Thank you! These results are very helpful.
Here are the results formatted a bit nicer:
Code:
Command line used: iozone -I -a -s 256M -r 4k -r 64k -r 512k -i 0 -i 1 -i 2
Output is in Kbytes/sec
[...]
random random
KB reclen write rewrite read reread read write
262144 4 1419 1499 4513 4448 3802 24
262144 64 5954 6818 18530 18580 18516 379
262144 512 7242 7781 15782 15826 16293 2160
These results show that when reading 4 KB blocks in random positions from the SSD, we get a throughput of 3800 KB/s (first row with a "reclen" of 4, in the "random read" column).
This is actually better than what I expected. With reference to my original post, it's about 8 times faster than a conventional hard drive (500 KB/s), but still about 6 times slower than an Intel SSD (22 MB/s). But the Intel SSD is probably around 10 times more expensive, so I definitely think these results are impressive.
3800 KB/s with a block size of 4 KB/s is around 1000 IOPS per second, that's definitely a step up from both a conventional hard drive and the earliest SSDs that made it into netbooks.
The random write speed, however, really falls throgh the floor: 24 KB/s, or 6 IOPS per second (that's a latency of ~170 milliseconds when doing small random writes to the disk!). But it seems we can't have everything, ie. the SSD in the Efika is probably optimized to do fast random 4 KB reads - and wisely so!
I probably should have asked you to perform the test like this though:
Code:
iozone -I -a -s 32M -r 4k -r 64k -r 512k -i 0 -i 1 -i 2
ie. writing a temporary
32 MB file to the disk instead of a
256 MB file (the test much have taken ages to complete at 24 KB/s random writes!). Thanks for enduring :).
By the way - does anyone know if it's a Sandisk iSSD that is in the Efikas?