How to use ops method in pytest-benchmark

Best Python code snippet using pytest-benchmark

inception_model.py

Source:inception_model.py

1# Copyright 2016 Google Inc. All Rights Reserved.2#3# Licensed under the Apache License, Version 2.0 (the "License");4# you may not use this file except in compliance with the License.5# You may obtain a copy of the License at6#7# http://www.apache.org/licenses/LICENSE-2.08#9# Unless required by applicable law or agreed to in writing, software10# distributed under the License is distributed on an "AS IS" BASIS,11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.12# See the License for the specific language governing permissions and13# limitations under the License.14# ==============================================================================15"""Inception-v3 expressed in TensorFlow-Slim.16  Usage:17  # Parameters for BatchNorm.18  batch_norm_params = {19      # Decay for the batch_norm moving averages.20      'decay': BATCHNORM_MOVING_AVERAGE_DECAY,21      # epsilon to prevent 0s in variance.22      'epsilon': 0.001,23  }24  # Set weight_decay for weights in Conv and FC layers.25  with slim.arg_scope([slim.ops.conv2d, slim.ops.fc], weight_decay=0.00004):26    with slim.arg_scope([slim.ops.conv2d],27                        stddev=0.1,28                        activation=tf.nn.relu,29                        batch_norm_params=batch_norm_params):30      # Force all Variables to reside on the CPU.31      with slim.arg_scope([slim.variables.variable], device='/cpu:0'):32        logits, endpoints = slim.inception.inception_v3(33            images,34            dropout_keep_prob=0.8,35            num_classes=num_classes,36            is_training=for_training,37            restore_logits=restore_logits,38            scope=scope)39"""40from __future__ import absolute_import41from __future__ import division42from __future__ import print_function43import tensorflow as tf44from inception.slim import ops45from inception.slim import scopes46def inception_v3(inputs,47                 dropout_keep_prob=0.8,48                 num_classes=1000,49                 is_training=True,50                 restore_logits=True,51                 scope=''):52  """Latest Inception from http://arxiv.org/abs/1512.00567.53    "Rethinking the Inception Architecture for Computer Vision"54    Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jonathon Shlens,55    Zbigniew Wojna56  Args:57    inputs: a tensor of size [batch_size, height, width, channels].58    dropout_keep_prob: dropout keep_prob.59    num_classes: number of predicted classes.60    is_training: whether is training or not.61    restore_logits: whether or not the logits layers should be restored.62      Useful for fine-tuning a model with different num_classes.63    scope: Optional scope for name_scope.64  Returns:65    a list containing 'logits', 'aux_logits' Tensors.66  """67  # end_points will collect relevant activations for external use, for example68  # summaries or losses.69  end_points = {}70  with tf.name_scope(scope, 'inception_v3', [inputs]):71    with scopes.arg_scope([ops.conv2d, ops.fc, ops.batch_norm, ops.dropout],72                          is_training=is_training):73      with scopes.arg_scope([ops.conv2d, ops.max_pool, ops.avg_pool],74                            stride=1, padding='VALID'):75        # 299 x 299 x 376        end_points['conv0'] = ops.conv2d(inputs, 32, [3, 3], stride=2,77                                         scope='conv0')78        # 149 x 149 x 3279        end_points['conv1'] = ops.conv2d(end_points['conv0'], 32, [3, 3],80                                         scope='conv1')81        # 147 x 147 x 3282        end_points['conv2'] = ops.conv2d(end_points['conv1'], 64, [3, 3],83                                         padding='SAME', scope='conv2')84        # 147 x 147 x 6485        end_points['pool1'] = ops.max_pool(end_points['conv2'], [3, 3],86                                           stride=2, scope='pool1')87        # 73 x 73 x 6488        end_points['conv3'] = ops.conv2d(end_points['pool1'], 80, [1, 1],89                                         scope='conv3')90        # 73 x 73 x 80.91        end_points['conv4'] = ops.conv2d(end_points['conv3'], 192, [3, 3],92                                         scope='conv4')93        # 71 x 71 x 192.94        end_points['pool2'] = ops.max_pool(end_points['conv4'], [3, 3],95                                           stride=2, scope='pool2')96        # 35 x 35 x 192.97        net = end_points['pool2']98      # Inception blocks99      with scopes.arg_scope([ops.conv2d, ops.max_pool, ops.avg_pool],100                            stride=1, padding='SAME'):101        # mixed: 35 x 35 x 256.102        with tf.variable_scope('mixed_35x35x256a'):103          with tf.variable_scope('branch1x1'):104            branch1x1 = ops.conv2d(net, 64, [1, 1])105          with tf.variable_scope('branch5x5'):106            branch5x5 = ops.conv2d(net, 48, [1, 1])107            branch5x5 = ops.conv2d(branch5x5, 64, [5, 5])108          with tf.variable_scope('branch3x3dbl'):109            branch3x3dbl = ops.conv2d(net, 64, [1, 1])110            branch3x3dbl = ops.conv2d(branch3x3dbl, 96, [3, 3])111            branch3x3dbl = ops.conv2d(branch3x3dbl, 96, [3, 3])112          with tf.variable_scope('branch_pool'):113            branch_pool = ops.avg_pool(net, [3, 3])114            branch_pool = ops.conv2d(branch_pool, 32, [1, 1])115          net = tf.concat([branch1x1, branch5x5, branch3x3dbl, branch_pool], 3)116          end_points['mixed_35x35x256a'] = net117        # mixed_1: 35 x 35 x 288.118        with tf.variable_scope('mixed_35x35x288a'):119          with tf.variable_scope('branch1x1'):120            branch1x1 = ops.conv2d(net, 64, [1, 1])121          with tf.variable_scope('branch5x5'):122            branch5x5 = ops.conv2d(net, 48, [1, 1])123            branch5x5 = ops.conv2d(branch5x5, 64, [5, 5])124          with tf.variable_scope('branch3x3dbl'):125            branch3x3dbl = ops.conv2d(net, 64, [1, 1])126            branch3x3dbl = ops.conv2d(branch3x3dbl, 96, [3, 3])127            branch3x3dbl = ops.conv2d(branch3x3dbl, 96, [3, 3])128          with tf.variable_scope('branch_pool'):129            branch_pool = ops.avg_pool(net, [3, 3])130            branch_pool = ops.conv2d(branch_pool, 64, [1, 1])131          net = tf.concat([branch1x1, branch5x5, branch3x3dbl, branch_pool], 3)132          end_points['mixed_35x35x288a'] = net133        # mixed_2: 35 x 35 x 288.134        with tf.variable_scope('mixed_35x35x288b'):135          with tf.variable_scope('branch1x1'):136            branch1x1 = ops.conv2d(net, 64, [1, 1])137          with tf.variable_scope('branch5x5'):138            branch5x5 = ops.conv2d(net, 48, [1, 1])139            branch5x5 = ops.conv2d(branch5x5, 64, [5, 5])140          with tf.variable_scope('branch3x3dbl'):141            branch3x3dbl = ops.conv2d(net, 64, [1, 1])142            branch3x3dbl = ops.conv2d(branch3x3dbl, 96, [3, 3])143            branch3x3dbl = ops.conv2d(branch3x3dbl, 96, [3, 3])144          with tf.variable_scope('branch_pool'):145            branch_pool = ops.avg_pool(net, [3, 3])146            branch_pool = ops.conv2d(branch_pool, 64, [1, 1])147          net = tf.concat([branch1x1, branch5x5, branch3x3dbl, branch_pool], 3)148          end_points['mixed_35x35x288b'] = net149        # mixed_3: 17 x 17 x 768.150        with tf.variable_scope('mixed_17x17x768a'):151          with tf.variable_scope('branch3x3'):152            branch3x3 = ops.conv2d(net, 384, [3, 3], stride=2, padding='VALID')153          with tf.variable_scope('branch3x3dbl'):154            branch3x3dbl = ops.conv2d(net, 64, [1, 1])155            branch3x3dbl = ops.conv2d(branch3x3dbl, 96, [3, 3])156            branch3x3dbl = ops.conv2d(branch3x3dbl, 96, [3, 3],157                                      stride=2, padding='VALID')158          with tf.variable_scope('branch_pool'):159            branch_pool = ops.max_pool(net, [3, 3], stride=2, padding='VALID')160          net = tf.concat([branch3x3, branch3x3dbl, branch_pool], 3)161          end_points['mixed_17x17x768a'] = net162        # mixed4: 17 x 17 x 768.163        with tf.variable_scope('mixed_17x17x768b'):164          with tf.variable_scope('branch1x1'):165            branch1x1 = ops.conv2d(net, 192, [1, 1])166          with tf.variable_scope('branch7x7'):167            branch7x7 = ops.conv2d(net, 128, [1, 1])168            branch7x7 = ops.conv2d(branch7x7, 128, [1, 7])169            branch7x7 = ops.conv2d(branch7x7, 192, [7, 1])170          with tf.variable_scope('branch7x7dbl'):171            branch7x7dbl = ops.conv2d(net, 128, [1, 1])172            branch7x7dbl = ops.conv2d(branch7x7dbl, 128, [7, 1])173            branch7x7dbl = ops.conv2d(branch7x7dbl, 128, [1, 7])174            branch7x7dbl = ops.conv2d(branch7x7dbl, 128, [7, 1])175            branch7x7dbl = ops.conv2d(branch7x7dbl, 192, [1, 7])176          with tf.variable_scope('branch_pool'):177            branch_pool = ops.avg_pool(net, [3, 3])178            branch_pool = ops.conv2d(branch_pool, 192, [1, 1])179          net = tf.concat([branch1x1, branch7x7, branch7x7dbl, branch_pool], 3)180          end_points['mixed_17x17x768b'] = net181        # mixed_5: 17 x 17 x 768.182        with tf.variable_scope('mixed_17x17x768c'):183          with tf.variable_scope('branch1x1'):184            branch1x1 = ops.conv2d(net, 192, [1, 1])185          with tf.variable_scope('branch7x7'):186            branch7x7 = ops.conv2d(net, 160, [1, 1])187            branch7x7 = ops.conv2d(branch7x7, 160, [1, 7])188            branch7x7 = ops.conv2d(branch7x7, 192, [7, 1])189          with tf.variable_scope('branch7x7dbl'):190            branch7x7dbl = ops.conv2d(net, 160, [1, 1])191            branch7x7dbl = ops.conv2d(branch7x7dbl, 160, [7, 1])192            branch7x7dbl = ops.conv2d(branch7x7dbl, 160, [1, 7])193            branch7x7dbl = ops.conv2d(branch7x7dbl, 160, [7, 1])194            branch7x7dbl = ops.conv2d(branch7x7dbl, 192, [1, 7])195          with tf.variable_scope('branch_pool'):196            branch_pool = ops.avg_pool(net, [3, 3])197            branch_pool = ops.conv2d(branch_pool, 192, [1, 1])198          net = tf.concat([branch1x1, branch7x7, branch7x7dbl, branch_pool], 3)199          end_points['mixed_17x17x768c'] = net200        # mixed_6: 17 x 17 x 768.201        with tf.variable_scope('mixed_17x17x768d'):202          with tf.variable_scope('branch1x1'):203            branch1x1 = ops.conv2d(net, 192, [1, 1])204          with tf.variable_scope('branch7x7'):205            branch7x7 = ops.conv2d(net, 160, [1, 1])206            branch7x7 = ops.conv2d(branch7x7, 160, [1, 7])207            branch7x7 = ops.conv2d(branch7x7, 192, [7, 1])208          with tf.variable_scope('branch7x7dbl'):209            branch7x7dbl = ops.conv2d(net, 160, [1, 1])210            branch7x7dbl = ops.conv2d(branch7x7dbl, 160, [7, 1])211            branch7x7dbl = ops.conv2d(branch7x7dbl, 160, [1, 7])212            branch7x7dbl = ops.conv2d(branch7x7dbl, 160, [7, 1])213            branch7x7dbl = ops.conv2d(branch7x7dbl, 192, [1, 7])214          with tf.variable_scope('branch_pool'):215            branch_pool = ops.avg_pool(net, [3, 3])216            branch_pool = ops.conv2d(branch_pool, 192, [1, 1])217          net = tf.concat([branch1x1, branch7x7, branch7x7dbl, branch_pool], 3)218          end_points['mixed_17x17x768d'] = net219        # mixed_7: 17 x 17 x 768.220        with tf.variable_scope('mixed_17x17x768e'):221          with tf.variable_scope('branch1x1'):222            branch1x1 = ops.conv2d(net, 192, [1, 1])223          with tf.variable_scope('branch7x7'):224            branch7x7 = ops.conv2d(net, 192, [1, 1])225            branch7x7 = ops.conv2d(branch7x7, 192, [1, 7])226            branch7x7 = ops.conv2d(branch7x7, 192, [7, 1])227          with tf.variable_scope('branch7x7dbl'):228            branch7x7dbl = ops.conv2d(net, 192, [1, 1])229            branch7x7dbl = ops.conv2d(branch7x7dbl, 192, [7, 1])230            branch7x7dbl = ops.conv2d(branch7x7dbl, 192, [1, 7])231            branch7x7dbl = ops.conv2d(branch7x7dbl, 192, [7, 1])232            branch7x7dbl = ops.conv2d(branch7x7dbl, 192, [1, 7])233          with tf.variable_scope('branch_pool'):234            branch_pool = ops.avg_pool(net, [3, 3])235            branch_pool = ops.conv2d(branch_pool, 192, [1, 1])236          net = tf.concat([branch1x1, branch7x7, branch7x7dbl, branch_pool], 3)237          end_points['mixed_17x17x768e'] = net238        # Auxiliary Head logits239        aux_logits = tf.identity(end_points['mixed_17x17x768e'])240        with tf.variable_scope('aux_logits'):241          aux_logits = ops.avg_pool(aux_logits, [5, 5], stride=3,242                                    padding='VALID')243          aux_logits = ops.conv2d(aux_logits, 128, [1, 1], scope='proj')244          # Shape of feature map before the final layer.245          shape = aux_logits.get_shape()246          aux_logits = ops.conv2d(aux_logits, 768, shape[1:3], stddev=0.01,247                                  padding='VALID')248          aux_logits = ops.flatten(aux_logits)249          aux_logits = ops.fc(aux_logits, num_classes, activation=None,250                              stddev=0.001, restore=restore_logits)251          end_points['aux_logits'] = aux_logits252        # mixed_8: 8 x 8 x 1280.253        # Note that the scope below is not changed to not void previous254        # checkpoints.255        # (TODO) Fix the scope when appropriate.256        with tf.variable_scope('mixed_17x17x1280a'):257          with tf.variable_scope('branch3x3'):258            branch3x3 = ops.conv2d(net, 192, [1, 1])259            branch3x3 = ops.conv2d(branch3x3, 320, [3, 3], stride=2,260                                   padding='VALID')261          with tf.variable_scope('branch7x7x3'):262            branch7x7x3 = ops.conv2d(net, 192, [1, 1])263            branch7x7x3 = ops.conv2d(branch7x7x3, 192, [1, 7])264            branch7x7x3 = ops.conv2d(branch7x7x3, 192, [7, 1])265            branch7x7x3 = ops.conv2d(branch7x7x3, 192, [3, 3],266                                     stride=2, padding='VALID')267          with tf.variable_scope('branch_pool'):268            branch_pool = ops.max_pool(net, [3, 3], stride=2, padding='VALID')269          net = tf.concat([branch3x3, branch7x7x3, branch_pool], 3)270          end_points['mixed_17x17x1280a'] = net271        # mixed_9: 8 x 8 x 2048.272        with tf.variable_scope('mixed_8x8x2048a'):273          with tf.variable_scope('branch1x1'):274            branch1x1 = ops.conv2d(net, 320, [1, 1])275          with tf.variable_scope('branch3x3'):276            branch3x3 = ops.conv2d(net, 384, [1, 1])277            branch3x3 = tf.concat([ops.conv2d(branch3x3, 384, [1, 3]),278                                   ops.conv2d(branch3x3, 384, [3, 1])], 3)279          with tf.variable_scope('branch3x3dbl'):280            branch3x3dbl = ops.conv2d(net, 448, [1, 1])281            branch3x3dbl = ops.conv2d(branch3x3dbl, 384, [3, 3])282            branch3x3dbl = tf.concat([ops.conv2d(branch3x3dbl, 384, [1, 3]),283                                      ops.conv2d(branch3x3dbl, 384, [3, 1])], 3)284          with tf.variable_scope('branch_pool'):285            branch_pool = ops.avg_pool(net, [3, 3])286            branch_pool = ops.conv2d(branch_pool, 192, [1, 1])287          net = tf.concat([branch1x1, branch3x3, branch3x3dbl, branch_pool], 3)288          end_points['mixed_8x8x2048a'] = net289        # mixed_10: 8 x 8 x 2048.290        with tf.variable_scope('mixed_8x8x2048b'):291          with tf.variable_scope('branch1x1'):292            branch1x1 = ops.conv2d(net, 320, [1, 1])293          with tf.variable_scope('branch3x3'):294            branch3x3 = ops.conv2d(net, 384, [1, 1])295            branch3x3 = tf.concat([ops.conv2d(branch3x3, 384, [1, 3]),296                                   ops.conv2d(branch3x3, 384, [3, 1])], 3)297          with tf.variable_scope('branch3x3dbl'):298            branch3x3dbl = ops.conv2d(net, 448, [1, 1])299            branch3x3dbl = ops.conv2d(branch3x3dbl, 384, [3, 3])300            branch3x3dbl = tf.concat([ops.conv2d(branch3x3dbl, 384, [1, 3]),301                                      ops.conv2d(branch3x3dbl, 384, [3, 1])], 3)302          with tf.variable_scope('branch_pool'):303            branch_pool = ops.avg_pool(net, [3, 3])304            branch_pool = ops.conv2d(branch_pool, 192, [1, 1])305          net = tf.concat([branch1x1, branch3x3, branch3x3dbl, branch_pool], 3)306          end_points['mixed_8x8x2048b'] = net307        # Final pooling and prediction308        with tf.variable_scope('logits'):309          shape = net.get_shape()310          net = ops.avg_pool(net, shape[1:3], padding='VALID', scope='pool')311          # 1 x 1 x 2048312          net = ops.dropout(net, dropout_keep_prob, scope='dropout')313          net = ops.flatten(net, scope='flatten')314          # 2048315          logits = ops.fc(net, num_classes, activation=None, scope='logits',316                          restore=restore_logits)317          # 1000318          end_points['logits'] = logits319          end_points['predictions'] = tf.nn.softmax(logits, name='predictions')320      return logits, end_points321def inception_v3_parameters(weight_decay=0.00004, stddev=0.1,322                            batch_norm_decay=0.9997, batch_norm_epsilon=0.001):323  """Yields the scope with the default parameters for inception_v3.324  Args:325    weight_decay: the weight decay for weights variables.326    stddev: standard deviation of the truncated guassian weight distribution.327    batch_norm_decay: decay for the moving average of batch_norm momentums.328    batch_norm_epsilon: small float added to variance to avoid dividing by zero.329  Yields:330    a arg_scope with the parameters needed for inception_v3.331  """332  # Set weight_decay for weights in Conv and FC layers.333  with scopes.arg_scope([ops.conv2d, ops.fc],334                        weight_decay=weight_decay):335    # Set stddev, activation and parameters for batch_norm.336    with scopes.arg_scope([ops.conv2d],337                          stddev=stddev,338                          activation=tf.nn.relu,339                          batch_norm_params={340                              'decay': batch_norm_decay,341                              'epsilon': batch_norm_epsilon}) as arg_scope:...

gen_stimuli.py

Source:gen_stimuli.py

1#!/usr/bin/env python-2.52import sys3import random4# Copyright 2017 ETH Zurich and University of Bologna.5# Copyright and related rights are licensed under the Solderpad Hardware6# License, Version 0.51 (the License); you may not use this file except in7# compliance with the License.  You may obtain a copy of the License at8# http://solderpad.org/licenses/SHL-0.51. Unless required by applicable law9# or agreed to in writing, software, hardware and materials distributed under10# this License is distributed on an AS IS BASIS, WITHOUT WARRANTIES OR11# CONDITIONS OF ANY KIND, either express or implied. See the License for the12# specific language governing permissions and limitations under the License.13def write_hex8_arr(f, name, arr):14    f.write('unsigned int %s[] = {\n' % name)15    for v in arr:16        f.write('0x%02X%02X%02X%02X,\n' % (v[0] & 0xFF, v[1] & 0xFF, v[2] & 0xFF, v[3] & 0xFF))17    f.write('};\n\n')18    return19def write_hex16_arr(f, name, arr):20    f.write('unsigned int %s[] = {\n' % name)21    for v in arr:22        f.write('0x%04X%04X,\n' % (v[0] & 0xFFFF, v[1] & 0xFFFF))23    f.write('};\n\n')24    return25def write_hex32_arr(f, name, arr):26    f.write('unsigned int %s[] = {\n' % name)27    for v in arr:28        f.write('0x%08X,\n' % (v & 0xFFFFFFFF))29    f.write('};\n\n')30    return31f = open('testALU_stimuli.h', 'w')32# number of test cases to generate33N = 1534################################################################################35################################################################################36#37# ADDITION38#39################################################################################40################################################################################41################################################################################42# generate testdata for l.add43################################################################################44ops_a    = []45ops_b    = []46exp_res  = []47for i in range(N):48    a = random.randint(-2**31, 2**31-1)49    b = random.randint(-2**31, 2**31-1)50    r = a + b51    ops_a.append(a)52    ops_b.append(b)53    exp_res.append(r)54write_hex32_arr(f, 'g_add_a', ops_a)55write_hex32_arr(f, 'g_add_b', ops_b)56write_hex32_arr(f, 'g_add_act', ops_a)57write_hex32_arr(f, 'g_add_exp', exp_res)58################################################################################59# generate testdata for l.addi60################################################################################61ops_a    = []62exp_res  = []63for i in range(N):64    a = random.randint(-2**31, 2**31-1)65    r = a + 10066    ops_a.append(a)67    exp_res.append(r)68write_hex32_arr(f, 'g_addi_a', ops_a)69write_hex32_arr(f, 'g_addi_act', ops_a)70write_hex32_arr(f, 'g_addi_exp', exp_res)71################################################################################72################################################################################73#74# SUBTRACTION75#76################################################################################77################################################################################78################################################################################79# generate testdata for l.sub80################################################################################81ops_a    = []82ops_b    = []83exp_res  = []84for i in range(N):85    a = random.randint(-2**31, 2**31-1)86    b = random.randint(-2**31, 2**31-1)87    r = a - b88    ops_a.append(a)89    ops_b.append(b)90    exp_res.append(r)91write_hex32_arr(f, 'g_sub_a', ops_a)92write_hex32_arr(f, 'g_sub_b', ops_b)93write_hex32_arr(f, 'g_sub_act', ops_a)94write_hex32_arr(f, 'g_sub_exp', exp_res)95################################################################################96################################################################################97#98# LOGIC STUFF99#100################################################################################101################################################################################102################################################################################103# generate testdata for l.and104################################################################################105ops_a    = []106ops_b    = []107exp_res  = []108for i in range(N):109    a = random.randint(-2**31, 2**31-1)110    b = random.randint(-2**31, 2**31-1)111    r = a & b112    ops_a.append(a)113    ops_b.append(b)114    exp_res.append(r)115write_hex32_arr(f, 'g_and_a', ops_a)116write_hex32_arr(f, 'g_and_b', ops_b)117write_hex32_arr(f, 'g_and_act', ops_a)118write_hex32_arr(f, 'g_and_exp', exp_res)119################################################################################120# generate testdata for l.or121################################################################################122ops_a    = []123ops_b    = []124exp_res  = []125for i in range(N):126    a = random.randint(-2**31, 2**31-1)127    b = random.randint(-2**31, 2**31-1)128    r = a | b129    ops_a.append(a)130    ops_b.append(b)131    exp_res.append(r)132write_hex32_arr(f, 'g_or_a', ops_a)133write_hex32_arr(f, 'g_or_b', ops_b)134write_hex32_arr(f, 'g_or_act', ops_a)135write_hex32_arr(f, 'g_or_exp', exp_res)136################################################################################137# generate testdata for l.xor138################################################################################139ops_a    = []140ops_b    = []141exp_res  = []142for i in range(N):143    a = random.randint(-2**31, 2**31-1)144    b = random.randint(-2**31, 2**31-1)145    r = a ^ b146    ops_a.append(a)147    ops_b.append(b)148    exp_res.append(r)149write_hex32_arr(f, 'g_xor_a', ops_a)150write_hex32_arr(f, 'g_xor_b', ops_b)151write_hex32_arr(f, 'g_xor_act', ops_a)152write_hex32_arr(f, 'g_xor_exp', exp_res)153################################################################################154################################################################################155#156# Shifting157#158################################################################################159################################################################################160################################################################################161# generate testdata for l.sll162################################################################################163ops_a    = []164ops_b    = []165exp_res  = []166for i in range(N):167    a = random.randint(-2**31, 2**31-1)168    b = random.randint(0, 31)169    r = a << b170    ops_a.append(a)171    ops_b.append(b)172    exp_res.append(r)173write_hex32_arr(f, 'g_sll_a', ops_a)174write_hex32_arr(f, 'g_sll_b', ops_b)175write_hex32_arr(f, 'g_sll_act', ops_a)176write_hex32_arr(f, 'g_sll_exp', exp_res)177################################################################################178# generate testdata for l.srl179################################################################################180ops_a    = []181ops_b    = []182exp_res  = []183for i in range(N):184    a = random.randint(-2**31, 2**31-1)185    b = random.randint(0, 31)186    # python does not provide logical right shifts, so we have to hack187    # something together188    r = (a % 0x100000000) >> b189    ops_a.append(a)190    ops_b.append(b)191    exp_res.append(r)192write_hex32_arr(f, 'g_srl_a', ops_a)193write_hex32_arr(f, 'g_srl_b', ops_b)194write_hex32_arr(f, 'g_srl_act', ops_a)195write_hex32_arr(f, 'g_srl_exp', exp_res)196################################################################################197# generate testdata for l.sra198################################################################################199ops_a    = []200ops_b    = []201exp_res  = []202for i in range(N):203    a = random.randint(-2**31, 2**31-1)204    b = random.randint(0, 31)205    r = a >> b206    ops_a.append(a)207    ops_b.append(b)208    exp_res.append(r)209write_hex32_arr(f, 'g_sra_a', ops_a)210write_hex32_arr(f, 'g_sra_b', ops_b)211write_hex32_arr(f, 'g_sra_act', ops_a)212write_hex32_arr(f, 'g_sra_exp', exp_res)213################################################################################214# generate testdata for l.slli215################################################################################216ops_a    = []217exp_res  = []218for i in range(N):219    a = random.randint(-2**31, 2**31-1)220    r = a << 10221    ops_a.append(a)222    exp_res.append(r)223write_hex32_arr(f, 'g_slli_a', ops_a)224write_hex32_arr(f, 'g_slli_act', ops_a)225write_hex32_arr(f, 'g_slli_exp', exp_res)226################################################################################227# generate testdata for l.srli228################################################################################229ops_a    = []230exp_res  = []231for i in range(N):232    a = random.randint(-2**31, 2**31-1)233    # python does not provide logical right shifts, so we have to hack234    # something together235    r = (a % 0x100000000) >> 9236    ops_a.append(a)237    exp_res.append(r)238write_hex32_arr(f, 'g_srli_a', ops_a)239write_hex32_arr(f, 'g_srli_act', ops_a)240write_hex32_arr(f, 'g_srli_exp', exp_res)241################################################################################242# generate testdata for l.srai243################################################################################244ops_a    = []245exp_res  = []246for i in range(N):247    a = random.randint(-2**31, 2**31-1)248    r = a >> 21249    ops_a.append(a)250    exp_res.append(r)251write_hex32_arr(f, 'g_srai_a', ops_a)252write_hex32_arr(f, 'g_srai_act', ops_a)253write_hex32_arr(f, 'g_srai_exp', exp_res)254# ################################################################################255# # generate testdata for l.cmov256# ################################################################################257# ops_a    = []258# ops_b    = []259# exp_res  = []260#261# for i in range(N):262#     a = random.randint(0, 2**32-1)263#     b = random.randint(0, 2**32-1)264#265#     if(a > b):266#         r = a267#     else:268#         r = b269#270#     ops_a.append(a)271#     ops_b.append(b)272#     exp_res.append(r)273#274# write_hex32_arr(f, 'g_cmov_a', ops_a)275# write_hex32_arr(f, 'g_cmov_b', ops_b)276# write_hex32_arr(f, 'g_cmov_act', ops_a)...

count_hooks.py

Source:count_hooks.py

1import argparse2import torch3import torch.nn as nn4multiply_adds = 15def count_convNd(m, x, y):6    x = x[0]7    cin = m.in_channels8    batch_size = x.size(0)9    kernel_ops = m.weight.size()[2:].numel()10    bias_ops = 1 if m.bias is not None else 011    ops_per_element = kernel_ops + bias_ops12    output_elements = y.nelement()13    # cout x oW x oH14    total_ops = batch_size * cin * output_elements * ops_per_element // m.groups15    # total_ops = batch_size * output_elements * (cin * kernel_ops // m.groups + bias_ops)16    m.total_ops = torch.Tensor([int(total_ops)])17def count_conv2d(m, x, y):18    x = x[0]19    cin = m.in_channels20    cout = m.out_channels21    kh, kw = m.kernel_size22    batch_size = x.size()[0]23    out_h = y.size(2)24    out_w = y.size(3)25    # ops per output element26    # kernel_mul = kh * kw * cin27    # kernel_add = kh * kw * cin - 128    kernel_ops = multiply_adds * kh * kw29    bias_ops = 1 if m.bias is not None else 030    ops_per_element = kernel_ops + bias_ops31    # total ops32    # num_out_elements = y.numel()33    output_elements = batch_size * out_w * out_h * cout34    total_ops = output_elements * ops_per_element * cin // m.groups35    m.total_ops = torch.Tensor([int(total_ops)])36def count_convtranspose2d(m, x, y):37    x = x[0]38    cin = m.in_channels39    cout = m.out_channels40    kh, kw = m.kernel_size41    batch_size = x.size()[0]42    out_h = y.size(2)43    out_w = y.size(3)44    # ops per output element45    # kernel_mul = kh * kw * cin46    # kernel_add = kh * kw * cin - 147    kernel_ops = multiply_adds * kh * kw * cin // m.groups48    bias_ops = 1 if m.bias is not None else 049    ops_per_element = kernel_ops + bias_ops50    # total ops51    # num_out_elements = y.numel()52    # output_elements = batch_size * out_w * out_h * cout53    ops_per_element = m.weight.nelement()54    output_elements = y.nelement()55    total_ops = output_elements * ops_per_element56    m.total_ops = torch.Tensor([int(total_ops)])57def count_bn(m, x, y):58    x = x[0]59    nelements = x.numel()60    # subtract, divide, gamma, beta61    total_ops = 4 * nelements62    m.total_ops = torch.Tensor([int(total_ops)])63def count_relu(m, x, y):64    x = x[0]65    nelements = x.numel()66    total_ops = nelements67    m.total_ops = torch.Tensor([int(total_ops)])68def count_sigmoid(m, x, y):69    x = x[0]70    nelements = x.numel()71    total_exp = nelements72    total_add = nelements73    total_div = nelements74    total_ops = total_exp + total_add + total_div75    m.total_ops = torch.Tensor([int(total_ops)])76def count_pixelshuffle(m, x, y):77    x = x[0]78    nelements = x.numel()79    total_ops = nelements80    m.total_ops = torch.Tensor([int(total_ops)])81def count_softmax(m, x, y):82    x = x[0]83    batch_size, nfeatures = x.size()84    total_exp = nfeatures85    total_add = nfeatures - 186    total_div = nfeatures87    total_ops = batch_size * (total_exp + total_add + total_div)88    m.total_ops = torch.Tensor([int(total_ops)])89def count_maxpool(m, x, y):90    kernel_ops = torch.prod(torch.Tensor([m.kernel_size]))91    num_elements = y.numel()92    total_ops = kernel_ops * num_elements93    m.total_ops = torch.Tensor([int(total_ops)])94def count_adap_maxpool(m, x, y):95    kernel = torch.Tensor([*(x[0].shape[2:])]) // torch.Tensor(list((m.output_size,))).squeeze()96    kernel_ops = torch.prod(kernel)97    num_elements = y.numel()98    total_ops = kernel_ops * num_elements99    m.total_ops = torch.Tensor([int(total_ops)])100def count_avgpool(m, x, y):101    total_add = torch.prod(torch.Tensor([m.kernel_size]))102    total_div = 1103    kernel_ops = total_add + total_div104    num_elements = y.numel()105    total_ops = kernel_ops * num_elements106    m.total_ops = torch.Tensor([int(total_ops)])107def count_adap_avgpool(m, x, y):108    kernel = torch.Tensor([*(x[0].shape[2:])]) // torch.Tensor(list((m.output_size,))).squeeze()109    total_add = torch.prod(kernel)110    total_div = 1111    kernel_ops = total_add + total_div112    num_elements = y.numel()113    total_ops = kernel_ops * num_elements114    m.total_ops = torch.Tensor([int(total_ops)])115def count_linear(m, x, y):116    # per output element117    total_mul = m.in_features118    total_add = m.in_features - 1119    num_elements = y.numel()120    total_ops = (total_mul + total_add) * num_elements...

config.js

Source:config.js

1/**2 * Build widget configuration3 */4(function () {5    'use strict';6    angular.module('devops-dashboard').controller('ChatOpsConfigController',7        ChatOpsConfigController);8    ChatOpsConfigController.$inject = ['modalData', '$modalInstance',9        'collectorData'];10    function ChatOpsConfigController(modalData, $modalInstance, collectorData) {11        var ctrl = this;12        var widgetConfig = modalData.widgetConfig;13        ctrl.chatOpsOptions = [{14            name: 'HipChat',15            value: 'HipChat'16        }, {17            name: 'Slack',18            value: 'Slack (Not implemented)'19        }, {20            name: 'Gitter',21            value: 'Gitter (Not implemented)'22        }];23        if (!widgetConfig.options.chatOpsOption) {24            ctrl.chatOpsOption = "";25        }26        else {27            var myindex;28            for (var v = 0; v < ctrl.chatOpsOptions.length; v++) {29                if (ctrl.chatOpsOptions[v].name == widgetConfig.options.chatOpsOption.name) {30                    myindex = v;31                    break;32                }33            }34            ctrl.chatOpsOption = ctrl.chatOpsOptions[myindex];35        }36        ctrl.chatOpsRoomName=widgetConfig.options.chatOpsRoomName;37        ctrl.chatOpsRoomAuthToken=widgetConfig.options.chatOpsRoomAuthToken;38        ctrl.chatOpsServerUrl=widgetConfig.options.chatOpsServerUrl;39        // public variables40        ctrl.submitted = false;41        ctrl.collectors = [];42        // public methods43        ctrl.submit = submitForm;44        // Request collecters45        collectorData.collectorsByType('ChatOps').then(processCollectorsResponse);46        function processCollectorsResponse(data) {47            console.log(data);48            ctrl.collectors = data;49        }50        /*51         * function submitForm(valid, url) { ctrl.submitted = true; if (valid &&52         * ctrl.collectors.length) {53         * createCollectorItem(url).then(processCollectorItemResponse); } }54         */55        function submitForm(valid, chatOpsOption, chatOpsRoomAuthToken, chatOpsServerUrl, chatOpsRoomName) {56            ctrl.submitted = true;57            if (valid && ctrl.collectors.length) {58                    createCollectorItem(chatOpsOption, chatOpsRoomAuthToken, chatOpsServerUrl, chatOpsRoomName).then(59                        processCollectorItemResponse);60                }61            }62        function createCollectorItem(chatOpsOption, chatOpsRoomAuthToken, chatOpsServerUrl, chatOpsRoomName) {63            var item = {64                    collectorId: _.findWhere(ctrl.collectors, {name: 'ChatOps'}).id,65                    options: {66                        chatOpsOption: chatOpsOption,67                        chatOpsRoomAuthToken: chatOpsRoomAuthToken,68                        chatOpsServerUrl: chatOpsServerUrl,69                        chatOpsRoomName: chatOpsRoomName70                    }71                };72            return collectorData.createCollectorItem(item);73        }74        function processCollectorItemResponse(response) {75            var postObj = {76                name: "ChatOps",77                options: {78                    id: widgetConfig.options.id,79                    chatOpsOption: ctrl.chatOpsOption,80                    chatOpsRoomName:ctrl.chatOpsRoomName,81                    chatOpsRoomAuthToken: ctrl.chatOpsRoomAuthToken,82                    chatOpsServerUrl: ctrl.chatOpsServerUrl83                },84                componentId: modalData.dashboard.application.components[0].id,85                collectorItemId: response.data.id86            };87            console.log("PostObject is:"+JSON.stringify(postObj));88            // pass this new config to the modal closing so it's saved89            $modalInstance.close(postObj);90        }91    }...

Automation Testing Tutorials

Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.