mirror of
https://github.com/saymrwulf/pytorch.git
synced 2026-05-14 20:57:59 +00:00
1ad0048b64
Headers under torch/csrc/distributed may be referened with relative path, e.g., "<c10d/...>". However, relative path cannot be gracefully handled by Meta internal build when the NCCL PG is hipified to support AMD/RCCL because the "hipified" header files are generated in other directories. Moreover, using absolute path for header inclusion is the state-of-the-art in most components in Pytorch. Thus, this patch refactors all header paths in torch/csrc/distributed to be absolute. See D39835774 for more details about Meta internal complication. **How to test**: commit 9e5d199 removes -I./torch/csrc/distributed in compile options. Thus use it to verify we don't miss any relative path use of torch/csrc/distributed headers. Pull Request resolved: https://github.com/pytorch/pytorch/pull/85780 Approved by: https://github.com/kumpera, https://github.com/huydhn
63 lines
1.9 KiB
C++
63 lines
1.9 KiB
C++
#include <gtest/gtest.h>
|
|
|
|
#include "e2e_test_base.h"
|
|
|
|
#include <torch/csrc/distributed/c10d/ProcessGroupGloo.hpp>
|
|
#include <torch/csrc/distributed/rpc/request_callback_no_python.h>
|
|
#include <torch/csrc/distributed/rpc/tensorpipe_agent.h>
|
|
#include <torch/torch.h>
|
|
|
|
namespace torch {
|
|
namespace distributed {
|
|
namespace rpc {
|
|
|
|
#ifdef USE_TENSORPIPE
|
|
|
|
class TestE2ETensorPipe : public TestE2EBase {
|
|
protected:
|
|
void buildRpcAgent() override {
|
|
auto options = c10d::ProcessGroupGloo::Options::create();
|
|
options->devices.push_back(
|
|
::c10d::ProcessGroupGloo::createDeviceForHostname(serverAddress));
|
|
float rpcTimeout = 30;
|
|
|
|
TensorPipeRpcBackendOptions opts(
|
|
/*numWorkerThreads=*/std::max(16U, std::thread::hardware_concurrency()),
|
|
/*transports=*/nullopt,
|
|
/*channels=*/nullopt,
|
|
/*rpc_timeout=*/rpcTimeout,
|
|
/*init_method=*/"unused");
|
|
|
|
rpcAgent = std::make_shared<TensorPipeAgent>(
|
|
store,
|
|
"worker",
|
|
0,
|
|
numWorkers,
|
|
opts,
|
|
std::unordered_map<std::string, DeviceMap>{},
|
|
std::vector<c10::Device>{},
|
|
std::make_unique<RequestCallbackNoPython>());
|
|
}
|
|
};
|
|
|
|
// End to end training loop test in C++ so that we can run LSAN on this test to
|
|
// catch memory leaks. Enabling LSAN with python multiprocessing has been
|
|
// challenging and we don't have a good solution yet.
|
|
TEST_F(TestE2ETensorPipe, TestTrainingLoop) {
|
|
runTrainingLoop();
|
|
// Ensure the tensorpipe internal state is cleared up.
|
|
auto tensorpipeAgent = std::static_pointer_cast<TensorPipeAgent>(rpcAgent);
|
|
|
|
// Shutdown RPC agent for all RPCs to clean up.
|
|
tensorpipeAgent->join();
|
|
tensorpipeAgent->shutdown();
|
|
ASSERT_EQ(0, tensorpipeAgent->numPendingResponses());
|
|
ASSERT_EQ(0, tensorpipeAgent->timeoutMapSize());
|
|
ASSERT_EQ(0, tensorpipeAgent->messageIdToTimeoutMapSize());
|
|
}
|
|
|
|
#endif
|
|
|
|
} // namespace rpc
|
|
} // namespace distributed
|
|
} // namespace torch
|