I didn't test with more than 256 MB dictionary because this laptop only has 8 GB of memory (512MB dict + 4 threads requires 12GB of ram to compress) and because usually the effectiveness of a dictionary starts to decrease at around 1/8th of the total file size. It's a good mix of files to test a compressor, it has source code, binary files (DLLs), images, archives (jar files) etc etc. Select download as zip, you'll get a 1.2 GB zip which unpacks in 2666 MB. * WinRAR has the single algorithm for version 5, while 7zip allows you to use algorithms like BZip2 or PPMd if you plan to archive mostly TEXT documents (these algorithms work best with text, LZMA and LZMA2 work best with everything)Īs a proof of what i mean, I downloaded Google Chrome source code from here: * WinRAR 5 is basically some variant of LZMA or LZMA2 algorithm if I remember correctly, which is the default of 7zip since it was launched I think you can select thread count from command line but who does that

* WinRAR will auto pick the number of threads to compress, while 7zip allows you to use only 1 thread if you want absolutely bestest compression possible (more threads results in slightly lower compression ratio). If you get a 32+ core processors like the current Threadripper/Epyc, you won't use them to the max with WinRAR, unless you run multiple instances at same time. * WinRAR is limited to maximum 32 threads, not sure about 7zip but I think it can do more. In contrast, 7zip will let you use even 1.5 GB dictionary size and its supported regardless of version you have. IF you go with RAR5, the dictionary size can increase to 1 GB. * IF you want backwards compatibility with people that use older WinRAR on unRAR programs, RAR 4 format is limited to a 4 MB dictionary. If you're using the defaults (ex fast, best compression), then WinRAR may get you better compression out of the box, but 7zip gives you a lot of flexibility and you're not as restricted. 7zip outperforms WinRAR at everything, if you know what you're doing.

Rar file extractor free