNode(nodeId)`	Lists cores that belong to a given NUMA node

Lazy Caching

CpuTopology caches:

Core count
Isolated cores
NUMA layout

This minimizes redundant syscalls and file reads.

Usage Example

CpuTopology topology(createSystemInterface());

if (topology.isNumaAvailable()) {
  auto nodes = topology.getNumaTopology().nodes;
  for (const auto& node : nodes) {
    std::cout << "NUMA Node " << node.nodeId << ": ";
    for (int core : node.cpuCores)
      std::cout << core << " ";
    std::cout << "\n";
  }
}

Integration

CpuTopology is used by:

CpuAffinity for pinning and scheduling
CoreAssignmentManager for optimized component placement
NumaAffinityGuard for temporary core+memory locality enforcement

It provides the necessary insights into hardware layout to guide all thread and memory affinity operations.