Object code optimizer: Difference between revisions

Revision as of 07:23, 21 August 2022

Latest revision as of 22:35, 29 July 2025

(12 intermediate revisions by 9 users not shown)

Line 1:

{{Short description|Aspect of software compilation}}

{{Multiple issues|

{{original research|date=November 2016}}

Line 6 ⟶ 7:

{{Use list-defined references|date=August 2022}}

[[Image:Binary optimizer.png|thumb|180px|A binary optimizer takes the existing output from a compiler and produces a better execution file with the same functionality.]]

An '''object code optimizer''', sometimes also known as a '''post pass optimizer''' or, for small sections of code, [[peephole optimizer]], forms part of a software [[compiler]]. It takes the output from athe source language compile step - the object code or [[binary file]] - and tries to replace identifiable sections of the code with replacement code that is more [[algorithmic efficiency|algorithmically efficient]] (usually improved speed).

==Examples==

* The earliest "COBOL Optimizer" was developed by [[Capex Corporation]] in the mid 1970s for [[COBOL]]. This type of optimizer depended, in this case, upon knowledge of '"weaknesses'" in the standard IBM COBOL compiler, and actually replaced (or [[patch (computing)|patch]]ed) sections of the object code with more efficient code. The replacement code might replace a linear [[lookup table|table lookup]] with a [[binary search algorithm|binary search]] for example or sometimes simply replace a relatively slow instruction with a known faster one that was otherwise functionally equivalent within its context. This technique is now known as [[strength reduction]]. For example, on the [[IBM/360]] hardware the <code>CLI</code> instruction was, depending on the particular model, between twice and 5 times as fast as a <code>CLC</code> instruction for single byte comparisons.<ref name="IBM"/><ref name="Evans_1982"/>

* "[[IBM]] Automatic Binary Optimizer for z/OS<ref name="IBM_ABO_2015"/>" (ABO) was introduced in 2015 as a cutting-edge technology designed to optimize the performance of [[COBOL]] applications on [[IBM Z]]<ref name="IBM_Z"/> mainframes without the need for recompiling source. It uses advanced optimization technology shipped in the latest Enterprise COBOL.<ref name="COBOL-zOS"/> ABO optimizes compiled [[binaries]] without affecting program logic. As a result, the application runs faster but behavior remains unchanged so testing effort could be reduced. Clients normally don't [[recompile]] 100 percent of their code when they upgrade to new compiler or IBM Z hardware levels, so code that's not recompiled wouldn't be able to take advantage of features in new IBM Z hardware. Now with ABO, clients have one more option to reduce [[CPU]] utilization and operating costs of their business-critical COBOL applications. You can try ABO out with an improved, easy-to-use ABO Trial Cloud Service <ref name="IBM_ABO_2020"/> without installing ABO on your system.

* The earliest "COBOL Optimizer" was developed by [[Capex Corporation]] in the mid 1970s for [[COBOL]]. This type of optimizer depended, in this case, upon knowledge of 'weaknesses' in the standard IBM COBOL compiler, and actually replaced (or [[patch (computing)|patch]]ed) sections of the object code with more efficient code. The replacement code might replace a linear [[lookup table|table lookup]] with a [[binary search algorithm|binary search]] for example or sometimes simply replace a relatively slow instruction with a known faster one that was otherwise functionally equivalent within its context. This technique is now known as [[strength reduction]]. For example, on the [[IBM/360]] hardware the <code>CLI</code> instruction was, depending on the particular model, between twice and 5 times as fast as a <code>CLC</code> instruction for single byte comparisons.<ref name="IBM"/><ref name="Evans_1982"/>

===Advantages===

Line 25:

===Recent developments===

More recently developed '"binary optimizers'" for various platforms, some claiming [[novelty (patent)|novelty]] but, nevertheless, essentially using the same (or similar) techniques described above, include:

* [[IBM]] Automaticautomatic Binarybinary Optimizeroptimizer for z/OS (ABO) (2015)<ref name="IBM_ABO_2015"/>

* [[IBM]] Automaticautomatic Binarybinary Optimizeroptimizer for z/OS (ABO) Trialtrial Cloudcloud Serviceservice (2020)<ref name="IBM_ABO_2020"/>

* [[The Sun Studio Binarybinary Codecode Optimizer]]optimizer<ref name="Binopt"/> — which requires a [[software performance analysis|profile]] phase beforehand

* Design and Engineering of a Dynamic Binary Optimizer — from [[IBM]] T. J. Watson Res. Center (February 2005)<ref name="Duesterwald_2005"/><ref name="Xu-Li-Bao-Wang-Huang_2007"/>

* QuaC: Binarybinary Optimizationoptimization for Fastfast [[self-modifying code|Runtimeruntime Codecode Generationgeneration]] in [[C (programming language)|C]]<ref name="CSD_1994"/> — (which appears to include some elements of JIT)

* [[DynamoRIO]]

* COBRA: Anan Adaptiveadaptive Runtimeruntime Binarybinary Optimizationoptimization Frameworkframework for Multithreadedmultithreaded Applicationsapplications<ref name="Kim-Hse-Yew_2007"/>

* Spike Executableexecutable Optimizeroptimizer (Unix kernel)<ref name="Spike_2010"/>

* "SOLAR" Softwaresoftware Optimizationoptimization at Linklink-time Andand Runrun-time<ref name="SOLAR"/>

* Dynimize: CPU Performanceperformance Virtualizationvirtualization<ref name="Dynimize"/>

* BOLT: post-link optimizer built on top of the [[LLVM]] framework. Utilizing sample-based profiling, BOLT improves the performance of real-world applications even for highly optimized binaries built with both [[Feedback Directeddirected Optimizationoptimization|feedback directed optimization]] and [[Linklink-time optimization]]. For [[GNU Compiler Collection|GCC]] and [[Clang]] compilers, BOLT speeds up their binaries by up to 20.4% on top of FDO and LTO, and up to 52.1% if the binaries are built without FDO and LTO.<ref name="Panchenko-Auler_2019"/>

==See also==

Line 47:

<ref name="IBM_ABO_2015">{{cite web |date=2015 |url=https://www.ibm.com/products/automatic-binary-optimizer-zos |title=IBM Automatic Binary Optimizer for z/OS - Overview |website=www.ibm.com |access-date=2020-05-15 |archive-date=2020-10-18 |archive-url=https://web.archive.org/web/20201018172402/https://www.ibm.com/products/automatic-binary-optimizer-zos |url-status=live}}</ref>

<ref name="IBM_ABO_2020">{{cite web |date=2020 |url=https://optimizer.ibm.com/ |title=IBM Automatic Binary Optimizer for z/OS Trial Cloud Service |website=optimizer.ibm.com |access-date=2021-10-27 |archive-date=2021-01-19 |archive-url=https://web.archive.org/web/20210119201500/https://optimizer.ibm.com/ |url-status=live}}</ref>

<ref name="Panchenko-Auler_2019">{{cite journalbook |author-last1=Panchenko |author-first1=Maksim |author-last2=Auler |author-first2=Rafael |author-last3=Nell |author-first3=Bill |author-last4=Ottoni |author-first4=Guilherme |title=BOLT:2019 aIEEE/ACM practicalInternational binarySymposium optimizeron forCode data centersGeneration and beyondOptimization (CGO) |journalchapter=ProceedingsBOLT: ofA thePractical 2019Binary IEEE/ACMOptimizer Internationalfor SymposiumData on Code GenerationCenters and OptimizationBeyond |date=2019-02-16 |pages=2–14 |doi=10.1109/CGO.2019.8661201 |arxiv=1807.06735 |isbn=978-1-7281-1436-1 |s2cid=49869552}}</ref>

<ref name="IBM_Z">{{cite web |url=https://www.ibm.com/it-infrastructure/z |title=IBM Z Mainframe Servers and Software |website=www.ibm.com |access-date=2018-08-23 |archive-date=2017-12-22 |archive-url=https://web.archive.org/web/20171222051401/https://www.ibm.com/it-infrastructure/z |url-status=live}}</ref>

<ref name="COBOL-zOS">{{cite web |url=https://www.ibm.com/products/cobol-compiler-zos |title=IBM Enterprise COBOL for z/OS - Overview |website=www.ibm.com |access-date=2021-10-27 |archive-date=2021-09-23 |archive-url=https://web.archive.org/web/20210923110834/https://www.ibm.com/products/cobol-compiler-zos |url-status=live}}</ref>

<ref name="IBM">{{cite web |url=http://www.bitsavers.org/pdf/ibm/360/A22_6825-1_360instrTiming.pdf |title=Archived copy |access-date=2010-01-07 |archive-date=2010-07-11 |archive-url=https://web.archive.org/web/20100711101128/http://bitsavers.org/pdf/ibm/360/A22_6825-1_360instrTiming.pdf |url-status=dead}}</ref>

<ref name="Evans_1982">{{cite journal |url=https://doi.org/10.1145/358728.358732|title=Software engineering for the Cobol environment |author-first=Michael |author-last=Evans |date=1982-12-01 |journal=Communications of the ACM |volume=25 |issue=12 |pages=874–882 |doi=10.1145/358728.358732 |s2cid=17268690 |doi-access-date=2021-10-27free |archive-date=2021-10-27 |archive-url=https://web.archive.org/web/20211027171912/https://dl.acm.org/doi/10.1145/358728.358732 |url-status=live}}</ref>

<ref name="Binopt">{{cite web |url=http://developers.sun.com/solaris/articles/binopt.html |title=The Binary Code Optimizer |access-date=2010-01-07 |archive-date=2010-07-22 |archive-url=https://web.archive.org/web/20100722060415/http://developers.sun.com/solaris/articles/binopt.html |url-status=live}}</ref>

<ref name="Duesterwald_2005">{{cite journal |doi=10.1109/JPROC.2004.840302 |title=Design and Engineering of a Dynamic Binary Optimizer |date=2005 |author-last=Duesterwald |author-first=E. |journal=Proceedings of the IEEE |volume=93 |issue=2 |pages=436–448 |s2cid=2217101}}</ref>