PostsynapticPerspectiveConvolveDelivery.cpp

/*
 * PostsynapticPerspectiveConvolveDelivery.cpp
 *
 *  Created on: Aug 24, 2017
 *      Author: Pete Schultz
 */

#include "PostsynapticPerspectiveConvolveDelivery.hpp"
#include "columns/HyPerCol.hpp"

namespace PV {

PostsynapticPerspectiveConvolveDelivery::PostsynapticPerspectiveConvolveDelivery(
      char const *name,
      HyPerCol *hc) {
   initialize(name, hc);
}

PostsynapticPerspectiveConvolveDelivery::PostsynapticPerspectiveConvolveDelivery() {}

PostsynapticPerspectiveConvolveDelivery::~PostsynapticPerspectiveConvolveDelivery() {}

int PostsynapticPerspectiveConvolveDelivery::initialize(char const *name, HyPerCol *hc) {
   return BaseObject::initialize(name, hc);
}

void PostsynapticPerspectiveConvolveDelivery::setObjectType() {
   mObjectType = "PostsynapticPerspectiveConvolveDelivery";
}

int PostsynapticPerspectiveConvolveDelivery::ioParamsFillGroup(enum ParamsIOFlag ioFlag) {
   int status = HyPerDelivery::ioParamsFillGroup(ioFlag);
   return status;
}

void PostsynapticPerspectiveConvolveDelivery::ioParam_receiveGpu(enum ParamsIOFlag ioFlag) {
   mReceiveGpu = false; // If it's true, we should be using a different class.
}

Response::Status PostsynapticPerspectiveConvolveDelivery::communicateInitInfo(
      std::shared_ptr<CommunicateInitInfoMessage const> message) {
   auto status = HyPerDelivery::communicateInitInfo(message);
   if (!Response::completed(status)) {
      return status;
   }
   // HyPerDelivery::communicateInitInfo() postpones until mWeightsPair communicates.
   pvAssert(mWeightsPair and mWeightsPair->getInitInfoCommunicatedFlag());
   if (!mWeightsPair->getInitInfoCommunicatedFlag()) {
      return Response::POSTPONE;
   }
   mWeightsPair->needPost();
   return Response::SUCCESS;
}

Response::Status PostsynapticPerspectiveConvolveDelivery::allocateDataStructures() {
   auto status = HyPerDelivery::allocateDataStructures();
   return status;
}

void PostsynapticPerspectiveConvolveDelivery::deliver() {
   // Check if we need to update based on connection's channel
   if (getChannelCode() == CHANNEL_NOUPDATE) {
      return;
   }
   float *postChannel = mPostLayer->getChannel(getChannelCode());
   pvAssert(postChannel);

   int numAxonalArbors = mArborList->getNumAxonalArbors();
   for (int arbor = 0; arbor < numAxonalArbors; arbor++) {
      int delay                = mArborList->getDelay(arbor);
      PVLayerCube activityCube = mPreLayer->getPublisher()->createCube(delay);

      // Get number of neurons restricted target
      const int numPostRestricted = mPostLayer->getNumNeurons();

      const PVLayerLoc *sourceLoc = mPreLayer->getLayerLoc();
      const PVLayerLoc *targetLoc = mPostLayer->getLayerLoc();

      const int sourceNx = sourceLoc->nx;
      const int sourceNy = sourceLoc->ny;
      const int sourceNf = sourceLoc->nf;
      const int targetNx = targetLoc->nx;
      const int targetNy = targetLoc->ny;
      const int targetNf = targetLoc->nf;
      const int nbatch   = targetLoc->nbatch;

      const PVHalo *sourceHalo = &sourceLoc->halo;
      const PVHalo *targetHalo = &targetLoc->halo;

      // get source layer's extended y stride
      int sy = (sourceNx + sourceHalo->lt + sourceHalo->rt) * sourceNf;

      // The start of the gsyn buffer
      float *gSynPatchHead = mPostLayer->getChannel(getChannelCode());

      // Get source layer's patch y stride
      Weights *postWeights  = mWeightsPair->getPostWeights();
      int syp               = postWeights->getPatchStrideY();
      int yPatchSize        = postWeights->getPatchSizeY();
      int numPerStride      = postWeights->getPatchSizeX() * postWeights->getPatchSizeF();
      int neuronIndexStride = targetNf < 4 ? 1 : targetNf / 4;

      for (int b = 0; b < nbatch; b++) {
         int sourceNxExt       = sourceNx + sourceHalo->rt + sourceHalo->lt;
         int sourceNyExt       = sourceNy + sourceHalo->dn + sourceHalo->up;
         int sourceNumExtended = sourceNxExt * sourceNyExt * sourceNf;

         float *activityBatch      = activityCube.data + b * sourceNumExtended;
         float *gSynPatchHeadBatch = gSynPatchHead + b * numPostRestricted;

         // Iterate over each line in the y axis, the goal is to keep weights in the cache
         for (int ky = 0; ky < yPatchSize; ky++) {
// Threading over feature was the important change that improved cache performance by
// 5-10x. dynamic scheduling also gave another performance increase over static.
#ifdef PV_USE_OPENMP_THREADS
#pragma omp parallel for schedule(static)
#endif
            for (int feature = 0; feature < neuronIndexStride; feature++) {
               for (int idx = feature; idx < numPostRestricted; idx += neuronIndexStride) {
                  float *gSyn = gSynPatchHeadBatch + idx;

                  int idxExtended = kIndexExtended(
                        idx,
                        targetNx,
                        targetNy,
                        targetNf,
                        targetHalo->lt,
                        targetHalo->rt,
                        targetHalo->dn,
                        targetHalo->up);
                  int startSourceExt = postWeights->getGeometry()->getUnshrunkenStart(idxExtended);
                  float *a           = activityBatch + startSourceExt + ky * sy;

                  int kTargetExt = kIndexExtended(
                        idx,
                        targetNx,
                        targetNy,
                        targetNf,
                        targetHalo->lt,
                        targetHalo->rt,
                        targetHalo->dn,
                        targetHalo->up);
                  float *weightBuf    = postWeights->getDataFromPatchIndex(arbor, kTargetExt);
                  float *weightValues = weightBuf + ky * syp;

                  float dv = 0.0f;
                  for (int k = 0; k < numPerStride; ++k) {
                     dv += a[k] * weightValues[k];
                  }
                  *gSyn += mDeltaTimeFactor * dv;
               }
            }
         }
      }
   }
#ifdef PV_USE_CUDA
   // CPU updated GSyn, now need to update GSyn on GPU
   mPostLayer->setUpdatedDeviceGSynFlag(true);
#endif // PV_USE_CUDA
}

void PostsynapticPerspectiveConvolveDelivery::deliverUnitInput(

      float *recvBuffer) {
   // Get number of neurons restricted target
   const int numPostRestricted = mPostLayer->getNumNeurons();

   const PVLayerLoc *targetLoc = mPostLayer->getLayerLoc();

   const int targetNx = targetLoc->nx;
   const int targetNy = targetLoc->ny;
   const int targetNf = targetLoc->nf;
   const int nbatch   = targetLoc->nbatch;

   const PVHalo *targetHalo = &targetLoc->halo;

   // Get source layer's patch y stride
   Weights *postWeights  = mWeightsPair->getPostWeights();
   int syp               = postWeights->getPatchStrideY();
   int yPatchSize        = postWeights->getPatchSizeY();
   int numPerStride      = postWeights->getPatchSizeX() * postWeights->getPatchSizeF();
   int neuronIndexStride = targetNf < 4 ? 1 : targetNf / 4;

   int numAxonalArbors = mArborList->getNumAxonalArbors();
   for (int arbor = 0; arbor < numAxonalArbors; arbor++) {
      for (int b = 0; b < nbatch; b++) {
         float *recvBatch = recvBuffer + b * numPostRestricted;

         // Iterate over each line in the y axis, the goal is to keep weights in the cache
         for (int ky = 0; ky < yPatchSize; ky++) {
// Threading over feature was the important change that improved cache performance by
// 5-10x. dynamic scheduling also gave another performance increase over static.
#ifdef PV_USE_OPENMP_THREADS
#pragma omp parallel for schedule(static)
#endif
            for (int feature = 0; feature < neuronIndexStride; feature++) {
               for (int idx = feature; idx < numPostRestricted; idx += neuronIndexStride) {
                  float *recvLocation = recvBatch + idx;

                  int kTargetExt = kIndexExtended(
                        idx,
                        targetNx,
                        targetNy,
                        targetNf,
                        targetHalo->lt,
                        targetHalo->rt,
                        targetHalo->dn,
                        targetHalo->up);
                  float *weightBuf    = postWeights->getDataFromPatchIndex(arbor, kTargetExt);
                  float *weightValues = weightBuf + ky * syp;

                  float dv = 0.0f;
                  for (int k = 0; k < numPerStride; ++k) {
                     dv += weightValues[k];
                  }
                  *recvLocation += mDeltaTimeFactor * dv;
               }
            }
         }
      }
   }
}

} // end namespace PV