JamesKehr Though I think it is good that some of Microsofts source code is released as FOSS, I am not going to review the code for 2 (or 3) different applications just to find out that they are not measuring the actual performance of the system. I would not go through the design specs., blueprints and QA-documents for the speedometer of every car brand, even if it was available. I would just use a third party tool, that I can use with every car.
If all of the 3 MS test applications were only one single set of source code that was portable. I would perhaps be more inclined to review and use it. If it had the option to test all the network libraries that are possible to use in Windows, that would perhaps sell it even better. A tool like that could even help with black box network analysis of closed source applications.
Until then I will continue recommending iPerf3 or even a simple "time scp..." to do network performance testing. At least I am (hopefully) getting the OS worst case. Not a misleading high result.