Changeset: 461a7eb33720 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=461a7eb33720
Modified Files:
monetdb5/extras/pyapi/Benchmarks/graph.py
monetdb5/extras/pyapi/Benchmarks/monetdb_testing.py
monetdb5/extras/pyapi/Benchmarks/pyapi_test.sh
Branch: pyapi
Log Message:
Slightly redesign testing and add pyfits benchmark.
diffs (truncated from 649 to 300 lines):
diff --git a/monetdb5/extras/pyapi/Benchmarks/graph.py
b/monetdb5/extras/pyapi/Benchmarks/graph.py
--- a/monetdb5/extras/pyapi/Benchmarks/graph.py
+++ b/monetdb5/extras/pyapi/Benchmarks/graph.py
@@ -21,7 +21,7 @@ y_log = False
line_plot = False
fill_time = False
data_start = 3
-graph_colors = ['blue', 'green', 'red', 'cyan', 'magenta', 'yellow', 'black',
'white', 'orange', 'darkgreen', 'aliceblue', 'darkgoldenrod', 'darkorchid',
'darkred', 'forestgreen', 'lavenderblush', 'lightsalmon', 'midnightblue',
'moccasin', 'papayawhip', 'turquoise', 'violet']
+graph_colors = ['blue', 'green', 'red', 'cyan', 'magenta', 'yellow', 'black',
'white', 'orange', 'darkgreen', 'darkgray', 'gold', 'darkorchid', 'darkred',
'violet', 'lavenderblush', 'lightsalmon', 'midnightblue', 'moccasin',
'papayawhip', 'turquoise', 'violet']
for i in range(3, len(arguments)):
if '-xlog' in arguments[i]: x_log = True
elif '-ylog' in arguments[i]: y_log = True
diff --git a/monetdb5/extras/pyapi/Benchmarks/monetdb_testing.py
b/monetdb5/extras/pyapi/Benchmarks/monetdb_testing.py
--- a/monetdb5/extras/pyapi/Benchmarks/monetdb_testing.py
+++ b/monetdb5/extras/pyapi/Benchmarks/monetdb_testing.py
@@ -127,6 +127,8 @@ if len(split) > 1 and split[1].lower() =
hot_test = False
args_test_type = split[0]
+print("Start test %s-%s" % (args_input_database, args_test_type))
+
drop_cache = os.environ["DROP_CACHE_COMMAND"]
def drop_all_caches():
array = numpy.zeros(100000)
@@ -557,14 +559,18 @@ else:
for i in range(0, 2):
database_execute()
for i in range(0,test_count):
- if not hot_test: drop_all_caches() #drop caches everytime for
cold tests
+ if not hot_test:
+ drop_all_caches() #drop caches everytime for cold tests
+ database_final()
+ os.system("%s %s %s %s" % (input_dir + "/randomstrings",
size, input_type, input_file))
+ database_load()
start = time.time()
database_execute()
end = time.time()
f.write(format_output(size, end - start))
database_clear()
f.close()
- database_final
+ database_final()
input_type = "integer"
function_name = str(args_test_type).lower()
@@ -579,29 +585,34 @@ else:
return numpy.min(numpy.sqrt(numpy.abs(a)))
function = sqroot
return_value = "double precision"
+ if function_name == "quantile":
+ def quantile(a):
+ return numpy.percentile(a, 50)
+ function = quantile
+ return_value = "integer"
import inspect
inspect_result = inspect.getsourcelines(function)
source_code = "".join([" " + x.lstrip() for x in inspect_result[0][1:]])
- if str(args_input_database).lower() == "postgres":
+ if str(args_input_database).lower() == "plpython" or
str(args_input_database).lower() == "postgres":
client = os.environ["POSTGRES_CLIENT_COMMAND"]
dropdb = os.environ["POSTGRES_DROPDB_COMMAND"]
initdb = os.environ["POSTGRES_CREATEDB_COMMAND"]
+ input_dir = os.environ["POSTGRES_CWD"]
function_name = str(args_test_type).lower()
+ createdb_file = "%s/%s.createdb.sql" % (input_dir, function_name)
+ run_file = "%s/%s.sql" % (input_dir, function_name)
def postgres_init():
- input_dir = os.environ["POSTGRES_CWD"]
+ if str(args_input_database).lower() == "plpython":
+ if function_name == "identity":
+ source = " return a"
+ elif function_name == "sqroot":
+ source = " import math\n return math.sqrt(abs(a))"
+ else: raise Exception("Unsupported function %s" %
function_name)
- createdb_file = "%s/%s.createdb.sql" % (input_dir, function_name)
- run_file = "%s/%s.sql" % (input_dir, function_name)
-
- if function_name == "identity":
- source = " return a"
- if function_name == "sqroot":
- source = " import math\n return math.sqrt(abs(a))"
-
- createdb_sql = """
+ createdb_sql = """
CREATE TABLE integers(i integer);
COPY integers FROM '%s' DELIMITER ',' CSV;
@@ -612,9 +623,18 @@ AS $$
%s
$$ LANGUAGE plpythonu;""" % (input_file, function_name, return_value, source)
- run_sql = """
- SELECT MIN(%s(i)) FROM integers;
- """ % function_name
+ run_sql = """
+ SELECT MIN(%s(i)) FROM integers;
+ """ % function_name
+ elif str(args_input_database).lower() == "postgres":
+ if function_name == "quantile":
+ run_sql = "SELECT percentile_cont(0.5) WITHIN GROUP(ORDER
BY i) FROM integers;"
+ else: raise Exception("Unsupported function %s" %
function_name)
+ createdb_sql = """
+ CREATE TABLE integers(i integer);
+
+ COPY integers FROM '%s' DELIMITER ',' CSV;""" % input_file
+
createdb = open(createdb_file, 'w+')
createdb.write(createdb_sql)
@@ -641,7 +661,7 @@ AS $$
execute_test(input_type, postgres_init, postgres_load,
postgres_execute, postgres_clear, postgres_final)
elif str(args_input_database).lower() == "sqlitemem" or
str(args_input_database).lower() == "sqlitedb":
- import csv, sqlite3
+ import sqlite3
database_file = os.environ["SQLITE_DB_FILE"]
database_name = ":memory:" if str(args_input_database).lower() ==
"sqlitemem" else database_file
@@ -674,7 +694,7 @@ AS $$
conn.close()
execute_test(input_type, sqlite_init, sqlite_load, sqlite_execute,
sqlite_clear, sqlite_final)
- elif str(args_input_database).lower() == "monetdbmapi" or
str(args_input_database).lower() == "pyapi" or str(args_input_database).lower()
== "pyapimap" or str(args_input_database).lower() == "rapi":
+ elif str(args_input_database).lower() == "monetdbmapi" or
str(args_input_database).lower() == "pyapi" or str(args_input_database).lower()
== "pyapimap" or str(args_input_database).lower() == "rapi" or
str(args_input_database).lower() == "monetdb":
import monetdb.sql
for i in range(0, max_retries):
try:
@@ -701,18 +721,26 @@ AS $$
elif str(args_input_database).lower() == "rapi":
if function_name == "identity":
c.execute("CREATE FUNCTION FUNC_%s(a integer) RETURNS %s
LANGUAGE R { min(a) };" % (function_name,return_value))
- if function_name == "sqroot":
+ elif function_name == "sqroot":
c.execute("CREATE FUNCTION FUNC_%s(a integer) RETURNS %s
LANGUAGE R { min(sqrt(abs(a))) };" % (function_name,return_value))
+ elif function_name == "quantile":
+ c.execute("CREATE FUNCTION FUNC_%s(a integer) RETURNS %s
LANGUAGE R { as.integer(quantile(a,0.5)) };" % (function_name,return_value))
+ else: raise Exception("Unsupported function %s" %
function_name)
if str(args_input_database).lower() == "pyapi" or
str(args_input_database).lower() == "pyapimap" or
str(args_input_database).lower() == "rapi":
def monetdb_execute():
c.execute("SELECT FUNC_%s(i) FROM integers;" % function_name)
result = c.fetchall()
- else:
+ elif str(args_input_database).lower() == "monetdbmapi":
def monetdb_execute():
c.execute('SELECT * FROM integers')
result = c.fetchall()
function(numpy.array(result, dtype=numpy.int32))
+ elif str(args_input_database).lower() == "monetdb":
+ def monetdb_execute():
+ if function_name == "quantile": c.execute("SELECT quantile(i,
0.5) FROM integers;")
+ else: raise Exception("Unsupported function %s" %
function_name)
+ result = c.fetchall()
def monetdb_clear():
c.execute("DROP TABLE integers;")
@@ -743,7 +771,7 @@ AS $$
def psycopg2_execute():
c.execute("SELECT * FROM integers;")
result = c.fetchall()
- print function(numpy.array(result, dtype=numpy.int32))
+ function(numpy.array(result, dtype=numpy.int32))
def psycopg2_clear():
c.execute("DROP TABLE integers;")
@@ -755,8 +783,7 @@ AS $$
execute_test(input_type, psycopg2_init, psycopg2_load,
psycopg2_execute, psycopg2_clear, psycopg2_final)
elif str(args_input_database).lower() == "pytables":
- import tables
- import csv
+ import tables, pandas as pd
table_file = 'testfile.h5'
@@ -771,12 +798,9 @@ AS $$
group = file.create_group('/', 'integers', 'integer_data')
table = file.create_table(group, 'values', description, "example")
values = table.row
- with open(input_file, 'rb') as csvfile:
- reader = csv.reader(csvfile)
- result = [x for x in reader]
- for x in result:
- values['i'] = int(x[0])
- values.append()
+ for x in pd.read_csv(input_file).values:
+ values['i'] = int(x)
+ values.append()
table.flush()
file.close()
@@ -814,18 +838,60 @@ AS $$
os.remove(input_file)
execute_test(input_type, csv_init, csv_load, csv_execute, csv_clear,
csv_final)
+ elif str(args_input_database).lower() == "pandascsv":
+ import pandas as pd
+ def csv_init():
+ return None
+
+ def csv_load():
+ return None
+
+ def csv_execute():
+ a = pd.read_csv(input_file)
+ function(a.values)
+
+ def csv_clear():
+ return None
+
+ def csv_final():
+ os.remove(input_file)
+
+ execute_test(input_type, csv_init, csv_load, csv_execute, csv_clear,
csv_final)
+ elif str(args_input_database).lower() == "pyfits":
+ import pyfits, pandas as pd
+ hdu_file = 'tempfile.fits'
+ def pyfits_init():
+ return None
+
+ def pyfits_load():
+ try: os.remove(hdu_file)
+ except: pass
+ array = pd.read_csv(input_file).values
+ numpy_array = numpy.array(array, dtype=numpy.int32)
+ hdu = pyfits.PrimaryHDU(numpy_array)
+ hdu.writeto(hdu_file)
+
+ def pyfits_execute():
+ hdulist = pyfits.open(hdu_file)
+ function(hdulist[0].data)
+
+ def pyfits_clear():
+ return None
+
+ def pyfits_final():
+ os.remove(input_file)
+ os.remove(hdu_file)
+
+ execute_test(input_type, pyfits_init, pyfits_load, pyfits_execute,
pyfits_clear, pyfits_final)
elif str(args_input_database).lower() == "numpybinary":
- import csv, numpy
+ import pandas as pd, numpy
numpy_binary = 'tempfile.npy'
def numpy_init():
return None
def numpy_load():
- with open(input_file, 'rb') as csvfile:
- reader = csv.reader(csvfile)
- result = [int(x[0]) for x in reader]
- numpy_array = numpy.array(result, dtype=numpy.int32)
- numpy.save(numpy_binary, numpy_array)
+ array = pd.read_csv(input_file).values
+ numpy.save(numpy_binary, numpy.array(array, dtype=numpy.int32))
def numpy_execute():
numpy_array = numpy.load(numpy_binary)
@@ -835,12 +901,12 @@ AS $$
return None
def numpy_final():
+ os.remove(numpy_binary)
os.remove(input_file)
- os.remove(numpy_binary)
execute_test(input_type, numpy_init, numpy_load, numpy_execute,
numpy_clear, numpy_final)
elif str(args_input_database).lower() == "castra":
- import castra, csv, shutil, pandas as pd
+ import castra, shutil, pandas as pd
castra_binary = 'data.castra'
def castra_init():
return None
@@ -856,7 +922,7 @@ AS $$
def castra_execute():
c = castra.Castra(castra_binary, readonly=True)
- print function(c[:, 'i'].values)
+ function(c[:, 'i'].values)
def castra_clear():
shutil.rmtree(castra_binary)
@@ -866,17 +932,15 @@ AS $$
execute_test(input_type, castra_init, castra_load, castra_execute,
castra_clear, castra_final)
elif str(args_input_database).lower() == "numpymemorymap":
- import csv, numpy
+ import pandas as pd, numpy
numpy_binary = 'tempfile.npy'
def numpy_init():
return None
def numpy_load():
- with open(input_file, 'rb') as csvfile:
- reader = csv.reader(csvfile)
- result = [int(x[0]) for x in reader]
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
