RPC Engine

This example showcases the RPC Engine capabilities. The code is structured as follows:

• source/ contains the different variants of this example corresponding to different backends

  • mpi.cpp corresponds to the MPI backend implementation

Topology Discovery

First, we use the topology manager to query the local HiCR instance’s topology

auto topology = topologyManager.queryTopology();

Then, we get the first available compute resource and memory space of the first device:

// Selecting first device
auto d = *topology.getDevices().begin();

// Get first memory space for RPC buffers
auto memSpaces = d->getMemorySpaceList();
auto bufferMemorySpace = *memSpaces.begin();

// Grabbing first compute resource for computing RPCs
auto computeResources = d->getComputeResourceList();
auto executeResource = *computeResources.begin();

RPC Registration

Then, create an RPC Engine define a lambda, and associate it with an RPC:

 // Creating RPC engine instance
HiCR::frontend::RPCEngine rpcEngine(...);

// Initialize RPC engine
rpcEngine.initialize();

// Creating execution unit to run as RPC
auto rpcExecutionUnit =
  std::make_shared<HiCR::backend::pthreads::ExecutionUnit>([&im](void *closure) {
    printf("Instance %lu: running Test RPC\n", im->getCurrentInstance()->getId());
  });

// Registering RPC to listen to
rpcEngine.addRPCTarget("Test RPC", rpcExecutionUnit);

SPMD: RPC Invokation and listening

The RPC is invoked by the root instance, the other instances listen to incoming RPCs:

if (currentInstance->isRootInstance())
 {
   for (auto &instance : im.getInstances())
     if (instance != currentInstance) rpcEngine.requestRPC(*instance, "Test RPC");
 }
 else
   rpcEngine.listen();

The result should look like the following:

Instance 1: running Test RPC
Instance 2: running Test RPC
Instance 4: running Test RPC
Instance 3: running Test RPC
Instance 5: running Test RPC