Text Editor Performance Tests

[TP109]

Tested three versions of MadEdit: wxMadEdit, MadEdit, and MadEdit-mod. Although some versions were faster than the others, all were fairly slow to load the 68MB file. These editors load text files according to the max text file setting in Tools->Options->General for MadEdit and MadEdit-mod, and Tools->Options->misc for wxMadEdit. Text files over this setting will be loaded as hex files and these load very quickly (almost instanenouly to the eye). The upper limit for text files appears to be 1,000,000,000 bytes, which is about 953MB, although it's not stated in the Tools->Options settings or anywhere else. It was necessary to change the default settings in order to test with the 68MB test file, and these were the only editors so far where this was required.

The next test file size is a 250MB text file, and MadEdit often crashed (MadEdit-mod and wxMadEdit not tested), but that depended on what other programs were running at the time. When it did load the file, it typically took about 3 mins or 180 secs. VM usage was up to 2GB and RAM usage over 1.5GB. CPU was at 95% at initial load and steadily reduced to around 15-30% halfway into the loading process and at 5-10% towards the end. The UI was unusable during the file load, which is typical.

Also, some clarification on the minimum RAM measurements. Minimum RAM is recorded after the UI is momentarily minimized to check that RAM usage decreases and remains at the lower setting, and for most editors, that is the case. A few however, tend to creep up over time to nearly 100% of the max RAM usage as before the UI was minimized. Although it's not a perfect test, it does identify those editors that have high RAM usage, probably due to memory leakage. Another test that could be performed is to determine if RAM usage decreases without minimizing the UI. However, that takes a lot of time, so it would be more efficient to test for that behavior when there are fewer editors to test.

The link for the test results is below and good for 28 days:
http://www.fileconvoy.com/dfl.php?id=g7 ... 36defe5966

[TP109]

Correction to previous post for MadEdit. Did some more investigation and it looks like the max text file option is in bytes, not bits. So the max text file size is about 953MB when set to 1,000,000,000. This editor doesn't provide a chm-type help file or other documentation. Had to experiment with different size files and the settings to determine that. Previous post corrected. This doesn't change the test results. Its loading speed for 68MB is too slow to test beyond that. I did do some "unofficial" testing at 250MB and those results are in the previous post.

[Minggang]

Good to know the results. I did some optimization on MadEdit-Mod and I think it should be 20~30% or even more better than 0.3.9/10 now.

Would you please share your text file to do the test and the steps so that I can verify it? I did not have a WinXP. However, the performance difference between MadEdit-Mod and NP++ is not that much on Win7 64 bits. I did a simple test with a 67MB text file.

If you have any question about MadEdit/MadEdit-Mod. Feel free to let me know.

Correction to previous post for MadEdit. Did some more investigation and it looks like the max text file option is in bytes, not bits. So the max text file size is about 953MB when set to 1,000,000,000. This editor doesn't provide a chm-type help file or other documentation. Had to experiment with different size files and the settings to determine that. Previous post corrected. This doesn't change the test results. It's loading speed for 68MB is too slow to test beyond that. I did do some "unofficial" testing at 250MB and those results are in the previous post.

[TP109]

Hi Minggang,

I assume you mean the steps for the 68MB file test. The file is actually 68,000kB/66.5MB, so 68MB is slightly off, but the same size was used throughout all testing. Here is the link for the 68MB test file:
http://www.fileconvoy.com/dfl.php?id=g0 ... 74a6617db3
expires in 28 days

As for the steps, SnapTimer or OnlyStopWatch_Unicode are used to time loading. VM, RAM, and CPU usage are measured with Windows task manager and System Explorer. All of these apps are available here at http://www.portablefreeware.com.

In case you are unaware of the method used, launch times with a test file are measured manually if they are over 3-4 secs. For empty launches and anything under 3-4 secs, then the AppTimer setup is used. A minimum of 6 launches are used to get an average launch time. The original setup documentation link is below:
http://www.fileconvoy.com/dfl.php?id=gd ... 1b900a8c74
expires in 28 days

Also, the results for 250MB load were for MadEdit only. I didn't specify that but should have (I already corrected that in my post). I didn't check it for MadEdit-mod or wxMadEdit, because generally I don't test any editor at 250MB or above when launch times are in the 15 sec range for 68MB.

Thought I'd add this bit of info about testing manually when editors take over 3-4 secs to load the test file. The reason for this is because editors that take longer than 3-4 secs to load a file oten use A LOT of CPU and RAM and when that happens, it crashes everything, including the AppTimer setup. And the larger the test file, the more resources that are used. So I do a trial run for each file size by dragging and dropping it into the editor to see how it performs. If it loads smoothly and quickly, then I will use the automated setup; otherwise, it's tested manually. Only a few editors are capable of being tested with the automated setup and the larger sized test files.

[TP109]

Tested MadEdit-mod v3.10. Results are nearly identical to v3.09, so no significant change observed as far as these tests go. Again, didn't test beyond 66.5MB because of the load time for the first test file. Corrected the spreadsheet to indicate a test file size of 66.5MB (68,000kB, a difference of about 2%). Checked the other test file sizes and they are correct. Link for results below. Good for 28 days:
http://www.fileconvoy.com/dfl.php?id=g9 ... b4263b1dee

[Minggang]

Hi, TP109, thanks for the information. It's OK.

Anyway, MadEdit is not famous for the performance but the extraordinary encoding support, cross-platform and Hex edit. I just want to make it better.

Thanks for your information that can help me to do performance tune.

[I am Baas]

I just want to make it better.

There is one annoyance that you should fix... I tested MadEdit-Mod 0.3.10 x64 and discovered that upon closing the "About MadEdit-Mod" window, it will attempt lo launch a Webpage.

[Minggang]

I just want to make it better.

There is one annoyance that you should fix... I tested MadEdit-Mod 0.3.10 x64 and discovered that upon closing the "About MadEdit-Mod" window, it will attempt lo launch a Webpage.

Hi, Baas, it said "Press OK to visit our HomePage.". You may click Cancel if you don't want to.

[Minggang]

Tested MadEdit-mod v3.10. Results are nearly identical to v3.09, so no significant change observed as far as these tests go. Again, didn't test beyond 66.5MB because of the load time for the first test file. Corrected the spreadsheet to indicate a test file size of 66.5MB (68,000kB, a difference of about 2%). Checked the other test file sizes and they are correct. Link for results below. Good for 28 days:
http://www.fileconvoy.com/dfl.php?id=g9 ... b4263b1dee

Hi, TP109, thanks for your work. Just as you had seen, performance of 0.3.10 should be the same as 0.3.9, which is a maintenance release for 0.3.x.
I'm working on latest code and did not release it yet since I have no idea for the major features that should be included as a new release besides the performance improvement.

Base on my estimation(I did not have a setup as what you did), the optimization might has great improvement without cheating---I mean cutting down any functions---only a pure performance improvement on low performance code.

I'm testing the alpha build to make sure there was no bug introduced by the code changes. Will let you know and ask you for helping me to do the benchmark testing against MadEdit-Mod 0.3.10

[TP109]

minggang,

The purpose of the test results posted to this forum is to provide TPFC members and visitors with independent information on the performance capabilities and limitations of portable and freeware text editors to help users make informed decisions for selecting suitable text editors according to their needs. This thread focuses on Freeware/OSS and portable programming and notepad class editors and their performance characteristics that would be important to most users such as speed/responsiveness, resource usage, file size capabilities, and reliability/stability. Because there are so many text editors to choose from, and because the information presented here is to be user focused, it's necessary to place limitations on what's included in order for it to be useful to as many users as possible.

Test data about Alpha and Beta versions are irrelevant to most users seeking information that can be immediately useful to them such as comparing an editor's performance data to other editors, and so that means that the information presented here should be limited to released versions only.

Any discussion regarding performance testing for Alpha and Beta versions should be conducted through other means, either through the PM system and/or private correspondence. I already sent you a PM.

[I am Baas]

I just want to make it better.

There is one annoyance that you should fix... I tested MadEdit-Mod 0.3.10 x64 and discovered that upon closing the "About MadEdit-Mod" window, it will attempt lo launch a Webpage.

Hi, Baas, it said "Press OK to visit our HomePage.". You may click Cancel if you don't want to.

Yes, it does. I just did not notice it the first time (and did not expect it to be there).

[Minggang]

Yes, it does. I just did not notice it the first time (and did not expect it to be there).

It's not a big duel to choose OK or Cancel or add a new button for this. It was designed by the founder. So, I did not change it since then.

[Minggang]

minggang,

The purpose of the test results posted to this forum is to provide TPFC members and visitors with independent information on the performance capabilities and limitations of portable and freeware text editors to help users make informed decisions for selecting suitable text editors according to their needs. This thread focuses on Freeware/OSS and portable programming and notepad class editors and their performance characteristics that would be important to most users such as speed/responsiveness, resource usage, file size capabilities, and reliability/stability. Because there are so many text editors to choose from, and because the information presented here is to be user focused, it's necessary to place limitations on what's included in order for it to be useful to as many users as possible.

Test data about Alpha and Beta versions are irrelevant to most users seeking information that can be immediately useful to them such as comparing an editor's performance data to other editors, and so that means that the information presented here should be limited to released versions only.

Any discussion regarding performance testing for Alpha and Beta versions should be conducted through other means, either through the PM system and/or private correspondence. I already sent you a PM.

Not got the PM yet. Do you mean email?
Don't bother. I did a rough test and the performance almost doubled. The rough estimation is OK for me.

[billon]

TP109, how about CudaText (from the author of SynWrite)?

[TP109]

TP109, how about CudaText (from the author of SynWrite)?

Thanks, I will look at it.