ZIP vs. 7z: Using Optimal 7-ZIP Compression Presets

Photo by JJ Ying on Unsplash
Photo by JJ Ying on Unsplash

Recently I have cleaned up my files and folders which had accrued over more than 20 years. (Btw tools like AllDup and FileJuggler have been tremendously helpful). In the process I also compressed some larger folders using the open-source tool 7-ZIP.

That was the point where I asked myself which would be the best 7-ZIP compression settings to compress a typical “work” folder with some data, PDFs, Microsoft Office files, some images. Not that it would be necessary given how affordable storage has become, but nevertheless I wanted to know for lil’ Geek’s peace of mind.So what I did is this: I created a typical “office and work” folder which looks like this:

7-ZIP Benchmark Folder

…and ran it through all 7z and ZIP compression profiles which are available in 7-ZIP.  The result:

7-ZIP Benchmark Results

There are basically 3 tiers of speed vs. compression.

  1. Around 55% compression, there is no option but to use 7z, because ZIP just can’t compress the files that good, not anywhere near. The best preset here is “7z Normal”. You can save a little more space with “7z Maximum” or even “Ultra”, yet that comes with a hefty hit in terms of speed.
  2. Around 30% compression, there is not better preset than “7z Fastest”. Especially if compared to ZIP, you will get ~5 times the speed for exactly the same compression ratio. If you’re after the best compression in this tier, go for “7z Fast”. 
  3. Finally there is “7z Save” and “ZIP Save”, which does not compress anything at all, but wraps files up into a single ZIP or 7z archive. Both of these options happen at full drive speed (here: ~2.5 GB/s), where 7z has a minimal advantage because the overhead data added to the archive file is marginally less compared to ZIP.


ZIP is the most widespread format out there, but compared to 7z (LZMA2) it is just inefficient. At only a fraction of the speed of 7z, you will get max. ~30% compression, while there is no way to get up to ~55% compression with ZIP at all. And even at 55% compression with “7z Normal” preset, 7z will be 50% faster than ZIP for ~30% compression performance. Yes, it is that clear.

Sure, there are other considerations to be taken into account, like compatibility (ZIP is surely most widespread, while you will need 7-ZIP to decompress 7z files), storage of error correction data (which 7z is not capable of, compared to i.e. WinRAR). But beyond those (obviously relevant) aspects, there is just no way around 7z today.


Completely uninstall printer driver or printer port from Windows 10 [solved]

Canon Printer
Photo by Joshua Fuller on Unsplash

In the digital age, with switching locations and offices frequently, it may be necessary to uninstall a specific printer driver or printer port under Windows 10. However, it wouldn’t be Windows if that would work flawlessly under all circumstances.

In my case, Windows 10 threw up an error message which read Driver cannot be uninstalled. The printing queue is not empty. or The driver is currently being used. (I translated this from an actual German error message, so it could read different in English versions.)

Here is what to do about that.



VeraCrypt boot problems after Windows 10 update [solved]

Windows 10 Bluescreen

VeraCrypt 1.2x, Windows 10 x64 v1809

If you want to work with Full Disk Encryption (FDE), plus pre-boot authentication and you don’t want to use Windows BitLocker for some reason, it is likely you will end up with VeraCrypt. Being an OpenSource tool based on the original TrueCrypt, it has been significantly improved over its predesessor and is widespread in the community.

However, there are some problems connected to VeraCrypt and UEFI / Windows 10’s update mechanism / bootloader handling.

Problems with VeraCrypt and Windows 10 bootloader

What you experience is, that the UEFI boot entry for VeraCrypt is gone. The following applies if you encrypted your whole system drive / partition using VeraCrypt. After an update of your mainboard BIOS, or after major Windows 10 updates, you may experience a boot-loop or bluescreen, where Windows 10 will try to “repair” your Windows 10 installation over and over, without succeeding. The error message is:

Automatic Repair couldn’t repair your PC
Press “Advanced options” to try other options to repair your PC or “Shut down” to turn off your PC


Are deep neural networks dramatically overfitted?

Geek alert (headline = link btw). That is an excellent piece on overfitting effects in what many people refer to as ‘AI’ (Artificial Intelligence) or ‘ML’ (Machine Learning).

Overfitting: the nature of the models widely used, in detail some methods like backpropagation, amplifies multicollinearity within these models, which basically means to artificially create stronger correlations between independent variables and the dependent variable in a development sample. That can lead to extraordinary predictive accuracy of the model – which is limited to the test sample, while it performs much worse using e.g. real-life data. The model then has been ‘over-fitted’ to that particular data sample.

Despite (European) startups claiming to use ‘AI’ in many cases but just don’t do that, always remember as a rule of thumb if someone talks to you about ‘AI’:

In your pitch, it’s AI.
In your specs, it’s ML.
In your pilot, it’s linear regression.
In your product, it’s printf();

Don’t know the source (thanks for any hint!), I recently read it on LinkedIn and had a good laugh 🙂


FutureID3: Biometrics & automated decision making

Picture courtesy of Mia Harbitz

It has been a pleasure to be part of FutureID3 at Jesus & St. John’s Colleges at Cambridge as a speaker, thanks for having me.

Part of the journey were many interesting and meaningful insights into biometrics in ID registration, automated decision making, its applications in the field, policy making, but also its societal effects, and challenges. Highly recommended to anyone who engages in that area.

My part was about the relevance of social acceptance of variables in credit scoring, and their actual predictive power when calculating a credit score. An appendix to that was about how we apply some of the resulting principles at awamo in East Africa.

Overall a very interesting event, with lots of insights from recent research, and the actual field applications, with lots of inspiring participants from all over the world. Special thanks go Keith Breckenridge from University of Witwatersrand (South Africa) for organizing the event, and to Simon Szreter for giving us the grand tour on Jesus & St. John’s Colleges.