The SSD get its speed from 2 things (related to cluster size)
1) The very fast seek/access time in write and read of SSD. They perform write/read in parallel.
2) Cache in both write and read
In read: Having a smaller cluster size, you’ll get a penalty in read speed (because it’ll have to read in more clusters for the data but remember they read in parallel) but since the SATA speed is the bottle neck in the SSD read performance, you’ll may not even see the penalty in test, let alone actual usage experience.
In write: The penalty is small because of the write cache. The SSD write in chunk of 4096k bytes or 8 of your clusters. Since you always get the benefit of a write cache, so what if the SSD waits a little before commit writing to 8 different clusters on your SSD? If you don’t have cache in the SSD, then you’ll take a larger penalty in this department. In the worst case, the penalty you see is comparable to having a SSD with less parallel channels. So your 512GB is as “slow” as a 128GB SSD, that’s still smoking fast.
In the old mechanical HD, clusters are located physically close to each other to reduce seek time. A smaller cluster size means less chance of a file fitting in continuous cluster block and the head has to jump around to different HD track to read a file (fragmentation). In SSD since reading are done in parallel, there’s no fragmentation issue and therefore penalty in cluster size is significantly reduced.
Facebook
Twitter