Mixed Workload And Steady State
80 Percent Sequential Mixed Workload
Our mixed workload testing is described in detail here, and our steady state tests are described here.
Mixed workloads give us a better measuring stick for determining real-world client SSD performance. After all, it's rare to read or write exclusively on a drive with an operating system installed on it, since so much activity happens in the background that you never see.
In previous tests, the SSD370S and two other SM2246EN-based SSDs clumped together on the charts with very little separation. Now we do see some white space between them as Transcend's SSD370S trails the BX100 and Reactor.
80 Percent Random Mixed Workload
Our random mix also differentiates the SM2246EN-powered drives. This time, the Reactor and SSD370S trail Crucial's BX100. All of the SSDs trail the 850 EVO though (a common theme in every mid-range review these days).Â
Sequential Steady State
SATA is a half-duplex interface; it can't read and write at the same time. Some drives have problems queuing data, so when commands are stacked in a mixed workloads, performance is affected, resulting in the curve with high reads and writes like you see here. Most of the TLC-based drives give this test a new purpose. In it, we're able to identify native TLC write speed using 128KB blocks.
Again, the SSD370S trails both of the other SM2246EN-based drives. Interestingly, in the workstation (70 percent read) and desktop (80 percent read) workloads, some of the TLC drives actually perform a little better.
Random Write Steady State
We don't use the random 4KB steady state test to evaluate this enterprise-oriented workload condition. After all, these drives weren't meant for that application. Rather, we run it as a means of gauging performance consistency, and to identify the drives that may work well in desktop RAID. We've yet to find a value-class SSD that performs exceptionally well consistently once it's in steady state. The Transcend SSD370S is no exception.