HeterogeneousCore/CUDAUtilities/src/EventCache.cc - master - githubmirror/cmssw

HeterogeneousCore/CUDAUtilities/src/EventCache.cc
71 строка3 KB
Andrea Bocci
Rename GPU check macros to uppercase
24 июл 2026, 18:36
24 июл 2026, 18:364ca8016
#include "FWCore/Utilities/interface/thread_safety_macros.h"
#include "HeterogeneousCore/CUDAUtilities/interface/EventCache.h"
#include "HeterogeneousCore/CUDAUtilities/interface/cudaCheck.h"
#include "HeterogeneousCore/CUDAUtilities/interface/currentDevice.h"
#include "HeterogeneousCore/CUDAUtilities/interface/deviceCount.h"
#include "HeterogeneousCore/CUDAUtilities/interface/eventWorkHasCompleted.h"
#include "HeterogeneousCore/CUDAUtilities/interface/ScopedSetDevice.h"

namespace cms::cuda {
  void EventCache::Deleter::operator()(cudaEvent_t event) const {
    if (device_ != -1) {
      ScopedSetDevice deviceGuard{device_};
      CUDA_CHECK(cudaEventDestroy(event));
    }
  }

  // EventCache should be constructed by the first call to
  // getEventCache() only if we have CUDA devices present
  EventCache::EventCache() : cache_(deviceCount()) {}

  SharedEventPtr EventCache::get() {
    const auto dev = currentDevice();
    auto event = makeOrGet(dev);
    // captured work has completed, or a just-created event
    if (eventWorkHasCompleted(event.get())) {
      return event;
    }

    // Got an event with incomplete captured work. Try again until we
    // get a completed (or a just-created) event. Need to keep all
    // incomplete events until a completed event is found in order to
    // avoid ping-pong with an incomplete event.
    std::vector<SharedEventPtr> ptrs{std::move(event)};
    bool completed;
    do {
      event = makeOrGet(dev);
      completed = eventWorkHasCompleted(event.get());
      if (not completed) {
        ptrs.emplace_back(std::move(event));
      }
    } while (not completed);
    return event;
  }

  SharedEventPtr EventCache::makeOrGet(int dev) {
    return cache_[dev].makeOrGet([dev]() {
      cudaEvent_t event;
      // it should be a bit faster to ignore timings
      // cudaEventBlockingSync is needed to let the thread calling
      // cudaEventSynchronize() to sleep instead of spinning the CPU
      CUDA_CHECK(cudaEventCreateWithFlags(&event, cudaEventDisableTiming | cudaEventBlockingSync));
      return std::unique_ptr<BareEvent, Deleter>(event, Deleter{dev});
    });
  }

  void EventCache::clear() {
    // Reset the contents of the caches, but leave an
    // edm::ReusableObjectHolder alive for each device. This is needed
    // mostly for the unit tests, where the function-static
    // EventCache lives through multiple tests (and go through
    // multiple shutdowns of the framework).
    cache_.clear();
    cache_.resize(deviceCount());
  }

  EventCache& getEventCache() {
    // the public interface is thread safe
    CMS_THREAD_SAFE static EventCache cache;
    return cache;
  }
}  // namespace cms::cuda