I bought a "PNY 16GB Attaché 3 USB 2.0 Flash Drive 10-Pack" from Amazon for $27.98, to transfer files from some ancient CentOS machines to replacement Debian machines.
The four I've tested so far format to more than 30GB (!!!) of ext3 file systems. Hard to believe, so I read/write- tested two of them with random patterns for a few days. No bits lost. I haven't "debugged" this strange miracle deeply; I don't know whether they actually store 32*8*(2^30) bits, or a slightly smaller fraction of that. My guess is that the chip inside is for a 32GB product, with some spare memory pages beyond that, but a few too many pages were defective to sell as a 32 GB product. So PNY formatted the "best" pages to look like a 16GB Windows flash drive and file system, and sold the drive as that. Formatting that I ignored and overwrote; I don't have any Windows machines around. If any of you wants to deeply examine one of these drives, I would be glad to give one to you, in return for data and opinions about that data. To do the job right, you might need a variable 5-volt"-ish" power supply and a temperature oven; some pages may fail near spec limits. I might loan a power supply, but my lab oven is too big to move. I can't guarantee that other PNY "10 packs" will behave the same. But as a retired chip designer, schooled in the idea that "a defect is a treasure" (hoard and test and understand them, never sell them to lesser beings), it is marvelous to encounter an "anti-defect". Keith L. -- Keith Lofstrom kei...@keithl.com
