This contains my bachelors thesis and associated tex files, code snippets and maybe more. Topic: Data Movement in Heterogeneous Memories with Intel Data Streaming Accelerator
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 

91 lines
2.9 KiB

import os
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
from common import calc_throughput, index_from_element, load_time_mesurements
runid = "Run ID"
x_label = "Thread Count"
y_label = "Throughput in GiB/s"
var_label = "Transfer Size"
thread_counts = ["1t", "2t", "12t"]
thread_counts_nice = ["1 Thread", "2 Threads", "12 Threads"]
size_labels = ["1mib", "1gib"]
size_labels_nice = ["1 MiB", "1 GiB"]
title = \
"""Total Throughput showing cost of MT Submit\n
Copying 120x split on n Threads Intra-Node on DDR\n
"""
description = \
"""Total Throughput showing cost of MT Submit\n
Running 120 Copy Operations split on n Threads\n
Copying Intra-Node on DDR performed for multiple Configurations\n
"""
index = [runid, x_label, var_label]
data = []
# loads the measurements from a given file and processes them
# so that they are normalized, meaning that the timings returned
# are nanoseconds per element transfered
def get_timing(file_path,thread_count):
divisor = 0
if thread_count == "1t": divisor = 1
elif thread_count == "2t" : divisor = 2
elif thread_count == "12t" : divisor = 12
return [ x / divisor for x in load_time_mesurements(file_path)]
# procceses a single file and appends the desired timings
# to the global data-array, handles multiple runs with a runid
# and ignores if the given file is not found as some
# configurations may not be benchmarked
def process_file_to_dataset(file_path, size_label, thread_count):
size_index = index_from_element(size_label,size_labels)
size_nice = size_labels_nice[size_index]
threadc_index = index_from_element(thread_count, thread_counts)
thread_count_nice = thread_counts_nice[threadc_index]
data_size = 0
if size_label == "1gib" : data_size = 1024*1024*1024
elif size_label == "1mib" : data_size = 1024*1024
timing = get_timing(file_path, thread_count)
run_idx = 0
for t in timing:
data.append({ runid : run_idx, x_label: thread_count_nice, var_label : size_nice, y_label : calc_throughput(data_size, t)})
run_idx = run_idx + 1
# loops over all possible configuration combinations and calls
# process_file_to_dataset for them in order to build a dataframe
# which is then displayed and saved
def main():
result_path = "benchmark-results/"
output_path = "benchmark-plots/"
for size in size_labels:
for thread_count in thread_counts:
file = os.path.join(result_path, f"mtsubmit-{thread_count}-{size}.json")
process_file_to_dataset(file, size, thread_count)
df = pd.DataFrame(data)
df.set_index(index, inplace=True)
plt.figure(figsize=(4, 4))
plt.ylim(0, 30)
sns.barplot(x=x_label, y=y_label, hue=var_label, data=df, palette="mako", errorbar="sd")
plt.savefig(os.path.join(output_path, "plot-mtsubmit.pdf"), bbox_inches='tight')
plt.show()
if __name__ == "__main__":
main()