2 changed files with 0 additions and 223 deletions
-
112benchmarks/benchmark-plotters/plot-perf-enginelocation.py
-
111benchmarks/benchmark-plotters/plot-perf-peakthroughput.py
@ -1,112 +0,0 @@ |
|||
import os |
|||
import json |
|||
import pandas as pd |
|||
import seaborn as sns |
|||
import matplotlib.pyplot as plt |
|||
|
|||
from common import calc_throughput, index_from_element |
|||
|
|||
runid = "Run ID" |
|||
x_label = "Copy Type" |
|||
y_label = "Throughput in GiB/s" |
|||
var_label = "Configuration" |
|||
types = ["intersock-n0ton4-1mib", "internode-n0ton1-1mib", "intersock-n0ton4-1gib", "internode-n0ton1-1gib"] |
|||
types_nice = ["Inter-Socket 1MiB", "Inter-Node 1MiB", "Inter-Socket 1GiB", "Inter-Node 1GiB"] |
|||
copy_methods = ["dstcopy", "srccopy", "xcopy", "srcoutsidercopy", "dstoutsidercopy", "sockoutsidercopy", "nodeoutsidercopy"] |
|||
copy_methods_nice = [ "Engine on DST-Node", "Engine on SRC-Node", "Cross-Copy / Both Engines", "Engine on SRC-Socket, not SRC-Node", "Engine on DST-Socket, not DST-Node", "Engine on different Socket", "Engine on same Socket"] |
|||
|
|||
title = \ |
|||
"""Throughput showing impact of Engine Location\n |
|||
Copy Operation on DDR with 1 Engine per WQ""" |
|||
|
|||
description = \ |
|||
"""Throughput showing impact of Engine Location\n |
|||
Some Configurations missing as they are not feesible\n |
|||
Copy Operation on DDR with 1 Engine per WQ""" |
|||
|
|||
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 load_time_mesurements(file_path,method_label): |
|||
with open(file_path, 'r') as file: |
|||
data = json.load(file) |
|||
iterations = data["list"][0]["task"]["iterations"] |
|||
|
|||
if method_label == "xcopy": |
|||
# xcopy runs on two engines that both copy 1/2 of the entire |
|||
# specified size of 1gib, therefore the maximum time between |
|||
# these two is going to be the total time for copy |
|||
|
|||
time0 = data["list"][0]["report"]["time"] |
|||
time1 = data["list"][1]["report"]["time"] |
|||
|
|||
return { |
|||
"total": max(time0["total"],time1["total"]) / iterations, |
|||
"combined" : [max(x,y) for x,y in zip(time0["combined"], time1["combined"])], |
|||
"submission" : [max(x,y) for x,y in zip(time0["completion"], time1["completion"])], |
|||
"submission" : [max(x,y) for x,y in zip(time0["completion"], time1["completion"])], |
|||
} |
|||
|
|||
else: |
|||
return { |
|||
"total": data["list"][0]["report"]["time"]["total"] / iterations, |
|||
"combined": data["list"][0]["report"]["time"]["combined"], |
|||
"submission": data["list"][0]["report"]["time"]["submission"], |
|||
"completion": data["list"][0]["report"]["time"]["completion"] |
|||
} |
|||
|
|||
|
|||
# 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 create_copy_dataset(file_path, method_label, type_label): |
|||
method_index = index_from_element(method_label,copy_methods) |
|||
method_nice = copy_methods_nice[method_index] |
|||
type_index = index_from_element(type_label, types) |
|||
type_nice = types_nice[type_index] |
|||
data_size = 0 |
|||
|
|||
|
|||
if type_label in ["internode-n0ton1-1gib", "intersock-n0ton4-1gib"]: data_size = 1024*1024*1024 |
|||
elif type_label in ["internode-n0ton1-1mib", "intersock-n0ton4-1mib"]: data_size = 1024 * 1024 |
|||
else: data_size = 0 |
|||
|
|||
try: |
|||
run_idx = 0 |
|||
time = [load_time_mesurements(file_path,method_label)["total"]] |
|||
for t in time: |
|||
data.append({ runid : run_idx, x_label: type_nice, var_label : method_nice, y_label : calc_throughput(data_size, t)}) |
|||
run_idx = run_idx + 1 |
|||
except FileNotFoundError: |
|||
return |
|||
|
|||
|
|||
# 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 method_label in copy_methods: |
|||
for type_label in types: |
|||
file = os.path.join(result_path, f"{method_label}-{type_label}-1e.json") |
|||
create_copy_dataset(file, method_label, type_label) |
|||
|
|||
df = pd.DataFrame(data) |
|||
df.set_index(index, inplace=True) |
|||
df = df.sort_values(y_label) |
|||
|
|||
sns.barplot(x=x_label, y=y_label, hue=var_label, data=df, palette="rocket", errorbar="sd") |
|||
|
|||
plt.title(title) |
|||
plt.savefig(os.path.join(output_path, "plot-perf-enginelocation.png"), bbox_inches='tight') |
|||
plt.show() |
|||
|
|||
if __name__ == "__main__": |
|||
main() |
|||
@ -1,111 +0,0 @@ |
|||
import os |
|||
import json |
|||
import pandas as pd |
|||
from itertools import chain |
|||
import seaborn as sns |
|||
import matplotlib.pyplot as plt |
|||
|
|||
from common import calc_throughput |
|||
|
|||
result_path = "benchmark-results/" |
|||
output_path = "benchmark-plots/" |
|||
|
|||
runid = "Run ID" |
|||
x_label = "Destination Node" |
|||
y_label = "Source Node" |
|||
v_label = "Throughput" |
|||
|
|||
title_allnodes = \ |
|||
"""Copy Throughput in GiB/s tested for 1GiB Elements\n |
|||
Using all 8 DSA Chiplets available on the System""" |
|||
title_smartnodes = \ |
|||
"""Copy Throughput in GiB/s tested for 1GiB Elements\n |
|||
Using Cross-Copy for Intersocket and all 4 Chiplets of Socket for Intrasocket""" |
|||
title_difference = \ |
|||
"""Gain in Copy Throughput in GiB/s of All-DSA vs. Smart Assignment""" |
|||
|
|||
description_smartnodes = \ |
|||
"""Copy Throughput in GiB/s tested for 1GiB Elements\n |
|||
Nodes of {8...15} are HBM accessors for their counterparts (minus 8)\n |
|||
Using all 4 DSA Chiplets of a Socket for Intra-Socket Operation\n |
|||
And using only the Source and Destination Nodes DSA for Inter-Socket""" |
|||
description_allnodes = \ |
|||
"""Copy Throughput in GiB/s tested for 1GiB Elements\n |
|||
Nodes of {8...15} are HBM accessors for their counterparts (minus 8)\n |
|||
Using all 8 DSA Chiplets available on the System""" |
|||
|
|||
index = [ runid, x_label, y_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 load_time_mesurements(file_path): |
|||
with open(file_path, 'r') as file: |
|||
data = json.load(file) |
|||
count = data["count"] |
|||
batch_size = data["list"][0]["task"]["batching"]["batch_size"] if data["list"][0]["task"]["batching"]["batch_size"] > 0 else 1 |
|||
iterations = data["list"][0]["task"]["iterations"] |
|||
|
|||
return { |
|||
"size": data["list"][0]["task"]["size"], |
|||
"total": sum([x / (iterations * batch_size * count * count) for x in list(chain([data["list"][i]["report"]["time"]["total"] for i in range(count)]))]), |
|||
"combined": [ x / (count * batch_size) for x in list(chain(*[data["list"][i]["report"]["time"]["combined"] for i in range(count)]))], |
|||
"submission": [ x / (count * batch_size) for x in list(chain(*[data["list"][i]["report"]["time"]["submission"] for i in range(count)]))], |
|||
"completion": [ x / (count * batch_size) for x in list(chain(*[data["list"][i]["report"]["time"]["completion"] for i in range(count)]))] |
|||
} |
|||
|
|||
|
|||
# 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, src_node, dst_node): |
|||
try: |
|||
file_data = load_time_mesurements(file_path) |
|||
time = [file_data["total"]] |
|||
run_idx = 0 |
|||
for t in time: |
|||
data.append({ runid : run_idx, x_label : dst_node, y_label : src_node, v_label: calc_throughput(file_data["size"], t)}) |
|||
run_idx = run_idx + 1 |
|||
except FileNotFoundError: |
|||
return |
|||
|
|||
|
|||
def plot_heatmap(table,title,node_config): |
|||
plt.figure(figsize=(8, 6)) |
|||
|
|||
sns.heatmap(table, annot=True, cmap="rocket_r", fmt=".0f") |
|||
|
|||
plt.title(title) |
|||
plt.savefig(os.path.join(output_path, f"plot-perf-{node_config}-throughput.png"), bbox_inches='tight') |
|||
plt.show() |
|||
|
|||
|
|||
# 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(node_config,title): |
|||
for src_node in range(16): |
|||
for dst_node in range(16): |
|||
size = "512mib" if node_config == "allnodes" and src_node == dst_node and src_node >= 8 else "1gib" |
|||
file = os.path.join(result_path, f"copy-n{src_node}ton{dst_node}-{size}-{node_config}-1e.json") |
|||
process_file_to_dataset(file, src_node, dst_node) |
|||
|
|||
df = pd.DataFrame(data) |
|||
|
|||
data.clear() |
|||
df.set_index(index, inplace=True) |
|||
data_pivot = df.pivot_table(index=y_label, columns=x_label, values=v_label) |
|||
|
|||
plot_heatmap(data_pivot, title, node_config) |
|||
|
|||
return data_pivot |
|||
|
|||
|
|||
if __name__ == "__main__": |
|||
dall = main("allnodes", title_allnodes) |
|||
dsmart = main("smart", title_smartnodes) |
|||
ddiff = dall - dsmart |
|||
plot_heatmap(ddiff,title_difference,"diff") |
|||
Write
Preview
Loading…
Cancel
Save
Reference in new issue