OWC Mercury Electra 3G MAX 960GB Review: 1TB of NAND in 2.5" Form Factor
Performance Over Time & TRIM
To see how the Mercury Electra behaves when pushed into a corner, I filled the drive with incompressible data and then tortured it with incompressible 4KB random writes (100% LBA space, QD=32) for 60 minutes:
| OWC Mercury Electra 3G MAX 960GB - Resiliency - AS SSD Sequential Write Speed - 6Gbps | |||
| Clean | After Torture | After TRIM | |
| OWC Mercury Electra 3G MAX | 187.5MB/s | 185.0MB/s | 206.7MB/s |
The results I got are rather unusual, although logical; performance does not degrade noticeably when the drive is tortured. The reason for this is the massive capacity and slow random write speed. It takes two hours to just fill the drive with sequential data, so 60 minutes of torturing doesn't even do one full drive write. IOmeter showed the random write speed to be around 20MB/s, which equals 72GB of host writes in 60 minutes. For comparison, most SandForce drives are writing at over 100MB/s at the end of the torture, so that works out to be ~360GB in one hour.
It seems that TRIM actually improved performance, but I don't believe that's true. The problem with the 960GB Mercury Electra is that it cannot be secure erased due to the internal RAID controller. Hence we had to do sequential write passes to restore performance after every test, which unfortunately doesn't lead to as ideal a scenario as secure erase. Moreover, I saw huge performance variation when running AS-SSD multiple times. I reran AS-SSD straight after the post-TRIM run and write speed was only 132.7MB/s. A third run was ~190MB/s again so it's hard to draw any clear conclusions.
Originally published at https://www.anandtech.com/show/6200/owc-mercury-electra-3g-max-960gb-review-cramming-1tb-of-nand-inside-25-chassis (Wayback Machine).