Time and Throughput

For a disk with 128GB capacity and an average throughput of 85MB/s, this test takes around 50 minutes. The formula for computing the test duration is t=2*V/B (i.e., the time in seconds multiplied by the volume in megabytes divided by the average bandwidth in megabytes per second). The factor of 2 is attributable to transferring twice for writing and reading.

This results in a mean bandwidth of B=2*V/t. The exact size of the data medium can be discovered by lsblk -b /dev/$device. If you substitute the value into the formula, you can compute the bandwidth in bytes per second.

A buyer can assume a defect or a fake if measuring the data medium reveals a figure of less than a quarter of the "up to" figure stated by the manufacturer, or less than half of the nominal value. However, you do need to use a connection with the maximum speed for the test, for example, a UHS-II reader for a UHS-II memory card.

Happy Ending

Returning to my two 128GB test cards, bad block reported 122,973,232 defective blocks from a total 131,071,999 blocks (of 1024 bytes each) for the first card – in other words, 94 percent of the blocks are broken. The second card, on the other hand, showed no errors. According to the formula t=2*V/B, the test duration should be about one hour, which was true of the second card, which passed the line after about 70 minutes. The test of the first card took too long – over five hours.

The test with badblocks thus exposed the following about the first card: it was a fake with many faults and was incredibly slow. When I confronted the vendor with these hard facts, I convinced him to take back the card and refund the amount paid. The second and intact card, however, works perfectly in combination with a short SD card adapter with a pull tab as the second SSD in my laptop.