Did you know ... | Search Documentation: |
Pack logtalk -- logtalk-3.85.0/examples/benchmarks/NOTES.md |
This file is part of Logtalk https://logtalk.org/ SPDX-FileCopyrightText: 1998-2023 Paulo Moura <pmoura@logtalk.org> SPDX-License-Identifier: Apache-2.0
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
To load this example and for sample queries, please see the SCRIPT.txt
file.
This folder provides simple benchmark tests for comparing Logtalk message sending performance with direct predicates calls in plain Prolog. These benchmarks may also be used for comparing Logtalk message sending performance across Prolog compilers.
This example is made of four loader files and five source files:
loader_events.lgt
loads all source files with event support turned onloader_no_events.lgt
loads all source files with event support turned offloader.lgt
the same as the loader_no_events.lgt
filebenchmarks.lgt
contains the benchmark goals and testing predicatesplain.lgt
contains definitions for list length and list reverse predicates
and a predicate for testing performance of the built-in predicates
assertz/1 and retract/1module.pl
(not loaded by default; see below)
contains the same definitions of the list length and list reverse
predicates encapsulated in a moduleobjects.lgt
contains an object encapsulating the same definitions of list length
and list reverse predicates, plus two descendant objects to simulate
a small hierarchy (used for testing calls to imported category
predicates)database.lgt
and database_other.lgt
contains predicates for testing the performance of the built-in
database methods assertz/1 and retract/1category.lgt
contains a single predicate used when comparing performance of
calls to imported category predicates using direct calls and using
messages to "self"maze.lgt
contains a simple maze search problem using a depth-first strategy
with loop detection
You may have noticed above that the benchmark predicates and the predicates
for plain Prolog testing are both encapsulated in Logtalk source files. The
Logtalk compiler just copies the plain Prolog code to the generated Prolog
files. The reason for using the `.lgt` extension for these files is simply
to make it possible to load the example code using calls to the predicates
logtalk_load/1-2
.
By default, the benchmark tests on the SCRIPT.txt
file use a list of 30
elements as an argument to the list length and list reverse predicates. When
dynamic binding is used, increasing the list length results in decreasing
performance differences between plain Prolog and Logtalk as the length and
reverse computation times far outweigh the overhead of the message sending
mechanism. Likewise, decreasing the list length leads to increasing performance
differences between plain Prolog and Logtalk (up to the point you will be
measuring the Logtalk message sending mechanism overhead compared to plain
Prolog predicate calls). In real-life applications, only testing can give
you a balanced view on the trade-offs between plain Prolog performance and
Logtalk programming features.
The loader files load a module version of the tests when a backend Prolog compiler that supports modules is used. For most Prolog module systems, the performance of module calls is close or even identical to the performance of plain Prolog calls when using imported predicates and implicit qualification. When using explicit module qualification, performance can be worse.
When static binding is used, messages to objects are, whenever possible, translated to direct predicate calls. Thus, performance should be about the same as in plain Prolog predicate calls. However, due to the overhead of one extra argument per object predicate (used for passing the execution context), the performance of Logtalk optimized calls might be slightly worse than the equivalent plain Prolog predicate calls.
When running the tests, pay special attention to the empty loop times. If the times for the empty loop are not stable across runs, try to use a higher value (than the default one) as argument of the run/1 predicate.
When the number of benchmark test repetitions is too low, you may get a "Number of calls per second: n/a" result.
The test that creates and abolishes empty objects may take a long time to complete when compared with all the other tests.