def compare_pig_results(spreadsheets):
spreadsheet_name = []
values = []
results = []
test_name = "pig"
for spreadsheetId in spreadsheets:
values.append(read_sheet(spreadsheetId, range=test_name))
spreadsheet_name.append(
get_sheet(spreadsheetId, range=[])["properties"]["title"]
)
spreadsheet_name = " vs ".join(spreadsheet_name)
for index, value in enumerate(values):
values[index] = (list(g) for k, g in groupby(value, key=lambda x: x != []) if k)
for value, ele in zip(values[0], values[1]):
results.append([""])
if value[0] == ele[0]:
results.append(value[0])
results = combine_two_array_alternating(results, value[1:], ele[1:])
spreadsheetId = create_spreadsheet(spreadsheet_name, test_name)
append_to_sheet(spreadsheetId, results, test_name)
graph_pig_data(spreadsheetId, test_name)
print(f"https://docs.google.com/spreadsheets/d/{spreadsheetId}")
return results
def compare_hammerdb_results(spreadsheets, test_name):
spreadsheet_name = []
values = []
results = []
for spreadsheetId in spreadsheets:
values.append(read_sheet(spreadsheetId, range=test_name))
spreadsheet_name.append(
get_sheet(spreadsheetId, range=[])["properties"]["title"])
spreadsheet_name = " vs ".join(spreadsheet_name)
for index, value in enumerate(values):
values[index] = (list(g)
for k, g in groupby(value, key=lambda x: x != [])
if k)
for value, ele in zip(values[0], values[1]):
results.append([""])
# Make sure it's same system family
# TODO: Maybe check for whole type instead of family
if value[0][1].split(".")[0] == ele[0][1].split(".")[0]:
results = combine_two_array_alternating(results, value, ele)
spreadsheetId = create_spreadsheet(spreadsheet_name, test_name)
append_to_sheet(spreadsheetId, results, test_name)
graph_hammerdb_data(spreadsheetId, test_name)
print(f"https://docs.google.com/spreadsheets/d/{spreadsheetId}")
return results
def process_results(results):
""""""
global test_name
spreadsheet_name = f"{config.cloud_type} {config.OS_TYPE}-{config.OS_RELEASE}"
if results:
if check_test_is_hammerdb(test_name):
results = create_summary_hammerdb_data(results)
else:
results = globals()[f"create_summary_{test_name}_data"](results)
if not config.spreadsheetId:
config.spreadsheetId = create_spreadsheet(spreadsheet_name,
test_name)
create_sheet(config.spreadsheetId, test_name)
append_to_sheet(config.spreadsheetId, results, test_name)
# Graphing up data
if check_test_is_hammerdb(test_name):
graph_hammerdb_data(config.spreadsheetId, test_name)
else:
globals()[f"graph_{test_name}_data"](config.spreadsheetId,
test_name)
return []
def compare_linpack_results(spreadsheets):
values = []
results = []
test_name = "linpack_8.2-8.3"
for spreadsheetId in spreadsheets:
values.append(read_sheet(spreadsheetId, range="linpack"))
for v in values[0]:
if v[0] == "System":
results.append([
"System",
"GFLOPS - 8.2",
"GFLOPS - 8.3",
"% Diff",
"GFLOP Scaling - 8.2",
"GFLOP Scaling - 8.3",
"Cost / hr",
"Price/Perf - 8.2",
"Price/Perf - 8.3",
"Price/Perf % Diff",
])
for y in values[1]:
if y[0] == "System":
continue
if v[0] in y[0]:
price_perf = []
perc_diff = (float(y[1]) - float(v[1])) / float(v[1])
price_perf.append(float(v[1]) / float(v[3]))
price_perf.append(float(y[1]) / float(y[3]))
price_perf_diff = (float(price_perf[1]) -
float(price_perf[0])) / float(price_perf[0])
results.append([
v[0],
v[1],
y[1],
perc_diff,
v[2],
y[2],
v[3],
price_perf[0],
price_perf[1],
price_perf_diff,
])
spreadsheetId = create_spreadsheet("Comparison 8.2 vs 8.3", test_name)
append_to_sheet(spreadsheetId, results, range=test_name)
graph_linpack_comparison(spreadsheetId, range=test_name)
print(f"https://docs.google.com/spreadsheets/d/{spreadsheetId}")
def compare_stream_results(spreadsheets,
test_name,
table_name=["Max Througput"]):
values = []
results = []
spreadsheet_name = []
for spreadsheetId in spreadsheets:
values.append(read_sheet(spreadsheetId, range=test_name))
spreadsheet_name.append(
get_sheet(spreadsheetId, range=[])["properties"]["title"])
spreadsheet_name = " vs ".join(spreadsheet_name)
for index, value in enumerate(values):
values[index] = (list(g)
for k, g in groupby(value, key=lambda x: x != [])
if k)
list_1 = list(values[0])
list_2 = list(values[1])
for value in list_1:
results.append([""])
for ele in list_2:
if value[0][0] in table_name and ele[0][0] in table_name:
results = combine_two_array_alternating(results, value, ele)
break
elif value[1][0] == ele[1][0]:
results.append(value[0])
results = combine_two_array_alternating(
results, value[1:], ele[1:])
break
spreadsheetId = create_spreadsheet(spreadsheet_name, test_name)
append_to_sheet(spreadsheetId, results, test_name)
graph_stream_data(spreadsheetId, test_name)
print(f"https://docs.google.com/spreadsheets/d/{spreadsheetId}")
return results
def graph_linpack_comparison(spreadsheetId, range="A:F"):
"""
Re-arrange data from the sheet into a dict grouped by machine name.
The sheet data & charts are then cleared excluding the header row.
The function then processes loops over each groups of machine and plots the
graphs.
Graphs:
- GFLOP and GFLOPS scaling
- Price/perf
:sheet: sheet API function
:spreadsheetId
:range: range to graph up the data, it will be mostly sheet name
"""
GFLOPS_PLOT_RANGE = "B"
PRICE_PER_PERF_RANGE = "H"
GRAPH_COL_INDEX = 5
GRAPH_ROW_INDEX = 0
data_dict = rearrange_linpack_data(spreadsheetId, range)
header_row = data_dict[0][0]
if data_dict:
clear_sheet_data(spreadsheetId, range)
clear_sheet_charts(spreadsheetId, range)
else:
raise Exception("Data sheet empty")
for data in data_dict:
machine_class = data[0][1].split(".")[0]
response = append_to_sheet(spreadsheetId, data, range)
updated_range = response["updates"]["updatedRange"]
title, sheet_range = updated_range.split("!")
sheet_range = sheet_range.split(":")
# apply_named_range(sheet, spreadsheetId, machine_class, updated_range)
sheetId = get_sheet(
spreadsheetId, updated_range)["sheets"][0]["properties"]["sheetId"]
# GFlops & GFlops scaling graph
requests = {
"addChart": {
"chart": {
"spec": {
"title":
"%s : %s and %s" %
(title, header_row[1], header_row[2]),
"basicChart": {
"chartType":
"COMBO",
"legendPosition":
"BOTTOM_LEGEND",
"axis": [
{
"position": "BOTTOM_AXIS",
"title": "%s" % (header_row[0]),
},
{
"position":
"LEFT_AXIS",
"title":
"%s, %s and %s, %s " % (
header_row[1],
header_row[2],
header_row[4],
header_row[5],
),
},
],
"domains": [{
"domain": {
"sourceRange": {
"sources": [{
"sheetId":
sheetId,
"startRowIndex":
int(sheet_range[0][1:]) - 1,
"endRowIndex":
sheet_range[1][1:],
"startColumnIndex":
0,
"endColumnIndex":
1,
}]
}
}
}],
"series": [
{
"series": {
"sourceRange": {
"sources": [{
"sheetId":
sheetId,
"startRowIndex":
int(sheet_range[0][1:]) - 1,
"endRowIndex":
sheet_range[1][1:],
"startColumnIndex":
ord(GFLOPS_PLOT_RANGE) % 65,
"endColumnIndex":
ord(GFLOPS_PLOT_RANGE) % 65 +
1,
}]
}
},
"targetAxis": "LEFT_AXIS",
"type": "COLUMN",
},
{
"series": {
"sourceRange": {
"sources": [{
"sheetId":
sheetId,
"startRowIndex":
int(sheet_range[0][1:]) - 1,
"endRowIndex":
sheet_range[1][1:],
"startColumnIndex":
ord(GFLOPS_PLOT_RANGE) % 65 +
1,
"endColumnIndex":
ord(GFLOPS_PLOT_RANGE) % 65 +
2,
}]
}
},
"targetAxis": "LEFT_AXIS",
"type": "COLUMN",
},
{
"series": {
"sourceRange": {
"sources": [{
"sheetId":
sheetId,
"startRowIndex":
int(sheet_range[0][1:]) - 1,
"endRowIndex":
sheet_range[1][1:],
"startColumnIndex":
ord(GFLOPS_PLOT_RANGE) % 65 +
3,
"endColumnIndex":
ord(GFLOPS_PLOT_RANGE) % 65 +
4,
}]
}
},
"targetAxis": "RIGHT_AXIS",
"type": "LINE",
},
{
"series": {
"sourceRange": {
"sources": [{
"sheetId":
sheetId,
"startRowIndex":
int(sheet_range[0][1:]) - 1,
"endRowIndex":
sheet_range[1][1:],
"startColumnIndex":
ord(GFLOPS_PLOT_RANGE) % 65 +
4,
"endColumnIndex":
ord(GFLOPS_PLOT_RANGE) % 65 +
5,
}]
}
},
"targetAxis": "RIGHT_AXIS",
"type": "LINE",
},
],
"headerCount":
1,
},
},
"position": {
"overlayPosition": {
"anchorCell": {
"sheetId": sheetId,
"rowIndex": GRAPH_ROW_INDEX,
"columnIndex":
ord(sheet_range[1][:1]) % 65 + 2,
}
}
},
}
}
}
# PRICE/PERF graph
body = {"requests": requests}
sheet.batchUpdate(spreadsheetId=spreadsheetId, body=body).execute()
requests = {
"addChart": {
"chart": {
"spec": {
"title": "%s : Price/Perf " % (title),
"basicChart": {
"chartType":
"COLUMN",
"legendPosition":
"BOTTOM_LEGEND",
"axis": [
{
"position": "BOTTOM_AXIS",
"title": "%s" % (header_row[0]),
},
{
"position": "LEFT_AXIS",
"title": "GFlops/$/hr "
},
],
"domains": [{
"domain": {
"sourceRange": {
"sources": [{
"sheetId":
sheetId,
"startRowIndex":
int(sheet_range[0][1:]) - 1,
"endRowIndex":
sheet_range[1][1:],
"startColumnIndex":
0,
"endColumnIndex":
1,
}]
}
}
}],
"series": [
{
"series": {
"sourceRange": {
"sources": [{
"sheetId":
sheetId,
"startRowIndex":
int(sheet_range[0][1:]) - 1,
"endRowIndex":
sheet_range[1][1:],
"startColumnIndex":
ord(PRICE_PER_PERF_RANGE) % 65,
"endColumnIndex":
ord(PRICE_PER_PERF_RANGE) % 65
+ 1,
}]
}
},
"targetAxis": "LEFT_AXIS",
"type": "COLUMN",
},
{
"series": {
"sourceRange": {
"sources": [{
"sheetId":
sheetId,
"startRowIndex":
int(sheet_range[0][1:]) - 1,
"endRowIndex":
sheet_range[1][1:],
"startColumnIndex":
ord(PRICE_PER_PERF_RANGE) % 65
+ 1,
"endColumnIndex":
ord(PRICE_PER_PERF_RANGE) % 65
+ 2,
}]
}
},
"targetAxis": "LEFT_AXIS",
"type": "COLUMN",
},
],
"headerCount":
1,
},
},
"position": {
"overlayPosition": {
"anchorCell": {
"sheetId": sheetId,
"rowIndex": GRAPH_ROW_INDEX,
"columnIndex":
ord(sheet_range[1][:1]) % 65 + 8,
}
}
},
}
}
}
body = {"requests": requests}
sheet.batchUpdate(spreadsheetId=spreadsheetId, body=body).execute()
GRAPH_ROW_INDEX += 20
