Dne Čt 6. září 2012 13:50:19 jste napsal(a):
Yes, this is a clear indication that the device has at most 8 MB erase blocks, not 16 MB. And that's good!
It's also clear that it can do random access inside of each erase block.
To explain what happens here, doing random access over three 16 MB sections on a device that has 8 MB erase blocks in reality means that the device actually uses six 8 MB erase blocks, which it cannot handle.
It is still possible that the actual erase block size is smaller than 8 MB (in which case you would find that it can handle more than 3 blocks), or that the erase blocks are 12 MB.
If you like to find out for sure, you can run
# this is slow for 8 MB erase blocks but fast for 12 MB erase blocks sudo flashbench /dev/sdb --open-au --erasesize=$[8*1024*1024] --blocksize=$[8*1024] --open-au-nr=3 --random
Sorry for a bit late answer, I was quite busy last days. Anyway, here are the results:
$ sudo flashbench /dev/sdb --open-au --erasesize=$[8*1024*1024] -- blocksize=$[8*1024] --open-au-nr=3 --random 8MiB 1.95M/s 4MiB 18.3M/s 2MiB 17.4M/s 1MiB 17.4M/s 512KiB 17M/s 256KiB 15.7M/s 128KiB 14.5M/s 64KiB 14.1M/s 32KiB 9.37M/s 16KiB 6.18M/s 8KiB 1.63M/s
# This one is fast, suggesting 8MB erase blocks?
# this is slow for 8 MB erase blocks but fast for 4 MB or 6 MB erase blocks sudo flashbench /dev/sdb --open-au --erasesize=$[4*1024*1024] --blocksize=$[8*1024] --open-au-nr=6 --random
$ sudo flashbench /dev/sdb --open-au --erasesize=$[4*1024*1024] -- blocksize=$[8*1024] --open-au-nr=6 --random 4MiB 1.09M/s 2MiB 690K/s 1MiB 387K/s 512KiB 212K/s ^C
# This one is really slow, so it's confirming 8 MB erase blocks?
# if the previous one was fast, this will be fast for 6 MB erase blocks but slow for 4 MB sudo flashbench /dev/sdb --open-au --erasesize=$[6*1024*1024] --blocksize=$[8*1024] --open-au-nr=6 --random
$ sudo flashbench /dev/sdb --open-au --erasesize=$[6*1024*1024] -- blocksize=$[6*1024] --open-au-nr=6 --random 6MiB 2.18M/s 3MiB 1.32M/s 1.5MiB 660K/s 768KiB 332K/s 384KiB 169K/s
# Here it's also slow, but that was expected.
If you come to a result with the above, please also send the output of 'lsusb' so I have all the information to add to the wiki page.
Here's the relevant lsusb line:
ID 125f:311a A-DATA Technology Co., Ltd.
JM