S150 SX4 vs S150 TX4:

To compare our 2 controllers we used the results of the 4disk and 2disk striped arrays.



When comparing the results of the SX4 and the TX4, both running a 4disk striped array, we can clearly see the impact of the SX4-cache on test results when using small files: the WIN-read2 piles out above 80MB/s, while the TX4 hardly gets over 40MB/s! But we also must say that, in read1, the TX4 is faster than the SX4: over 10MB/s faster when using small files.

Another strange result: copying small files with the SX4 is slower than with the TX4. This is strange because a lot of these small files should be in the SX4s cache, meaning it doesn't have to perform as much read-operations as the TX4. Also remarkable: the big files were read 20MB/s faster with the TX4 controller (120MB/s <-> 100MB/s).

It seems that the SX4 is optimized for us with RAID5 arrays and less for other types. This is not a big problem of course, why else as for RAID5 would you buy this controller?

In Sandra we compared the random read/write results.


you can safely ignore the HDtach results as they are heavily flawed

Here the TX4 clearly is the king of the hill, as it writes 40MB/s faster than the SX4! Of course, the result of the SX4 isn't bad (80MB/s), but the difference between these two cousins is huge. Reading is also much faster on the TX4: with 70MB/s he's almost 40MB/s faster than the SX4.

As both controllers support 33Mhz and 66Mhz PCI, I decided to test how much difference with the higher bandwidth bus would be.




As you can see, the 66Mhz results are almost always higher than the 33Mhz’, but not earth shattering. The real advantage of the 66Mhz lies however in the fact that if you were to be using a gigabit NIC in one of the other 66Mhz PCI slots then both controllers have enough bandwidth to work at full capacity, where otherwise they would have been limited in a 33Mhz PCI bus environment.