/******************************************************************************/
/*!
 * \file timlCNNProfile.c
 */
/* Copyright (C) 2015 Texas Instruments Incorporated - http://www.ti.com/
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 *
 *    Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 *
 *    Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the
 *    distribution.
 *
 *    Neither the name of Texas Instruments Incorporated nor the names of
 *    its contributors may be used to endorse or promote products derived
 *    from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 ******************************************************************************/


/*******************************************************************************
 *
 * INCLUDES
 *
 ******************************************************************************/

#include "../api/timl.h"


/******************************************************************************/
/*!
 * \ingroup   cnn
 * \brief     Profile the CNN with both timing and memory allocation
 * \param[in] cnn        CNN
 * \param[in] data       Input data batch pointer
 * \param[in] dim        Data dimension
 * \param[in] batchSize  Data batch size
 * \param[in] label      Label ptr
 * \param[in] iter       Iterations
 * \return    Error code
 */
/******************************************************************************/

int timlCNNProfile(timlConvNeuralNetwork *cnn, float *image, int row, int col, int channel, int *label, int labelRow, int labelCol, int batchSize, int iter)
{
   int             j;
   struct timespec startTime;
   struct timespec endTime;
   struct timespec fpStartTime;
   struct timespec fpEndTime;
   struct timespec bpStartTime;
   struct timespec bpEndTime;
   struct timespec layerStartTime;
   struct timespec layerEndTime;
   float           time;
   int             err;
   timlCNNLayer    *bpStartLayer;
   timlCNNLayer    *layer;
   float           cost;
   long            mem;

   err   = 0;
   layer = cnn->head;
   timlCNNPrint(cnn);
   printf("Batch size = %d, iteration = %d.\n", batchSize, iter);

   // load image and label
   timlCNNLoadImage(cnn, image, row, col, channel, cnn->params.batchSize);
   timlCNNLoadLabel(cnn, label, labelRow, labelCol, cnn->params.batchSize);

   clock_gettime(CLOCK_REALTIME, &startTime);
   clock_gettime(CLOCK_REALTIME, &fpStartTime);

   clock_gettime(CLOCK_REALTIME, &layerStartTime);
   timlCNNInputForwardPropagation(layer);
   clock_gettime(CLOCK_REALTIME, &layerEndTime);
   time = timlUtilDiffTime(layerStartTime, layerEndTime);
   printf("[Forward] layer %d(%s) = %.2f us\n", layer->id, timlCNNLayerTypeStr(layer), time);

   while (layer->next != NULL) {
      clock_gettime(CLOCK_REALTIME, &layerStartTime);
      for (j = 0; j < iter; j++) {
         switch (layer->next->type) {
            case CNN_Softmax:
               timlCNNSoftmaxForwardPropagation(layer);
               break;
            case CNN_Accuracy:
               timlCNNAccuracyForwardPropagation(layer);
               break;
            case CNN_SoftmaxCost:
               timlCNNSoftmaxCostForwardPropagation(layer);
               break;
            case CNN_Conv:
               timlCNNConvForwardPropagation(layer);
               break;
            case CNN_Norm:
               timlCNNNormForwardPropagation(layer);
               break;
            case CNN_Pooling:
               timlCNNPoolingForwardPropagation(layer);
               break;
            case CNN_Linear:
               timlCNNLinearForwardPropagation(layer);
               break;
            case CNN_Nonlinear:
               timlCNNNonlinearForwardPropagation(layer);
               break;
            case CNN_Dropout:
               timlCNNDropoutForwardPropagation(layer);
               break;
            default:
               break;
         }
      }

      clock_gettime(CLOCK_REALTIME, &layerEndTime);
      time = timlUtilDiffTime(layerStartTime, layerEndTime);
      printf("[Forward] layer %d(%s) = %.2f us\n", layer->next->id, timlCNNLayerTypeStr(layer->next), time);

      layer = layer->next;

   }

   clock_gettime(CLOCK_REALTIME, &fpEndTime);
   time = timlUtilDiffTime(fpStartTime, fpEndTime);
   printf("[Forward] total time = %.2f us\n", time);

   if (cnn->params.phase == Util_Test) {
      return err;
   }

   // back propagation
   clock_gettime(CLOCK_REALTIME, &bpStartTime);

   while (layer->prev != NULL) {
      clock_gettime(CLOCK_REALTIME, &layerStartTime);
      for (j = 0; j < iter; j++) {
         switch (layer->type) {
            case CNN_Input:
               break;
            case CNN_Conv:
               timlCNNConvBackPropagation(layer);
               break;
            case CNN_Pooling:
               timlCNNPoolingBackPropagation(layer);
               break;
            case CNN_Norm:
               timlCNNNormBackPropagation(layer);
               break;
            case CNN_Nonlinear:
               timlCNNNonlinearBackPropagation(layer);
               break;
            case CNN_Linear:
               timlCNNLinearBackPropagation(layer);
               break;
            case CNN_Dropout:
               timlCNNDropoutBackPropagation(layer);
               break;
            case CNN_SoftmaxCost:
               timlCNNSoftmaxCostBackPropagation(layer);
               break;
            case CNN_Softmax:
               timlCNNSoftmaxBackPropagation(layer);
               break;
            default:
               break;
         }
      }
      clock_gettime(CLOCK_REALTIME, &layerEndTime);
      time = timlUtilDiffTime(layerStartTime, layerEndTime);
      printf("[Backward] layer %d(%s) = %.2f us\n", layer->id, timlCNNLayerTypeStr(layer), time);
      layer = layer->prev;
   }
   clock_gettime(CLOCK_REALTIME, &bpEndTime);
   clock_gettime(CLOCK_REALTIME, &endTime);
   time = timlUtilDiffTime(bpStartTime, bpEndTime);
   printf("[Backward] total time = %.2f us\n", time);

   time = timlUtilDiffTime(startTime, endTime);
   printf("total time = %.2f us\n", time);

   return err;
}