Text Editor Performance Tests

Discuss anything related to portable freeware here.
Message
Author
TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#121 Post by TP109 »

Excel 2003.

TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#122 Post by TP109 »

Tested SynWrite v6.20.2195, NFOPad v1.71, and Editbone v10.7.1.

Uploaded to link below:
http://s000.tinyupload.com/index.php?fi ... 7250222968

TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#123 Post by TP109 »

Performed additional testing to obtain more information about text editor start times for conducting a deeper comparative analysis of the launching process for possibly identifying parameters that could be useful for predicting or identifying start lag, lockups, crashing and freezing issues. The measurements made for these tests were avg launch time, launch time deviation, avg initialization time, initialization time deviation, min RAM, peak RAM, and VM. Not all the measurements were consistent since ideally, launch times should be shorter than full initialization times, but that was not the case for a few editors, but that could be caused by a compatibility issue between certain editors and the testing system.

Launch times were recorded when the system reported an editor as launched and visible (in actuality, the editor may or may not be visible or rendered). For initialization time, an editor is visible, fully loaded and ready to accept user input. The previous testing measured usability, or when the UI "appears" to the user as fully launched and ready. In contrast, this testing measured actual initialization times, independent of user perceptions.

Overall, the results are similar, but some editors did change position since the rankings are based on the launch and initialization times as measured by the test setup.

The first workbook sheet contains the raw data, the 2nd sheet is the sorted list by initialization time. The 3rd and 4th sheets are for future use. So far, I've only tested in empty launch time mode, but this provides a good idea of how quick these editors are in relation to one another and confirms the previous test results. The editors were launched a minimum of 12 times each for these tests.

Results uploaded to the link below:
http://s000.tinyupload.com/index.php?fi ... 6884839394
Last edited by TP109 on Fri Mar 25, 2016 9:07 am, edited 4 times in total.

xor
Posts: 106
Joined: Sat Apr 18, 2015 11:02 pm

Re: Text Editor Performance Tests

#124 Post by xor »

plea: Any chance to have the .xls file uploaded to somewhere else?

TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#125 Post by TP109 »

Please give a reason why it should be uploaded elsewhere. If there is a problem, I would like to know about it.

TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#126 Post by TP109 »

Here are the latest stability test results. I ranked them as best, good, and average. That seems like the best way to categorize them without appearing overly biased:

Best - experienced no crashes, freezes etc.
AkelPad 4.97
BVD NotePad 5.2
Editor2 2.9.1.13 Ansi
EditpadLite7 7.5
EverEdit 2.9.0.2074
NFOPad 1.68
NFOPad 1.7
Notepad2 4.2.25
Notepad2BE 4.2.25 R4
PSPad 4.60
SSLPgmEdit 3.0
TedNotePad 6.0.2
TextAdept 8.4
TopGun 2.4
Win32pad 1.5.10.4
X-ConTEXT 0.98.6 rev3

Good
AkelPad 4.9.2
JANE 1.69
NotePad++ 2.7.1
SavageEd2 0.02.00
SynPad 1.2.11.69

Average
Aedit 4.0-SX r3
CudaText 1.2.12.0
Editbone 10.1.1
Editbone 10.2.1
Editbone 10.3.1
Editbone 9.3.4
EditPadLite7 7.3.7
Extreme Editor 7.1.2.5
GigaEdit 1.0.0.0
Kudaz 2.2.1.2
MadEdit 2.9.1
MadEdit-mod 0.3.9
Min Profit (GUI) 5.2.10
Notepad2-mod 4.2.25 r964
Notepad3 bld 289
nPad2 3.1.3.36
PlainEdit 1.7.6.0 (non-NET)
QuickEditor 3.5
SynWrite 6.16.2010
SynWrite 6.19.2150
X-Scite 3.5.1
Geany Portable 1.24

A little more info about the previous launch time analysis test results. Some of the times may appear off from previous results, but that is because the launch time analysis tests used actual data, and that doesn't always correlate with what the user sees. So an editor like PSPad, which appears on the bottom, isn't actually the slowest. It's at the bottom because I couldn't obtain an initialization time. Another editor like Geany Portable, with an initialization time of 11 secs seems excessive, but it's a Linux native and probably isn't fully integrated with Windows, so it's debatable whether that's actually correct. There are other factors involved too that affect the times. In any case, editors that appear in the top rankings for all the test results are more likely to be the quickest and most compatible with Windows (at least on XP). And just because an editor is fast doesn't mean it's stable, so the stability tests need to be taken into consideration.

The variation measurements/calculations probably don't mean much at around 100 or so, and they are more important for initialization time than for launch time, but anything in the multiple hundreds or thousands is likely to be an indicator of instability.
Last edited by TP109 on Wed Nov 02, 2016 3:47 am, edited 2 times in total.

xor
Posts: 106
Joined: Sat Apr 18, 2015 11:02 pm

Re: Text Editor Performance Tests

#127 Post by xor »

TP109 wrote:Please give a reason why it should be uploaded elsewhere. If there is a problem, I would like to know about it.
I keep getting a xls file containing only the source code (html css js) of tinyupload.com's download link page, weird! but atm I can only open the xls file in Android (no Windows) dunno wheter to worry or feel ashamed for being missing something. I dared think it is a tinyupload issue.

TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#128 Post by TP109 »

xor wrote:
TP109 wrote:Please give a reason why it should be uploaded elsewhere. If there is a problem, I would like to know about it.
I keep getting a xls file containing only the source code (html css js) of tinyupload.com's download link page, weird! but atm I can only open the xls file in Android (no Windows) dunno wheter to worry or feel ashamed for being missing something. I dared think it is a tinyupload issue.
Nobody else has complained about this. I use TinyUpload because it doesn't require registration and I can get a download count, although I can't find out who has downloaded the files or when they were downloaded. So it provides limited stats without a lot of hassle. To determine if TinyUpload is the problem, I uploaded the file to the link below:
http://www.fileconvoy.com/dfl.php?id=g4 ... 5ff236b284

Let me know if that works or not.

xor
Posts: 106
Joined: Sat Apr 18, 2015 11:02 pm

Re: Text Editor Performance Tests

#129 Post by xor »

this time (fileconvoy) the contents of the xls file is correct. thanks, and good job.

TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#130 Post by TP109 »

Tested HiEditor v1.0.1.7 and TextEditorPlus v1.1. TextEditorPlus saves its settings to appdata. Didn't test HiEditor for portability, but its ini file is in the application folder.

The following charts are based on the latest text editor versions. These should make it easier to evaluate how they compare to one another and provide some idea of their maximum file size handling capabilities.
te_empty.jpg
te_68mb.jpg
te_250mb.jpg
Last edited by TP109 on Tue Oct 04, 2016 5:10 pm, edited 1 time in total.

TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#131 Post by TP109 »

Posts are limited to 3 attachments, so here are the remainder of the charts for the 500MB and 750MB launch times.

Keep in mind that although an editor may be able to load a large file like a 500MB or 750MB text file, many take an unacceptably long time to do so and they often become unstable too. Those that can load a large file quickly (say under 5 secs) are usually the most stable.
te_500mb.jpg
te_750mb.jpg

User avatar
webfork
Posts: 10818
Joined: Wed Apr 11, 2007 8:06 pm
Location: US, Texas
Contact:

Re: Text Editor Performance Tests

#132 Post by webfork »

Interesting stuff. Some questions here:

* I wonder what Savageed is doing for 2 minutes with a 750 meg file. Maybe the program is indexing the file so it can run searches? Does Clipedit or Editpadlite have any trouble searching the file after the initial (almost instant) file load?

* Did EditPad crash in the 750 meg test?
TP109 wrote:Posts are limited to 3 attachments, so here are the remainder of the charts for the 500MB and 750MB launch times.
Just for reference, Imgur has free hosting and you can list as many as you want. ShareX has a great interface for uploading multiple images.

TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#133 Post by TP109 »

webfork wrote:Interesting stuff. Some questions here:

* I wonder what Savageed is doing for 2 minutes with a 750 meg file. Maybe the program is indexing the file so it can run searches? Does Clipedit or Editpadlite have any trouble searching the file after the initial (almost instant) file load?

* Did EditPad crash in the 750 meg test?
TP109 wrote:Posts are limited to 3 attachments, so here are the remainder of the charts for the 500MB and 750MB launch times.
Just for reference, Imgur has free hosting and you can list as many as you want. ShareX has a great interface for uploading multiple images.
SavageEd2, QuickEdit and HiEditor basically show blank text interfaces while loading large files although the editors themselves are visible. They are nonfunctional during that time and can't be used in any way. The positive part is that they don't usually crash while loading large files.

EditPadLite and GigaEdit incrementally load files and display the loaded portions of the file during the loading process.

Editpadlite can search and find text that has already loaded, but not text which hasn't loaded yet, although a search term can be entered in advance and it will find it once that particular portion of the file has loaded. Editpadlite's UI is also usable during load.

GigaEdit is the fastest loading editor for large files. It behaves similarly to Editpadlite, but searches are not performed in the same way. Searches are very processor intensive and cause GigaEdit to either stop loading or to drastically reduce it's loading speed, so it's best to let it fully load before searching.

Yes, the newer versions of EditPadLite crash with an out of memory error while loading 750MB test files.

The charts are in order of the test file used, starting from no file to 750MB. As the test file size increased, the number of editors that could load those files decreased. At 750MB, only four editors were left.

The results are somewhat deceptive. Just because an editor can load a certain size file doesn't mean it's usable.

Actually, launch times can be further categorized this way:
1. Fastest to fully load and usable during load
2. Fastest to fully load while not being usable during load
3. Fastest to display a usable UI while loading

The charts above don't make a distinction between these three categories, but place all the editors into category 3, which is more of a "user perception" measurement. So GigaEdit, Editpadlite and other so-called fast editors don't actually load the test file in the times shown in these charts, but show a faster time because they have usable interfaces and display partially loaded results while loading. They also inform the system that they are "initialized" even though they are actually still loading the file. Many of the other editors don't display anything at all and are unusable while loading. A user in those cases may become uncertain if the editor will load the file at all or if it will crash. For editors that have "frozen" UIs while loading, the times are correct since they don't display results or have usable UIs until the test file is fully loaded. For editors that incrementally load files and have usable UIs, the times are deceptive and don't really indicate a fully loaded file, they only "appear" to have fully loaded the file. In other words, the charts above better indicate which editors have the "ability" to load these large test files rather than the actual time it takes to load them.

Anyway, fully loaded launch time tests will require separate charts. I will provide those in the near future.

Hope that answers your questions. At least for now.

Yes, I've used and I'm aware of Imgur. In order to more accurately assess member interest, I thought it would be better to post these particular charts directly into this forum. For other topics such as submissions and such, I would use Imgur.

TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#134 Post by TP109 »

Webfork's comments prompted me to take a closer look at the launch/load tests. I discovered that several apparently loaded editors were not actually fully loading the test files. For instance, some editors such as QuickEditor and SavageEd2 only appear to completely load a 750MB test file since the system reports them as initialized, their file contents are visible, they don't crash or freeze, and they provide some type of visual indication reporting the test file as fully loaded. After trying to scroll or search to locate the end of the test files, I found that a few editors such as QE and SE2 just stopped loading at some point. In other words, they never fully load the test file, explaining why some of the tested editors never crashed when loading large files. I knew there were some differences between launch and load times, but wasn't aware of that particular issue or that the gap between launch and load times would be as large as I found out.

I also noticed some unexplained behavior, such as some editors being able to load a 750MB or 500MB file faster than a 250MB file.

After marking the end of each test file for easy identification, each editor was retested for complete file loading and verified by using search/find to match the last line of the test file, and ensuring that no crashing, freezing, errors or any other problems occurred during that process. Editors not able to fully load the test file or displaying any issues after loading a test file were dropped from the charts.

The charts below compare launch time to load time for each editor. The empty load chart includes all tested editors. After that, only those editors verified to fully load test files without problems are included in the 68MB, 250MB, 500MB, and 750MB category charts. Launch times are labeled as "Perceived" and fully loaded times are labeled as "Actual". Many editors, such as GigaEdit or QuickEditor have fast launch/initialization times, but have slower load times - often by several factors when loading large files. Editors with this type of behavior can be easily identified using the charts. All of the following charts are sorted by load time (actual), indicated by the yellow bars. Launch/initialization times (perceived) are in purple.
te_empty.jpg
te_68mb.jpg
te_250mb.jpg
continued in next post ....
Last edited by TP109 on Sun Oct 09, 2016 1:51 pm, edited 10 times in total.

TP109
Posts: 571
Joined: Sat Apr 08, 2006 7:12 pm
Location: Midwestern US

Re: Text Editor Performance Tests

#135 Post by TP109 »

te_500mb.jpg
te_750mb.jpg

Post Reply