manfid = 0x000003 and, oemid = 0x5344

$ sudo ./flashbench -a /dev/mmcblk0 --blocksize=$[4 * 1024] align 2147483648        pre 3.47ms      on 3.74ms       post 2.81ms diff 601µs align 1073741824        pre 3.84ms      on 3.99ms       post 3.15ms diff 494µs align 536870912 pre 3.95ms      on 4.05ms       post 3.46ms     diff 349µs align 268435456 pre 3.42ms      on 3.68ms       post 3.2ms      diff 372µs align 134217728 pre 2.86ms      on 3.13ms       post 2.72ms     diff 341µs align 67108864  pre 2.93ms      on 3.35ms       post 2.84ms     diff 458µs align 33554432  pre 3.9ms       on 4.34ms       post 3.82ms     diff 479µs align 16777216  pre 2.96ms      on 3.23ms       post 2.26ms     diff 624µs align 8388608   pre 2.06ms      on 2.66ms       post 2.22ms     diff 519µs align 4194304   pre 2.98ms      on 3.31ms       post 2.23ms     diff 699µs align 2097152   pre 4.21ms      on 4.5ms        post 3.86ms     diff 461µs align 1048576   pre 4.38ms      on 4.63ms       post 3.95ms     diff 467µs align 524288    pre 3.03ms      on 3.22ms       post 2.85ms     diff 280µs align 262144    pre 2.76ms      on 2.98ms       post 2.45ms     diff 371µs align 131072    pre 2.47ms      on 2.7ms        post 2.48ms     diff 228µs align 65536     pre 2.58ms      on 2.97ms       post 2.46ms     diff 451µs align 32768     pre 2.48ms      on 2.69ms       post 2.5ms      diff 194µs align 16384     pre 3.09ms      on 3.28ms       post 2.7ms      diff 383µs align 8192      pre 2.78ms      on 3.01ms       post 2.76ms     diff 238µs

Erase Block size assumed to be 4MB.

$ sudo ./flashbench -O --erasesize=$[4 * 1024 * 1024] --blocksize=$[4 * 1024] /dev/mmcblk0 --open-au --random 4MiB    4.51M/s 2MiB    4.15M/s 1MiB    3.58M/s 512KiB  3.25M/s 256KiB  4.01M/s 128KiB  3.46M/s 64KiB   3.43M/s 32KiB   2.95M/s 16KiB   2.97M/s 8KiB    2.19M/s 4KiB    1.56M/s

$ sudo ./flashbench -O --erasesize=$[4 * 1024 * 1024] --blocksize=$[4 * 1024] /dev/mmcblk0 --open-au 4MiB    4.11M/s 2MiB    3.01M/s 1MiB    2.95M/s 512KiB  2.72M/s 256KiB  2.52M/s 128KiB  2.57M/s 64KiB   3.36M/s 32KiB   2.68M/s 16KiB   2.3M/s 8KiB    1.8M/s 4KiB    1.51M/s

$ sudo ./flashbench -O --erasesize=$[4 * 1024 * 1024] --blocksize=$[4 * 1024] /dev/mmcblk0 --open-au-nr=1 --random 4MiB    4.21M/s 2MiB    4.28M/s 1MiB    3.16M/s 512KiB  4.4M/s 256KiB  3.5M/s 128KiB  3.52M/s 64KiB   4.17M/s 32KiB   3.75M/s 16KiB   2.56M/s 8KiB    2.66M/s 4KiB    1.75M/s

$ sudo ./flashbench -O --erasesize=$[4 * 1024 * 1024] --blocksize=$[4 * 1024] /dev/mmcblk0 --open-au-nr=1 4MiB    4.2M/s 2MiB    3.84M/s 1MiB    4.34M/s 512KiB  4.36M/s 256KiB  3.53M/s 128KiB  3.8M/s 64KiB   4.32M/s 32KiB   3.85M/s 16KiB   3.78M/s 8KiB    3.34M/s 4KiB    2.95M/s

It gave me a consistent throughput as long as upto 7 erase blocks were open. Performance with 8 open erase blocks is not that bad either.

$ sudo ./flashbench -O --erasesize=$[4 * 1024 * 1024] --blocksize=$[4 * 1024] /dev/mmcblk0 --open-au-nr=7 --random 4MiB    3.73M/s 2MiB    3.02M/s 1MiB    3.14M/s 512KiB  2.88M/s 256KiB  2.48M/s 128KiB  2.34M/s 64KiB   2.16M/s 32KiB   1.96M/s 16KiB   1.55M/s 8KiB    1.04M/s 4KiB    1.01M/s

$ sudo ./flashbench -O --erasesize=$[4 * 1024 * 1024] --blocksize=$[4 * 1024] /dev/mmcblk0 --open-au-nr=7 4MiB    3.4M/s 2MiB    3.1M/s 1MiB    3.01M/s 512KiB  3.12M/s 256KiB  2.91M/s 128KiB  2.87M/s 64KiB   2.58M/s 32KiB   2.78M/s 16KiB   2.37M/s 8KiB    1.85M/s 4KiB    1.28M/s

$ sudo ./flashbench -O --erasesize=$[4 * 1024 * 1024] --blocksize=$[4 * 1024] /dev/mmcblk0 --open-au-nr=8 --random 4MiB    3.66M/s 2MiB    2.94M/s 1MiB    3.04M/s 512KiB  2.55M/s 256KiB  2.46M/s 128KiB  2.27M/s 64KiB   1.96M/s 32KiB   1.53M/s 16KiB   1.38M/s 8KiB    955K/s 4KiB    823K/s

$ sudo ./flashbench -O --erasesize=$[4 * 1024 * 1024] --blocksize=$[4 * 1024] /dev/mmcblk0 --open-au-nr=8 4MiB    3.36M/s 2MiB    2.69M/s 1MiB    2.66M/s 512KiB  2.56M/s 256KiB  2.41M/s 128KiB  2.4M/s 64KiB   2.14M/s 32KiB   2.04M/s 16KiB   2.01M/s 8KiB    1.65M/s 4KiB    1.2M/s