2
edits
Changes
→[0.1] Optimize and run a benchmark program specifically designed to make use of glibc
== Project Contributor(s) ==
IRC:<br />
<ul><li>'''#ubuntu-arm''' (persia)<br /></li><li>'''#gentoo-embedded''' (solar and steev_)</li></ul> SENECA: '''Tae Hee (Tyler) Lee''' <blockquote>My fellow classmate, a great and helpful guy working on the [http://zenit.senecac.on.ca/wiki/index.php/To_Thumb_or_Not_to_Thumb To Thumb or Not to Thumb Project]. We both have the same project objective; to benchmark a system. (his project deals with 16-bit codes) We share access to the test builder cdot-beagleXM-0-3 and even ran tests both at the same time that could have toasted it! Tyler helped me out with with scripting/'makefiles' and using 'screen'.</blockquote>
== Project Details ==
'''ARMv7''' Technologies:
There are currently 2 '''ARMv7''' (beagleboard & beagleboard XM) builders in the Fedora ARM farm. These builders are running builds on '''ARMv5tel''' . This project will focus in using the beagleboardXM builder <code>cdot-beagleXM-0-3</code> ===Specifications for the cdot-beagleboardXM-0-3 builder===----[[Image:Bb.jpg|thumb|right|400px| cdot-beagleboardXM-0-3 builder]]'''beagleboardXM specific'''<br /><blockquote>[http://beagleboard.org/hardware-xM beagleboardXM hardware page]</blockquote>'''cat /proc/version'''<blockquote>Linux version 2.6.32 (ubuntu@ip-10-204-115-71) (gcc version 4.3.3 (GCC) ) #3 PREEMPT Wed Aug 18 15:53:03 UTC 2010</blockquote>'''cat /proc/cpuinfo'''<blockquote>Processor : ARMv7 Processor rev 2 (v7l)<br />BogoMIPS : 515.72<br />Features : swp half thumb fastmult vfp edsp thumbee neon vfpv3<br />CPU implementer : 0x41<br />CPU architecture which means optimum performance is not achieved: 7<br />CPU variant : 0x3<br />CPU part : 0xc08<br />CPU revision : 2<br />Hardware : OMAP3 Beagle Board<br />Revision : 0020<br />Serial : 0000000000000000<br /></blockquote>'''cat /proc/meminfo'''<br /><blockquote>MemTotal: 498716 kB</blockquote> '''rpm -q glibc'''<br /><blockquote>glibc-2.11-2.fa3.armv5tel (Subjected for upgrade to armv7)</blockquote>
== Project Plan ==
Goals for each release:
Any packages compiled for the beagleboards can be installed without optimization. So far, in the case of <code>cdot-beagleXM-0-3</code> packages are compiled without it. Without optimizations, software installed in a system can only run on sub-optimal performance. In order to make use of ARMv7 architecture features, editing the <code>CFLAGS</code> to use <code>arm</code> optimization options will let the compiler attempt to improve the performance and/or code size of the program; resulting in a more efficient/faster system.
The goal of this release is to run a benchmark software named [http://en.wikipedia.org/wiki/Dhrystone Dhrystone] on <code>cdot-beagleXM-0-3</code> and record the results. Three (3) runs are required: '''No optimization''', '''Optimized for armv5tel''', and '''Optimized for armv7'''. Dhrystone is chosen as the benchmark software mainly, to test the general system performance of <code>cdot-beagleXM-0-3</code> and to test how much performance gain can be expected from optimizing a program. Other reasons include:
<blockquote>
<b><font style="font-size:100%">Reasons for using Dhrystone</font></b>
*ARM® recognizes the program and uses it as a performance attribute of their processors.
*Dhrystone provides a more meaningful MIPS (Million Instructions Per Second) because results are compared to a reference machine.
*Dhrystone numbers reflect the performance of the C compiler and libraries more so than the performance of the processor itself. (considered as a weakness of the program)
*Check if armv7 optimization options and armv5tel optimization options differ significantly in program performance
</blockquote>
[[Image:graph.png|thumb|450px|right| A graph showing the overall system performance of cdot-beagleXM-0-3]]
'''Test Result (in DMIPS):'''<br />
Normal = 758.869322709 DMIPS<br />
Optimized for armv5tel = 1034.82179852 DMIPS<br />
Optimized for armv7 = 1034.82179852 DMIPS<br />
The benchmark graph shows that optimization increased the overall performance of cdot-beagleXM-0-3 by '''36%''' (Normal run vs. Optimized for armv5tel/armv7). The results for both armv5tel and armv7 optimizations are the same. (It can be assumed that the armv5tel glibc impacts the performance of C library dependent programs such as Dhrystone). Another possible reason is that the compiler used is already armv7 optimized (since dhrystone also relies on the compiler efficiency) The data gathered can be used as a reference for conducting 0.2 project release.
<br />
<br />
<br />
<br />
<br />
<br />
<br />
<br />
===[0.2] Install an armv7 glibc and re-run the benchmark using dhrystone===
Since optimizing the benchmark program using both architecture optimizations revealed the same results, optimizing the system binaries and re-running the benchmark should be able to provide a more concrete contrast between the 2 architectures.
The goal of this release is to find out if upgrading the armv5tel glibc would affect the performance of a C library dependent program such as dhrystone. By re-installing glibc using armv7 optimizations and re-running an armv7 optimized dhrystone; a better benchmark result is expected (Higher DMIPS). The result would be beneficial for Fedora-ARM, for it will help the community decide if armv5tel codes should continually be supported.
'''Requires:'''
<blockquote>[http://arm.koji.fedoraproject.org/koji/buildinfo?buildID=8721 glibc binaries]</blockquote>
In order for a successful glibc armv7 build, the file ''/usr/lib/rpm/redhat/rpmrc'' needs edit to use armv7 build options. Using mock; a downloaded glibc source can be rebuilt. Once finished, the binaries can then be installed locally to the system using rpm OR a local repository can be created enabling the use of yum.
'''Test Result:'''
The Dhrystone (DMIPS) results never changed. The benchmark brought the same, exact number of DMIPS. Although Dhrystone does make use of C library functions and is assumed that the glibc would have effects on the program; the results proved that upgrading the glibc did not bring what's expected.
It is proven that armv7 and armv5 arch optimizations provide the same level of performance especially when running C library dependent programs on cdot-beagleXM-0-3 builder. Why is it possible when armv7 architecture is supposed to be better than armv5tel? One big answer is that the system tested was built to use an [[Imagehttp:graph//en.wikipedia.png|widthpx| ]org/wiki/Application_binary_interface ABI]called "softfp". Although beagleboardXM (cortex-a8)supports the "hard floating-point" ABI, The Fedora-ARM currently can't afford to waste time building a system that supports "hardfp" solely on test purposes. To make things a little clearer, cdot-beagleXM-0-3 can't use the technology offered by ARM-cortex-a8 processor because of how the system (down to the lowest level) is built.
== Things to learn ==
-rpmmacros
-Dhrystone 2.1<br />
-gcc ARM optimizations<br />
-Ways of benchmarking ARM processors<br />
-gcc install options<br />
-Compiling kernel and glibc<br />
-Familiarization with ARM hardware<br />
=== HOW TOs ===
[http://zenit.senecac.on.ca/wiki/index.php/Dhrystone_howto A guide for using Dhrystone benchmark]
== Project News ==
''December 16th, 2010'' - [0.2] and [0.3] Release updated and finalized
''December 15th, 2010'' - Added the [http://zenit.senecac.on.ca/wiki/index.php/Dhrystone_howto Dhrystone How to Page]
''December 9th, 2010'' - Project page update (0.1 Release)
''November 22nd, 2010'' - Release 0.1 test results posted
''November 4th, 2010'' - Compiler optimization options ready for testing, project page updated