128 KB Sequential Performance: RAW, Windows, And Mac
Sequential Read Performance
In sequential read testing, the 256 GB and 512 GB Vertex 4s deliver identical performance, though it's hard to tell the lines are overlapping in the chart above. Regardless, it's clear that testing physical blocks doesn't favor OCZ's newest SSD. With speeds just above 200 MB/s, the Vertex 4 doesn't appear to live up to the Vertex namesake. Even Intel's SSD 320 enables faster transfers.
The results clearly change when we switch to testing within a file system. In NTFS specifically, both Vertex 4s start out at ~210 MB/s and climb as high as 480 MB/s after hitting a queue depth of 16. Although Samsung's 830 and OCZ's Vertex 3 deliver better performance, the Vertex 4 is not far behind.
Almost all of the SSDs are affected by host caching in a Mac environment, which is why you see most drives start around 500 MB/s. Interestingly, the 512 GB Vertex 4 exceeds the Vertex 3's performance, following a similar performance path as the 256 GB Crucial m4. The 256 GB Vertex 4 falls a bit behind, though. It's still able to beat Intel's SSD 320, but falls shy of OCZ's 240 GB Vertex 3.
Sequential Write Performance
Examples include Application Installation, Document Backup
SandForce-based SSDs reign supreme when it comes to writing compressible data, which explains why the 240 GB Vertex 3 maintains such a huge lead. At a queue depth of two, the Vertex 3 is the only SSD able to break through 500 MB/s. However, the 512 GB Vertex 4 doesn't fall far behind, which is encouraging considering that OCZ actually cites lower write speeds on the 480 GB Vertex 3, meaning the bigger drive would be slower than what we see here.
Zeroing-in on drive performance faced with incompressible information, the Vertex 3 delivers close to 300 MB/s. Meanwhile, the Vertex 4 continues to excel since it doesn't rely on compression to achieve better results.
On our Mac, we again see a lead favoring the Vertex 4. Even at a queue depth of one, both Vertex 4s maintain a 15 MB/s edge over the older Vertex 3. This lead mostly continues as queue depth increases up to around eight.