Test System and Results
The test system used for benching these products is the same we use for all storage testing.
- ASUS Sabertooth 990FX Motherboard
- AMD FX-4100 Quad Core Bulldozer
- 8GB Corsair EVO 1866 9-10-9 DDR3
- EVGA 580GTX Superclock
- Corsair TX-750 PSU
- Western Digital 1TB Greenpower drive for OS
- Cooler Master 212+ Heatsink
- Custom Built Bench Station
- ADATA SP900 Premier Pro SSD 128 GB
- SilverStone Raven RS02 2.5” External Enclosure
Since none of these devices will be used for an operating system install, running the full gamut of storage testing software normally used seemed superfluous; Instead, only two software were used. ATTO Disk Benchmark was used with results in our standard cluster sizes 4K, 8K, 128K, 1MB and 8 MB. Crystal DiskMark X64 was run in both random and 0-fill data tests.
For comparison we pitted the EC03, EC04 and FP37 (connected to the EC03) against the onboard USB 3.0 controller on the Sabertooth 990FX, which utilizes an AsMedia ASM1042 controller.
Performance was fairly close on smaller file sizes, but the difference became more noticeable as the test file size increased. Despite the ATTO tests running a Queue Depth of 4, this is not really accurate as with no NCQ, performance is actually single threaded. This is notable in the very large file size tests as write speed surpasses read performance by 128K tests sizes on the SandForce driven SSD used for testing. The EC03 performed very well topping out at about 240 MBps Read and 250 MBs writes. The EC04 did not fare as well topping out at 200 MBps read speed and under 150 MBps writes. The onboard solution performed best overall with slightly lower read speeds at about 230 MBps but as high as 275 MBps write. With the FP37 connected to the EC03 and testing pass-through mode it managed to meet or even very slightly out perform the EC03 in small file testing. Once the file size increased however performance dropped to approximately 5% less than the EC03.
Crystal Diskmark results were similar to ATTO. With smaller file sizes performance differences were fairly small between any of the devices tested. At larger file sizes the EC03, FP37 and AsMedia controller traded blows for best performers. Due to the speed limits of the TI controller we saw in the ATTO test the EC04 performed just as well in the random data as it did in the 0 fill tests with only one exception. Lastly we see the high QD test only made a small improvement in those results across any device. Once again the FP37 performed at or less than 5% slower than he EC03 it was attached too.
