by Robert Tanner on August 23, 2010 in Storage
Looking to upgrade your PC with a fast SSD? How fast do you want it? If you answered “ultra fast”, then OCZ’s RevoDrive is worth a look. With read speeds of 500MB/s and beyond, a PCI Express interface, and a modest price premium, this SSD is hard to ignore. We’re on the bleeding-edge here, though, so this drive isn’t without a few caveats.
This test is perhaps the most important in our battery of benchmarks as it gives us truly real-world results. It is designed to simulate three kinds of multi-tasking scenarios in order to see how well the storage drive can cope with concurrent workloads. The better a drive performs here, the quicker and more responsive it should feel in everyday tasks. It shouldn’t need to be said that this is where SSDs shine and where traditional HDD bottlenecks are most directly experienced.
In addition to stressing the controller with a demanding, large queue to sort through (NCQ support helps here), this test will give any weak controller a complete panic attack by overwhelming it with simultaneous random read/write operations to juggle with large sequential writes, which is the Achilles’ heel of many cheaper SSDs.
Queue depth and IOPs optimizations have long been a strength of Intel’s own SSDs, however, there is enough differing types of workloads here that regardless of drive, every SSD should see some part of the workload playing to its own unique strengths and weaknesses in some fashion. A good-quality SSD should allow the system to remain responsive as the tasks are carried out in the background at all times. (Please for your own safety don’t try this at home on your HDD!)
Our Medium test consists of the following:
- Playback of a 56MB FLAC music file in Winamp.
- 50 ~8MB images queued to open in Photoshop CS4.
- Opening of three Excel, three Word, and one PowerPoint files (various large sizes, for example one Excel file consists of an actual 72MB database).
- Browsing to four different websites in Firefox.
- Extraction of a 1GB RAR containing numerous “program file” folders.
- Extraction of an 893MB ZIP containing 100 RAW images.
- Transfer of a 7.16GB file to a second partition on the same drive
- Viewing of two PDF documents.
- Viewing of two small RAR utility archives
- Execution of four small utilities
Our Heavy test consists of all-the-above in addition to a full Anti-Virus scan running concurrently in the background with the start of the test. The AV scan uses a static, unchanging 5.1GB test folder that contains 19,748 files and 2,414 sub-folders created from the Program Files directory.
Granted, even with a Core i7 processor, no computer user would be performing all of these tasks concurrently unless they wish to see their computer go unresponsive for up to 30 minutes at a time, but with an SSD, this is almost child’s play! If you think we are exaggerating then just look at what a fairly typical SATA HDD is able to offer, which is a representative sample for any other desktop HDD.
Last but not least, the Light test changes things slightly. This test is a batch file dropped into the Startup folder designed to load several programs as soon as Windows 7 reaches the desktop. This light test will open four websites in Firefox, load five images in Photoshop CS4, start playing our favorite 8 minute (56MB FLAC) music file in Winamp, and open a single large Word, Excel, and PowerPoint document, in addition to a single PDF file. For this specific test in particular, we start measuring from the moment the power button is pressed to the moment the last program and all files have been fully loaded ready for use.
In the time required for a regular hard disk drive computer to boot, it is possible for an SSD to boot and have fully loaded a multitude of programs and files that you regularly use, as the nearly two and a half minutes for the HDD clearly illustrates. For the test platform base level boot time is typically a rather sedate 55 seconds. Loading several Microsoft Office documents in addition to Photoshop only adds an average of 12-15 seconds, in contrast to the mechanical drive which sees its load time more than double by comparison.
Then we have a drive such as the Revo, which one-ups the average SSD and manages to complete the Light load scenario in less than one minute. The average computer user is lucky to have a responsive desktop by one minute, let alone all of their regular programs already loaded and running!
Moving on to the Medium workload scenario the Revo has no trouble shaving a bit over 40 seconds from the best times, which is more impressive considering the test platform is contributing to the bottleneck here. Just as with the Iometer results, it is apparent the Revo’s dual-controller advantage makes it the best choice for intensive I/O workloads.
To add a little perspective, in a little over four minutes the RevoDrive is able to almost halve the time required by the X25-M G1 SSD, partly due to the X25-M G1’s 80MB/s write limitation and lack of TRIM support , yet the competition fairs even worse. The discontinued 40GB V series drive only has half of the normal 10 flash channels and this, as well as the small drive capacity, limit it to a time more than double that (9:11) of the V+ series 128GB SSD. For those like me that are too lazy to do the math, the outclassed mechanical hard disk drive completes the identical scenario in not quite 14 minutes. That’s nearly 10 minutes of time saved just by upgrading to a good quality SSD right there.
As mentioned above the heavy test uses the exact same scenario as the medium test, but includes a concurrent anti-virus scan of a static test folder. Predictable or not, in a five minutes and thirty-two seconds the OCZ RevoDrive sets a new record here as well, knocking 99 seconds off the previous best result. Five and a half minutes is far more palatable than the platter-based drive time which on average required an eye-watering 23 minutes and 42 seconds to complete the same workload.
One thing not shown by our graphs that we like to address is both result consistency and system responsiveness. With the medium scenario the Revo performed all five runs with results that were consistent, although due to the inability for TRIM to pass through the RAID array results did vary a bit more than seen by either the Vertex, Vertex 2, or Nova drives which all support TRIM. As one might expect the Revo never suffered a loss in system responsiveness. Weaker controllers such as the “V” drive’s JMicron controller by comparison had some results differing by as much as several minutes between ones in addition to the system becoming unresponsive, which was clear evidence the controller was becoming overwhelmed from the workload.
Primarily what to take away from these results are which SSD controllers are best suited for high drive I/O workloads. Even with the most brutal of scenarios where several gigabytes of data are simultaneously accessed and written to the drive, SSDs have little to no trouble juggling the loads while maintaining a responsive system. The 120GB OCZ RevoDrive has no trouble stomping over all other single-controller drives in our graphs, and even after reaching a thoroughly “dirty” state results still easily outpaced the competition.