Você está na página 1de 23

TUU 2014 Performance Testing

Whitepaper
V1.0
Sandro Villinger
1 Introduction
1.1 Performance tests

This document is meant to protocol performance tests for AVG PC TuneUp. It is important to note that
all benchmarks were performed on sample machines and do not necessarily represent all laptops on
the market. However, it was made sure that both low-powered and high-powered devices were used in
order to give an accurate representation of devices currently available.

1.1.1 Preparing the test environment & test guidelines

Note

1 The tested devices resided in an environment compliant with ECMA-383:


a. Temperature: 23 degrees Celsius.
b. Relative humidity: 10 - 80 %.
c. Ambient light: 250 +/- 50 lux.
2 The Power Plan has been set to Balanced
3 All external devices have been unplugged (keyboard, mouse, hard disks).
4 Wireless radios have deliberately been enabled to recreate a typical user scenario
5 All tests were performed according to Microsoft Developer Central Hardware guidelines:
All machines have been used productively for 4 weeks.
Rundll32.exe advapi32.dll,ProcessIdleTasks command was used by internal testers
to make Windows perform file placement optimization tasks.
WindowsUpdate was successfully run on all machines 24 hours prior to testing and
the systems were each restarted five times. Windows Update was then disabled.
The Windows Experience Index was calculated.
On Windows 8, the new Automatic Maintenance feature was first executed and
then disabled.
The SuperFetch feature, which adapts to programs and Windows usage, was
enabled and properly trained.
The Windows Search index was fully build.
Scheduled Tasks have been performed.
6 We ran each benchmark test three times and used the average in our result table.

1 The tested devices resided in an environment compliant with ECMA-383:


d. Temperature: 23 degrees Celsius.
e. Relative humidity: 10 - 80 %.
Ambient light: 250 +/- 50 lux.
1.2 Methods used
To provide accurate test results, testers had to perform several steps and foll ow a specific flow. Both
an automated and a manual approach were used to get precise results:

1.2.1 Test tools in detail


Testers used various methods to determine performance.

Boot Time & Shutdown Tests

1 For boot time tests, we used Microsofts Windows Performance Toolkit, part of the
Windows Assessment Toolkit:

The assessment results are used to diagnose potential problems, so that the hardware and software that
you develop are both responsive and have a minimal impact on battery life, startup performance, and
shutdown time. The same assessments are available for OEM partners, ISV/IHV partners, enthusiasts, and
other members of the community, to establish a common framework to measure, compare, and review
aspects of quality.

This kit performs extensive boot traces and covers the entire boot process: From
UEFI/BIOS initialization to the last service is run. It gives an indication of the impact of 3 rd
party software on the boot process.
On both tablets, testers used the Fast Startup check which measures the
Hybrid Boot feature found on Windows 8.
On the ultrabook, testers were forced to perform a full boot which is similar to
the boot experience found on Windows 7 and earlier versions.

2 For shut down tests, Windows Performance Toolkit was used again to measure the PC
shutdown duration down to the millisecond.
3 For resource usage tests (number of processes and memory usage), Task-Manager
was used. In all scenarios, the PC was booted and the values were taken 15 minutes after
the boot procedure was completed to eliminate most post-boot interferences.
4 For photo editing benchmarks, the Photo Handling assessment from Microsoft
Assessment Toolkit was used. This measures the time it takes to complete user tasks
such as viewing, searching and manipulating a set of photos.
5 For file operation tasks, the File Handling assessment from Microsoft Assessment
Toolkit was used. This measures the duration of common file functions such as copy,
move, delete, and zip.

Page 3 of 23
1.2.2 Descriptions of Test PCs

Benchmarks were performed on recent machines (Q4 2011-Q2 2013) with Windows 8, which is the
most performance-optimized Windows OS for consumers to date.
2011-era Ultrabook: Core i7, 4 GB, 256 GB (Purchased: September 2011)
2013-era Microsoft Surface Pro: Core i5 1.7 GHz, 4 GB, 128 GB SSD (Purchased: June
2013)
2013-era Acer Iconia W3: Low-End Intel Atom Z2760 1.8 GHz, 2 GB, 64 GB SSD
(Purchased: June 2013)

It should be noted that the results are likely to be even greater on less performance-oriented operating
systems, such as Windows XP or Windows Vista.

1.2.3 Preparing the machines: Installing 150+ applications


The more programs users download from the web or install from a DVD, the more additional load is put
on the system. This load consists of:
Services
Background processes
Startup items
Plug-Ins and Explorer.exe hooks
Drivers (Some software products even install new drivers, for example virtual DVD drives.)

Additional background activity puts strain on a users machines, as it needs to assign a certain amount
of resources to the newly installed programs which reduces performance. Two factors are responsible
for this loss in performance:
Windows needs to distribute resources (CPU time, handles) on these background tasks
Less memory is available for active processes

The system becomes slower with each program that gets installed. In this experiment, a large (but not
unusual) amount of applications is being installed to see the effects on performance.

This is a typical scenario to test the effects of TuneUp Utilities 2014 on systems with a high load. The
experiment helps evaluate a) how a high application load has an effect on modern systems and b)
whether AVG PC TuneUp is capable of optimizing performance. The following steps were performed:

Page 4 of 23
Step 1: The three test machines used were deliberately left in an untouched condition to
simulate three typical users scenarios:
o Ultrabook: A ultrabook was clean-installed in January of 2013 and few programs
installed. It is an example of relatively newly installed machines with little load.
o Microsoft Surface Pro: The Surface Pro was a clean install and had no programs on
it. Testers needed to install benchmarking software and the appropriate tools to
determine the effect.
o Acer Iconia W3: This machine was pre-installed by the OEM and came with 15 built-in
applications and trial-versions.
Step 2: Testers prepared the test machines according to Microsoft guidelines (see above).
Step 3: Image of the clean install was taken using Microsoft s built-in recimg. To revert to
this clean state, Windows 8s Refresh feature could be used. All three PCs in their
untouched condition are referred to as Clean.
Step 4: To simulate workload, the machines were equipped with approx.. 150 additional
popular programs. Especially on first time users PCs, who do not know about the effects of
3rd party programs on performance, as well as IT pros, who require a significant amount of
applications, this is not an uncommon scenario. A list of installed programs is available (contact
Sandro Villinger, see information below).
o Identifying programs: Testers selected these 150 based on the most used programs
on Wakoopa (a software usage tracker), the top downloads at Downloads.com, from
various Top 10 Programs you need lists and from personal experience with what users
often install. Also, popular Ninite installer was used: http://ninite.com. This tool installs
50+ popular applications silently (double software was unchecked). It is an even mixture
of very popular and huge software suitessuch Microsoft Office or Nero and useful
freeware applications such as Skype and Opera.
o Installation process: Installers were run using the Standard options, which in many
cases installed additional toolbars or downloads. This causes an additional load on the
machine.

Step 5: After the installation was completed, all programs were started at least once on all
three test devices to determine their functionality or set them up.
Step 6: Each three systems were then rebooted 15 times over the course of 3 consecutive
days. This rules out most of the post-installation background activities applications tend to
perform. The states of these systems at this point is referred to as High Load. Acronis
TrueImage was used to perform complete system images.
Step 7: All benchmarks below were performed according to the Microsoft guidelines.
Step 8: TuneUp Utilities 2014 was installed on these devices and the following performance-
optimizing steps were performed:
o 1-Click-Maintenance: Run with default settings.
o Increase Performance Assistant: Profile set and all recommendations accepted
o Fix Problem Assistant: All problems were fixed.
o Turbo Mode: Enabled.
o Program Deactivator: All Programs were turned off (two reboots were necessary).
o Startup Manager: All remaining startup entries were turned off.
o Live Optimization: Enabled.
o Registry Defrag: Performed.
This optimized state is referred to as Optimized with TuneUp.

Page 5 of 23
1.2.4 Disclaimer
The following test results were performed under a highly controlled environment, in accordance to
industry standards, and with professional measurement software which was previously only used by
Microsoft engineers (WPT was an internal Microsoft tool to measure OS performance). The tests were
done with care and repeated several times. However, the testers cannot guarantee that these
performance tests are absolute accurate and can be reproduced on other machines. While performance
testing on clean PCs is straightforward, the installation of several programs introduces factors that
cannot (!) be controlled these include sudden interferences by update mechanisms or self-
maintenance tasks that the installed programs perform after a certain time or when triggered. These
variances were reduced to a minimum by several reboots and days of uptime however, they cannot
be eliminated. It is possible that the results were impacted due to the high load that was put on the
system. Still, the results represent a strong and accurate representation of systems under
high load and how an optimization product is capable of solving these problems.

Page 6 of 23
2.1 Boot Results
Boot time cannot be measured subjectively (e.g. stop watch technique), as Windows performs a
number of post-boot operations that have a severe impact on performance. Only when these
operations are done, users can fully enjoy the PCs performance.

Windows Performance Toolkit traces the entire boot procedure from UEFI/BIOS boot to the last
service being loaded and gives an accurate representation of boot time. The results:

2011-era Ultrabook: Core i7, 4 GB RAM, 256 GB SSD


Clean High Load Optimized with
TuneUp
Average* 62 seconds 164 seconds 60 seconds
*Windows Performance Toolkit performs 3 test runs in total and averages the results. All test details and test dumps are
available, please contact Sandro Villinger for more information.

Page 7 of 23
63% Boot Up Speed Improvement on
Ultrabook

200

150
Seconds

100 164

50 62 60

0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

The ultrabook saw problems with the Hybrid Boot feature of Windows 8, which is why the Full Boot
test suite was used. This performs a regular boot, which is similar to the procedure found in older
Windows versions.
In our tests, the ultrabooks boot procedure was completed after 62 seconds. However, once the 180
programs were installed, boot time went to 164 seconds After the optimization with TuneUp Utilities
2014, the values didnt just return to normal but were even lower (=better) than on the clean
system. This is a reduction of boot time of 63,41%.

Page 8 of 23
2013-era Microsoft Surface Pro: Core i5, 4 GB RAM, 128 GB SSD
Clean High Load Optimized with
TuneUp
Average* 7,2 seconds 57,8 7,8 seconds
*Windows Performance Toolkit performs 3 test runs in total and averages the results.

86% Boot Up Speed Improvement on


Microsoft Surface Pro

60
50
40
Seconds

57,8
30
20
10 7,2 7,8
0
Clean High Load (150+ Optimized (with TuneUp)
Programs)

Clean High Load (150+ Programs) Optimized (with TuneUp)

Total boot time went from 57,8 seconds (after the programs were installed) down to 7,8 seconds
which is almost as low as when the Surface Pro was tested in its clean condition (7,2 seconds). In
short, turning off all additional load of the system and enabling all performance-related features of
TuneUp Utilities 2014, the Microsoft Surface Pro saw a decrease in boot time of 86,51%.

Page 9 of 23
2.2 Shutdown Results
Restarting and shutting down the system is a valid indicator of how fast the PC is capable of ending
processes and services. Also, if a shutdown takes too long, users tend to get frustrated a shorter
shutdown means a generally better user experience.

2011-era Ultrabook: Core i7, 4 GB, 256 GB SSD


Clean High Load Optimized with
TuneUp
Average* 3,7 seconds 18,9 seconds 1,8 seconds
*Windows Performance Toolkit performs 3 test runs in total and averages the results.

90% Faster Shutdown on Ultrabook

20

15
Seconds

18,9
10

5
3,7 1,8
0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

Time it took to shut down the ultrabook went from 3,7 seconds to 18,9 once all the programs were
installed. TuneUp Utilities 2014 was able to optimize this PC significantly: After the optimization, it only
took 1,8 seconds to shut down the PC. This is a reduction of boot time of 90%!

Page 10 of 23
2013-era Microsoft Surface Pro: Core i5, 4 GB RAM, 128 GB SSD
Clean High Load Optimized with
TuneUp
Average* 4,5 seconds 11,3 seconds 6,0 seconds
*Windows Performance Toolkit performs 3 test runs in total and averages the results.

46.9% Faster Shutdown on Microsoft Surface


Pro

12
10
8
Seconds

11,3
6
4 6
4,5
2
0
Clean High Load (150+ Optimized (with TuneUp)
Programs)

Clean High Load (150+ Programs) Optimized (with TuneUp)

The shutdown procedure took slightly more than 11 seconds after all the programs were installed on
the Surface Pro. However, once TuneUp Utilities 2014 performed all tuning operations, this value
decreased noticeably from 11 seconds to 6. This is an improvement of 46.9%.

Page 11 of 23
2.3 File Copying Results
File I/O can be severely impacted by background processes consuming both hard disk cycles as well as
CPU time. Windows Assessment Toolkit performed large and small file operations repeatedly to
determine the hard disk performance. The results:

2011-era Ultrabook: Core i7, 4 GB, 256 GB SSD


Clean High Load Optimized with
TuneUp
Average* 8.7 seconds 9.5 seconds 8.6 seconds
*Windows Performance Toolkit performs 3 test runs in total and averages the results.

9.5% Faster File Copy Speed on Ultrabook

10
8
Seconds

6 9,5
8,7 8,6
4
2
0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

File copy operations on the ultrabook were only slightly impacted (9.5 versus 8,7 seconds) by the
additional load, which is due to the stronger Core i7 processor and the SSD. The optimization with
TuneUp Utilities returned this value back to normal. This is an improvement of 9.5%.

2013-era Microsoft Surface Pro: Core i5, 4 GB RAM, 128 GB SSD


Clean High Load Optimized with
TuneUp
Average* 5,6 seconds 6,6 seconds 5,8 seconds
*Windows Performance Toolkit performs 3 test runs in total and averages the results.

Page 12 of 23
12% Faster File Copy Speed on Surface Pro

5
Seconds

4 6,6
5,6 5,8
3

0
Clean High Load (150+ Programs) Optimized (with TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

The additional load of 150 programs had a slight impact on copy operations (6.6 versus 5.6 seconds),
which testers noted in several repetitions of the benchmarks. The optimization with TuneUp Utilities
improved the file copy results, but could not revert it back to the original value. Still, it is an
improvement of 12%.

Page 13 of 23
2013-era Acer Iconia W3: Low-End Intel Atom 1.8 GHz, 2 GB RAM, 64 GB SSD
Clean High Load Optimized with
TuneUp
Average* 70 seconds 152 seconds 64 seconds
*Windows Performance Toolkit performs 3 test runs in total and averages the results.

58% Faster File Copy Speed on Acer Iconia W3

160
140
120
100
Seconds

152
80
60
40 70 64

20
0
Clean High Load (150+ Programs) Optimized (with TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

The low-end processor, also commonly found in netbooks, was not designed to handle both intense
background processes and CPU-intense file copy operations. It took the device 152 second to complete
the task. After the optimization, it took the device only 64 seconds to complete the task. This is an
improvement of 58%.

Page 14 of 23
2.3 Photo Editing Results
This automated tasks measures the performance when working with photos (small to large). Tasks that
are measured several times are: Opening a large picture folder, building the thumbnail index, opening
photos, rotating pictures and manipulating pictures. The results are not shown in seconds. Instead,
Microsoft Assessment tools trace delays that are visible to the user. If there is no visible delay in the
operation on screen (e.g. opening a picture), the value of operations without delay is 100%.
However, if there is visible impact, the percentage is lower. For example, if the value is 92%, it means
that in 8% there was a noticeable lag in editing or viewing pictures.

2011-era Ultrabook: Core i7, 4 GB, 256 GB SSD


Clean High Load Optimized with
TuneUp
Responses without delays 96% 92% 96%
*Windows Performance Toolkit performs 3 test runs in total and averages the results.

4% Less Delays in Photo Editing on


Ultrabook

100
80
Percent

60 96 96
92
40
20
0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

On the ultrabook, the assessment tool dedected delays in only about 4% of all photo-related tasks.
After the system was under the burden of 150+ programs, this value increased to 8%. TuneUp
Utilities 2014 was able to restore performance to the original, which means there are 4% less
delays.

Page 15 of 23
2013-era Microsoft Surface Pro: Core i5, 4 GB RAM, 128 GB SSD
Clean High Load Optimized with
TuneUp
Responses without delays 100% 100% 100%
*Windows Performance Toolkit performs 3 test runs in total and averages the results.

0% Improvement in Photo Editing on


Surface Pro
2 0 1 3 - e r a Mi c r o s o f t Su r f a c e P r o : C o r e i 5, 4 G B R A M, 128 G B SSD

100

80
Percent

60 100 100 100


40

20

0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

The Surface Pro was the only device that did not suffer under the load of multiple applications being
run in the background. The photo performance did not change.

2.4 Resource Usage


Day-to-day performance and responsiveness is significantly determined by the amount of active
processes as well as free memory. To this end, we decided to measure the number of processes as well
as available memory before and after optimizing the test devices with TuneUp Utilities.
It should be noted that a high number of processes and memory usage has a noticeable effect on
responsiveness, as DWM (Desktop Window Manager), OS processes and also actively used 3 rd party
processes need to share their resources with other processes. This introduces delays, which in turn
lead to a bad user experience. Also, as the devices tested were under a constant stress from the 150+
programs, built-in components (CPU, RAM, HDD) could not enter low-powered states which also has a
significant impact on battery life.

2011-era Ultrabook: Core i7, 4 GB, 256 GB SSD


# of background processes Clean High Load Optimized with
TuneUp
Run 1 69 processes 146 processes 42 processes
Run 2 69 processes 149 processes 42 processes
Run 3 68 processes 148 processes 42 processes
Average 69 processes 148 processes 42 processes

Page 16 of 23
72% Less Active Background Processes
on Ultrabook

150

100
148

50 69
42

0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

The number of active processes jumped from 69 to 148 and reduced the responsiveness drastically.
The task-manager showed a constant activity of 10-30% (CPU usage). After the optimization was
complete, there were only 42 processes active a reduction of 72%.

Memory Usage (Committed) Clean High Load Optimized with


TuneUp
Run 1 1.1 GB 2.3 GB 0.8 GB
Run 2 1.1 GB 2.3 GB 0.8 GB
Run 3 1.1 GB 2.3 GB 0.8 GB
Average 1.1 GB 2.3 GB 0.8 GB

1.6 GB of Memory Saved on Ultrabook

2,5
2
Gigabytes

1,5
2,3
1
1,1
0,5 0,7

0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

Page 17 of 23
The amount of memory used on this 4 GB system increased from 1,1 Gigabytes to 2,3. This means less
available resources to actively used programs. TuneUp Utilities optimizations saved 1.6 GB of RAM.

Page 18 of 23
2013-era Microsoft Surface Pro: Core i5, 4 GB RAM, 128 GB SSD
# of background processes Clean High Load Optimized with
TuneUp
Run 1 40 processes 133 processes 45 processes
Run 2 40 processes 132 processes 45 processes
Run 3 40 processes 133 processes 45 processes
Average 40 processes 133 processes 45 processes

66% Less Active Background Processes


on Microsoft Surface Pro

150

100
133
50
40 45

0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

On the clean Microsoft Surface Pro, ony 40 processes were active as only Windows 8 and just a
handful of programs were installed. However, after the 150+ products were installed and the system
was rebooted 15 times and used for several days, this number jumped to 133. While TuneUp Utilities
2014 improved the situation (45 processes, a reduction of 66%), it wasnt quite able to get it back
into its original shape.

Memory Usage (Committed) Clean High Load Optimized with


AVG PC TuneUp
Run 1 0.9 GB 2.2 GB 1.1 GB
Run 2 0.9 GB 2.2 GB 1.1 GB
Run 3 0.9 GB 2.2 GB 1.1 GB
Average 0.9 GB 2.2 GB 1.1 GB

Page 19 of 23
1.1 GB Memory Saved on Microsoft Surface

2,5
2
Gigabytes

1,5
2,2
1
0,9 1,1
0,5
0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

The amount of memory consumed by the OS, drivers and 3rd party applications rose from 0,9 to 2.2
GB. TuneUp Utilities and its various optimization tools was able to reduce this number drastically by
saving 1.1 GB.

Page 20 of 23
2013-era Acer Iconia W3: Low-End Intel Atom 1.8 GHz, 2 GB RAM, 64 GB
SSD

# of background processes Clean High Load Optimized with


TuneUp
Run 1 51 processes 107 processes 46 processes
Run 2 52 processes 107 processes 46 processes
Run 3 51 processes 108 processes 46 processes
Average 51 processes 107 processes 46 processes

57% Less Background Processes on Acer


Iconia W3

120
100
80
60 107
40 51 46
20
0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

After the additional programs were installed, the low-powered Dual-Core CPU was under stress by 107
processes. It was no longer possible to perform work, launch apps, or even browse the web without
significant delays. After the optimization, only 46 processes were active and the system responded a
lot faster this is a reduction of processes by 57%.

Memory Usage (Committed) Clean High Load (180 Optimized with


Progams) TuneUp
Run 1 0.7 GB 1.4 GB 0.7 GB
Run 2 0.7 GB 1.4 GB 0.7 GB
Run 3 0.7 GB 1.4 GB 0.7 GB
Average 0.7 GB 1.4 GB 0.7 GB

Page 21 of 23
0.8 GB Memory Saved on Acer Iconia W3

1,5
Gigabytes

1
1,4

0,5 0,7 0,6

0
Clean High Load (150+ Optimized (with
Programs) TuneUp)

Clean High Load (150+ Programs) Optimized (with TuneUp)

As the Aconia W3 tablet had only 2 GB of RAM, having 1.4 GB completely eaten up after the
installation of 150+ applications saw a massive decrease in performance. TuneUp Utilities, however,
saved 800 Megabytes of memory on this tablet.

Page 22 of 23
2.5 Cleaning Tests
The following cleaning tests (using Disk Cleaner) were performed after the 150 programs were
installed. It should be noted that the following values represent snapshots and are changing hourly as
the system is used.

Core i7 Ultrabook Microsoft Surface Pro Acer Iconia W3


3,8 Gigabyte of unnecessary 6,8 Gigabyte of unnecessary 8,5 Gigabyte of unnecessary
files files files
14,782 wasteful files in total 17,613 wasteful files in total 41,418 wasteful files in total

Contact Information
The contact information below is provided in case you are working with particular individuals
or agencies on this review. If you have any questions around TuneUp Utilities, please contact:

Sandro Villinger
s.villinger@tuneup.de

Legal Disclaimer

Wi-Fi is a registered trademark of Wi-Fi Alliance.

Bluetooth is a registered trademark of Blueooth SIG.

Samsung and Samsung Series 7 are registered trademarks of Samsung Electronics Co.

Windows , Windows XP, Windows Vista , Windows 7, Windows 8,

Surface Pro, Skype , Microsoft Office

and Internet Explorer, are registered trademarks of Microsoft Corporation.

PowerMark is a registered trademark of Futuremark Corp.
Acer is a registered trademark of Acer Inc.
Opera is a registered of Opera Software ASA.

Page 23 of 23

Você também pode gostar