shape = m0.shape
ndims = len(spacing)
idx = 800 #shot[node_idx]
print("Process shot no.: ", idx)
dorig, sx, sz, gx, gz, tn, dt, nt = segy_get(bucket, data_path,
data_name + str(idx) + '.segy')
print("Shape: ", shape)
# Load previous iterations
if iteration == 1:
x = np.zeros(shape=shape, dtype='float32')
else:
x = array_get(
bucket,
variable_path + 'chunk_1/' + variable_name + str(iteration - 1))
if num_chunks > 1:
for chunk in range(1, num_chunks):
x_chunk = array_get(
bucket, variable_path + 'chunk_' + str(chunk + 1) + '/' +
variable_name + str(iteration - 1))
x = np.concatenate((x, x_chunk), axis=0)
x = x.reshape(shape[0], shape[1], order='F')
# Set up model structures
model = Model(shape=shape, origin=origin, spacing=spacing, vp=np.sqrt(1 / m0))
# Time axis
t0 = 0.
dt_comp = model.critical_dt
import numpy as np
import boto3, json, time, datetime, sys
import matplotlib.pyplot as plt
from pytz import timezone
from CloudExtras import array_get
# Read parameters
with open('parameters.json', 'r') as filename:
parameters = json.load(filename)
# EC2 bare metal results
key = 'timings_omp_bare_metal/strong_scaling_omp_bare_metal_numthreads_'
num_cores = np.array([1, 2, 4, 8, 16, 24])
num_files = 9
T = []
for i in range(num_files):
for run in range(3):
T = array_get(parameters['bucket_name'],
key + str(num_cores[i]) + '_run_' + str(run))
T.dump('results_bare_metal/timings_strong_scaling_omp_' +
str(num_cores[i]) + '_run_' + str(run))
break
time.sleep(10)
# Gather timings (unix time stamps)
print("Get timings")
T = np.zeros(shape=(3))
T[0] = array_job['createdAt']
T[1] = array_job['startedAt']
T[2] = array_job['stoppedAt']
# Get Devito runtimes and script runtimes
key = 'timings_hybrid/hybrid_scaling'
bucket = parameters['bucket_name']
while True:
try:
tcont = array_get(bucket, key)
break
except:
time.sleep(1)
T2 = np.zeros(len(tcont)) # should contain 3 values
for j in range(len(T2)):
T2[j] = tcont[j]
TT = np.concatenate(
(T, T2), axis=0
) # should be 4 + 3 values: creation, start, stop, gradient_timestamp, kernel time, python_runtime, devito_runtime
# Save results w/ current datetime stamp
timestamp = datetime.datetime.now()
timestamp = str(timestamp).replace(' ', '') # remove spaces
files_r5 = client.list_objects_v2(Bucket=bucket,
Prefix='pwitte/timings/r5/mpi_scaling')
num_timings_r5 = len(files_r5['Contents'])
file_list_r5 = []
for j in range(num_timings_r5):
file_list_r5.append(files_r5['Contents'][j]['Key'])
timings_r5 = []
for batchsize in num_instances:
run = 0
for filename in file_list_r5:
if filename[
0:int(41 + len(str(batchsize))
)] == 'pwitte/timings/r5/mpi_scaling_num_nodes_' + str(
batchsize) + '_':
t = array_get(bucket, filename)
t.dump('results/r5/timings_num_nodes_' + str(batchsize) + '_run_' +
str(run))
run += 1
files_c5n = client.list_objects_v2(Bucket=bucket,
Prefix='pwitte/timings/c5n/mpi_scaling')
num_timings_c5n = len(files_c5n['Contents'])
file_list_c5n = []
for j in range(num_timings_c5n):
file_list_c5n.append(files_c5n['Contents'][j]['Key'])
timings_c5n = []
for batchsize in num_instances:
run = 0
for filename in file_list_c5n:
