The test program writes a 64kB block of generated random data 10.000 times onto one single position on the stick and checks after each write if the data is stored correctly. When a 10.000 write cycle has passed all other cells on the stick are also checked, if they did loose their information which can happen as a side effect. This goes on until the stick throws errors. Yes, this destroyes the stick, but we gain some better understanding what sticks handle wear leveling well and which do not

Endurance will be measured by "rounds". A "round" is defined as how often the whole stick could have been completely filled by the written data volume (until the error occured). This takes a stick's size (which helps in doing wear leveling) into account and can give a size independent measure for endurance.
A heavy working Meteohub system uploading lots of graphs, feeding many weather networks is expected to produce a load of about 2 "rounds" per day. Newest Meteohub versions do display a accumulated round value since last reboot on "System Info" page. When you devide that value by the number of days the system is online you get the average "rounds" per day. In the attached example this is about 2 "rounds" per day. When you have a good storage like JetFlash130 in use, which can stand over 2000 rounds you can expect at least 3 years of untroubled operation. In reality I would expect more than the double of this, because the stress test by writing over and over data to one and the same blocks for a hundered of million times is much more demanding than what happens in real life.
I will post findings about no name and major brand sticks in this thread...