SATA 6Gbps vs SATA 3Gbps

SATA 6Gbps is often referred to as S-ATA3 or S-ATA III, which is actually wrong according to the SATA Naming Guidelines. It should be referred to as SATA 6Gbps or SATA Revision 3.0.

SATA 6Gbps will provide speeds up to 6Gbps (around 600MB/s when overhead is taken into account) and is fully compatible with earlier revisions and uses the same connectors like we know from earlier SATA revisions.



For now nor Intel or AMD have integrated SATA 6Gbps in their chipsets so we have to rely on an extra chip. In this case the Marvell 9128. Gigabyte supplied us with a P55A-UD6 motherboard and a Seagate Barracuda XT with SATA 6Gbps interface for testing.
To test the performance of the drive we use 3 benchmarks, Crystal DiskMark, HD Tune Pro and Attobench. We compared the scores with the same drive connected to the Intel controller.

Hardware used for this benchmark
CPU	Intel Core i5 750
Mainboard	Gigabyte P55-UD6
Videocard	Geforce 8500GT
Memory	2x 2GB PC3-12800 Corsair XMS @ 1.65V
Storage	Seagate Barracuda XT 2TB Buffalo D-H1.OTU3 1TB USB3.0
PSU	Zalman ZM500 HP
Operating System	Windows Vista Business 64b

When we look at the differences between the tests we see they are almost non-existant. The hard drive is capable of reaching speeds up to 144MB/s. Not bad for a mechanical hard drive, but based on this benchmark there is not really a need for faster SATA interface.




The same is true for the Atto Disk Benchmark tests. Comparing SATA 6Gbps to SATA 3Gbps, the first seems to outperform on read speeds, albeit slightly, but the difference is not more than a few MB/s.



When looking at the read speeds in HD Tune Pro we see a slightly better performance with the SATA 6Gbps controller in the first part of the test compared to the SATA 3Gbps. On this controller there is about 7 MB/s extra performance when compared to the SATA 3Gbps controller. The minimum speeds stays the same, as well as the access time that is comparable. The average is only slightly higher, due to the similar performance for the most of the test.