aboutsummaryrefslogtreecommitdiffstats
path: root/resources/tools/presentation/generator_tables.py
diff options
context:
space:
mode:
Diffstat (limited to 'resources/tools/presentation/generator_tables.py')
-rw-r--r--resources/tools/presentation/generator_tables.py84
1 files changed, 60 insertions, 24 deletions
diff --git a/resources/tools/presentation/generator_tables.py b/resources/tools/presentation/generator_tables.py
index 5efe66777b..cf8ac4598c 100644
--- a/resources/tools/presentation/generator_tables.py
+++ b/resources/tools/presentation/generator_tables.py
@@ -210,9 +210,9 @@ def table_performance_comparison(table, input_data):
header = ["Test case", ]
if table["include-tests"] == "MRR":
- hdr_param = "Receive Rate"
+ hdr_param = "Rec Rate"
else:
- hdr_param = "Throughput"
+ hdr_param = "Thput"
history = table.get("history", None)
if history:
@@ -235,6 +235,7 @@ def table_performance_comparison(table, input_data):
# Prepare data to the table:
tbl_dict = dict()
for job, builds in table["reference"]["data"].items():
+ topo = "2n-skx" if "2n-skx" in job else ""
for build in builds:
for tst_name, tst_data in data[job][str(build)].iteritems():
tst_name_mod = tst_name.replace("-ndrpdrdisc", "").\
@@ -385,6 +386,7 @@ def table_performance_comparison(table, input_data):
pass
tbl_lst = list()
+ footnote = False
for tst_name in tbl_dict.keys():
item = [tbl_dict[tst_name]["name"], ]
if history:
@@ -394,27 +396,28 @@ def table_performance_comparison(table, input_data):
item.append(round(mean(hist_data) / 1000000, 2))
item.append(round(stdev(hist_data) / 1000000, 2))
else:
- item.extend([None, None])
+ item.extend(["Not tested", "Not tested"])
else:
- item.extend([None, None])
+ item.extend(["Not tested", "Not tested"])
data_t = tbl_dict[tst_name]["ref-data"]
if data_t:
item.append(round(mean(data_t) / 1000000, 2))
item.append(round(stdev(data_t) / 1000000, 2))
else:
- item.extend([None, None])
+ item.extend(["Not tested", "Not tested"])
data_t = tbl_dict[tst_name]["cmp-data"]
if data_t:
item.append(round(mean(data_t) / 1000000, 2))
item.append(round(stdev(data_t) / 1000000, 2))
else:
- item.extend([None, None])
- if "dot1q" in tbl_dict[tst_name]["name"]:
- item.append("Changed methodology")
- elif item[-4] is not None and item[-2] is not None and item[-4] != 0:
+ item.extend(["Not tested", "Not tested"])
+ if item[-4] is not None and item[-2] is not None and item[-4] != 0:
item.append(int(relative_change(float(item[-4]), float(item[-2]))))
- else:
- item.append("n/a")
+ elif item[-4] is None or item[-2] is None or item[-4] == 0:
+ item.append("New in CSIT-1908")
+ elif topo == "2n-skx" and "dot1q" in tbl_dict[tst_name]["name"]:
+ item.append("See footnote [1]")
+ footnote = True
if (len(item) == len(header)) and (item[-3] is not None):
tbl_lst.append(item)
@@ -428,7 +431,22 @@ def table_performance_comparison(table, input_data):
for test in tbl_lst:
file_handler.write(",".join([str(item) for item in test]) + "\n")
- convert_csv_to_pretty_txt(csv_file, "{0}.txt".format(table["output-file"]))
+ txt_file_name = "{0}.txt".format(table["output-file"])
+ convert_csv_to_pretty_txt(csv_file, txt_file_name)
+
+ if footnote:
+ with open(txt_file_name, 'a') as txt_file:
+ txt_file.writelines([
+ "Footnotes:",
+ "[1] CSIT-1908 changed test methodology of dot1q tests in "
+ "2n-skx testbeds, dot1q encapsulation is now used on both "
+ "links of SUT.",
+ " Previously dot1q was used only on a single link with the "
+ "other link carrying untagged Ethernet frames. This change "
+ "results",
+ " in slightly lower throughput in CSIT-1908 for these "
+ "tests. See release notes."
+ ])
def table_performance_comparison_nic(table, input_data):
@@ -454,9 +472,9 @@ def table_performance_comparison_nic(table, input_data):
header = ["Test case", ]
if table["include-tests"] == "MRR":
- hdr_param = "Receive Rate"
+ hdr_param = "Rec Rate"
else:
- hdr_param = "Throughput"
+ hdr_param = "Thput"
history = table.get("history", None)
if history:
@@ -479,6 +497,7 @@ def table_performance_comparison_nic(table, input_data):
# Prepare data to the table:
tbl_dict = dict()
for job, builds in table["reference"]["data"].items():
+ topo = "2n-skx" if "2n-skx" in job else ""
for build in builds:
for tst_name, tst_data in data[job][str(build)].iteritems():
if table["reference"]["nic"] not in tst_data["tags"]:
@@ -635,6 +654,7 @@ def table_performance_comparison_nic(table, input_data):
pass
tbl_lst = list()
+ footnote = False
for tst_name in tbl_dict.keys():
item = [tbl_dict[tst_name]["name"], ]
if history:
@@ -644,27 +664,28 @@ def table_performance_comparison_nic(table, input_data):
item.append(round(mean(hist_data) / 1000000, 2))
item.append(round(stdev(hist_data) / 1000000, 2))
else:
- item.extend([None, None])
+ item.extend(["Not tested", "Not tested"])
else:
- item.extend([None, None])
+ item.extend(["Not tested", "Not tested"])
data_t = tbl_dict[tst_name]["ref-data"]
if data_t:
item.append(round(mean(data_t) / 1000000, 2))
item.append(round(stdev(data_t) / 1000000, 2))
else:
- item.extend([None, None])
+ item.extend(["Not tested", "Not tested"])
data_t = tbl_dict[tst_name]["cmp-data"]
if data_t:
item.append(round(mean(data_t) / 1000000, 2))
item.append(round(stdev(data_t) / 1000000, 2))
else:
- item.extend([None, None])
- if "dot1q" in tbl_dict[tst_name]["name"]:
- item.append("Changed methodology")
- elif item[-4] is not None and item[-2] is not None and item[-4] != 0:
+ item.extend(["Not tested", "Not tested"])
+ if item[-4] is not None and item[-2] is not None and item[-4] != 0:
item.append(int(relative_change(float(item[-4]), float(item[-2]))))
- else:
- item.append("n/a")
+ elif item[-4] is None or item[-2] is None or item[-4] == 0:
+ item.append("New in CSIT-1908")
+ elif topo == "2n-skx" and "dot1q" in tbl_dict[tst_name]["name"]:
+ item.append("See footnote [1]")
+ footnote = True
if (len(item) == len(header)) and (item[-3] is not None):
tbl_lst.append(item)
@@ -678,7 +699,22 @@ def table_performance_comparison_nic(table, input_data):
for test in tbl_lst:
file_handler.write(",".join([str(item) for item in test]) + "\n")
- convert_csv_to_pretty_txt(csv_file, "{0}.txt".format(table["output-file"]))
+ txt_file_name = "{0}.txt".format(table["output-file"])
+ convert_csv_to_pretty_txt(csv_file, txt_file_name)
+
+ if footnote:
+ with open(txt_file_name, 'a') as txt_file:
+ txt_file.writelines([
+ "Footnotes:",
+ "[1] CSIT-1908 changed test methodology of dot1q tests in "
+ "2n-skx testbeds, dot1q encapsulation is now used on both "
+ "links of SUT.",
+ " Previously dot1q was used only on a single link with the "
+ "other link carrying untagged Ethernet frames. This change "
+ "results",
+ " in slightly lower throughput in CSIT-1908 for these "
+ "tests. See release notes."
+ ])
def table_nics_comparison(table, input_data):