i bought 2pcs drives similar to this at $2.50 each (averaged incl shipping)...
https://shopee.com.my/Samsung-Metal-U-Disk-USB-3.0-Flash-Drive-2TB-High-Speed-Reading-Memory-Stick-Pen-i.308717336.9909165828 advertised as 2TB USB3, but unfortunately when they came, they only 117GB capacity (128GB drive?) a bit dissapointed because i cannot make a hack intended for them. nevermind... on to the data verification test, took many hours to fill each drive to 117GB (average 3-4MB/s write). at half way at 64GB, they passed verification test, so i continue until they are full. currently verifying all the data at 15-20MB/s read (after unmount and remount of course! to clear up disk caches). checking randomly with hex editor, all data up to 117GB seems legit, so its also a bit surprising to me that they have real 128GB meat in there. 128GB USB2 drive is still worth the money imho. so far my experience is mixed, on all the dodgy flashes that i purchased, but more toward positive side since they are not totally useless, in fact i can make them all really usefull for my need. the price is about right, as long as you look for the cheapest deal and dont go and hope for the price beyond maybe $10-50 nonsense... but this is of course purely my own experience and opinion, ymmv.
ps: i have to update my software to ver2 last night, to include filename tag inside the file data, to enable me eyeballing in hex editor, and the software automation ensures wrap around data is not happening (just in case). and "seed" blocks is now interchanging between non-random (sequential integers) and random numbers, just because for the purpose mentioned, we cant verify perfectly aligned wrapped around data on totally random numbers data (as used in ver1) either using SW automation or visual/manual check. and simple xor checksum is upgraded to yet another simple "xor and add 1" checksum to add more bulletproofness to detection of corrupted data "vector"
i'm not intending to do running blocks/bytes type of checksum as i think my fixed (cluster aligned) block checksum can do the job without doubt, running block will not be bulletproof from perfectly aligned overwritten (wrapped around) data anyway, so thats why the need for human readable data at the beginning of the first block. the software will make "simulation data" (cluster aligned) until the flash is filled up to the set GB and creates sequentially numbered files (1GB each or until space is zero). following me so far? no?
well, i attached the SW + VB source code, ver1 and ver2 in Original Post... fwiw..
edit: if one wonder why i waste my time to do this, its the same reason why others wasted their time reviewing T&M tools in youtube and here, except this is to test the limit of this dodgy technology (Dave did the same on some dodgy DMMs, tools and tech). and i wasted not much, since i can do other work while the SW is doing its job. and just typing here, no need hours of editing videos. the developed SW can be used in the future for other verification purposes. fwiw.
(ps: the SW is finished verifying, 1st run 117GB passed, i will keep the flash drives offline for few hours or days and make 2nd verification pass later to see if data can survive that long without power. in the mean time i will recompile and upload the SW ver2 (final touch up), retaining ver1 in this thread since i'm not going to keep the compiled exe any longer. here we go!)
edit: i tried Flash Tester tools in the net, they work sort of cluster/sector based verification, test is soooo long and no way to interrupt/verify in the middle of process. Windows Scandisk cant detect bad sectors on my previous 1TB drive, bummer! so i will use my own checksum tool for this
cheers and best regards.