We have been busily preparing a new storage test suite that features Vista Ultimate 64. Call us crazy for using Vista at this point, but in order to properly test new technologies like ReadyBoost, ReadyDrive, and Low-priority I/O, this meant a change to the latest (but not greatest) that Microsoft offers. This change, like most when moving to Vista, has not exactly been smooth. The loss of our Intel IPEAK test suite is one that stings, as it is not compatible with Vista, but we will still provide selected numbers when appropriate. However, we believe our final selection of benchmarks will be able to provide a well-rounded look at a storage device’s performance from an actual application viewpoint. We will still provide a select number of theoretical benchmarks but our emphasis will be on how the storage device performs in a variety of actual benchmarks ranging from gaming to database creation.
The drive manufacturers are currently busy offering second-generation hybrid and PMR based mechanical devices but the real news in storage technology continues to be the rapid developing SSD market. We provided an early first look at the latest generation SSD drive from MTRON and since then have been bombarded not only with requests for additional tests, but drives from a variety of manufacturers that feature Super Talent, Samsung, MTRON, and MEMORIGHT.
We were very excited when our friends at DV Nation contacted us about the recent arrival of the latest generation SSD products from MTRON and MEMORIGHT that feature sustained read rates starting at 80MB/sec and going up to 112MB/sec. While these read speeds are impressive, write speed is the one area that SSD drives have generally been weak. The latest controller technology is solving this weakness. As an example, sustained write speeds are up to 105MB/sec in our sample drive from MEMORIGHT . These specifications far exceed those of the latest SanDisk and Samsung consumer SSD products that average 67MB/sec read speeds and 45MB/sec write speeds.
Another area of interest is the capability of the PATA based SSD drives to compete with their SATA cousins. In fact, the overall application results of the PATA based drives are equal to and sometime exceed the SATA drives, although their theoretical results show significantly lower scores at times. We are still researching this phenomenon as it occurs on chipsets from Intel, JMicron, Marvell, AMD, and NVIDIA. We hope to have answers before we lose our hair or the roundup, whichever comes first. It appears at this time that both Dave and I will be bald first.
We received numerous requests for write results on the MTRON drives, so we are providing those results today along with initial results from the MEMORIGHT SATA and MTRON PATA drives. We will be back before CES coverage starts with our first SSD roundup that features six different drives along with some interesting RAID results from our 64GB SSD units. Our numbers today were generated with our new test bed that features an ASUS P5N-T Deluxe motherboard based on the NVIDIA 780i chipset, Intel Q6600 Quad Core processor, and 4GB of OCZ PC2-6400 Flex memory. Our change back to an NVIDIA platform is due to the Intel ICH8/ICH9 chipsets capping SSD performance to 80MB/sec rates, something our NVIDIA or AMD chipsets do not suffer from at this time.
HD Tach 3.01
MTRON 32GB MSP 7000
MTRON 32GB MSD PATA-3025
Memoright 32GB MR25.1-032S
Western Digital Raptor 74GB 0ADFD
Our HD Tach test results heavily favor the MTRON 32GB SATA unit in sustained read rates. The MTRON's average read rate of 112.3MB/sec is about 33% better than the WD Raptor drive at 75.5MB/sec. The biggest difference is the MTRON drive holds the 112MB/sec transfer rate across the entire drive while the Raptor progressively decreases to around 52MB/sec at the end of the drive. The previous MTRON STR results were 95.1MB/sec, which is about 17% slower than the new drive.
The most interesting aspect is the average 83.2MB/sec write speed of the MTRON 32GB SATA drive, just 7% slower overall than the Raptor. However, the minimum write speed of the MTRON SATA unit is around 68MB/sec compared to 20MB/sec on the Raptor. Both of these pale in comparison to the MEMORIGHT 32GB drive that features a sustained write speed of 104.9MB/sec, the fastest we have tested to date. The MTRON 32GB PATA drive features very competitive read rates at 103.4MB/sec but HD Tach reports average write speeds at 39.4MB/sec. However, write speed performance in application testing shows it to be extremely close to the MTRON SATA drives. We will have an update in our roundup, but for now, the PATA drive looks to be a winner for those needing to upgrade an older notebook or extreme benchmarkers needing the fastest IDE drive around.
The burst rates of the Raptor at 130.7MB/sec are slightly higher than the 96~118MB/sec rates on the SSD drives, but burst rates are not nearly as large of a factor as other indicators. Access times greatly favors the SSD drives at .1ms compared to 8.3ms on the Raptor drive. That about does it for today, we will be back shortly with a full performance review on these very fast, but also expensive drives.
Facebook
Twitter
RSS
18 Comments
View All Comments
Reflex - Thursday, December 27, 2007 - link
While the ability to write remains intact, they are far more durable than a standard HDD. Once you start running out of writes(cells typically can be written to only a few hundred thousand times, wear leveling can effectively make that a few million), the drive starts having issues, in many cases it will simply cease functioning.A few million writes sounds impressive, but on a typical user system thousands can happen in a day, and over time the drive will start to fail or simply give up the ghost suddenly. I have seen both scenerios, although typically the drive simply became inaccessible one day without warning.
ikjadoon - Thursday, December 27, 2007 - link
It is important to remember that access time and transfer rate measurements are mostly diagnostic in nature and not really measurements of "performance" per se. Assessing these two specs is quite similar to running a processor "benchmark" that confirms "yes, this processor really runs at 2.4 GHz and really does feature a 400 MHz FSB." Many additional factors combine to yield aggregate high-level hard disk performance above and beyond these two easily measured yet largely irrelevant metrics. In the end, drives, like all other PC components, should be evaluated via application-level performance. Over the next few pages, this is exactly what we will do. Read on!aperson2437 - Thursday, December 27, 2007 - link
Check out this review:http://www.nextlevelhardware.com/storage/battleshi...">http://www.nextlevelhardware.com/storage/battleshi...
Maybe AnandTech can do a similar review with a bunch of MTron drives in Raid0. Maybe Mtron has a new series of SSDs about to come out with new and improved controllers designed to work with the latest and greatest mobos for Intel and AMD processors. They should send them to AnandTech. AnandTech does a pretty good job in their reviews ... objective and well written stuff usually.
testmeplz - Thursday, December 27, 2007 - link
Those raptor write scores can't be right.Bladen - Thursday, December 27, 2007 - link
Aren't small random writes the biggest killer for SSDs in terms of performance?leboss - Friday, December 28, 2007 - link
Yes, indeed. It's very frustrating to read SSD reviews such as this one where this issue is completely ignored. Where are the random write performance numbers?Since SSDs have very poor random write performance, I don't see how they could be useful for most applications. I can see how they can be useful in laptops, but what do you do with them other than that and some niche applications (such as military where ruggedness is important)? I guess they can serve as a cache for mostly read-only traffic, but in that case the competition is DRAM, not disk drives, so the review should compare the price/performance of DRAM to that of flash-based SSDs.
Sivar - Thursday, December 27, 2007 - link
It's too bad that benchmarks like HDTach are being used, which have nothing to do with real-world performance.STR - Sustained Transfer Rate, is useful only if you are streaming very high-data rate video (and even then, only one or two video streams, since past that the hard drive has to grind).
Intel IPEAK is a great benchmark (otherwise StorageReview.com wouldn't have been the first review website to use it), and it is unfortunate that it is no longer usable, but until a replacement is made, please use something (say, a timed real application) rather than yet another STR benchmark which means nothing.
The performance bottleneck of most SSDs, at least those which use FLASH memory, is a large number of SMALL writes to disk. HDTach tests SUSTAINED transfer of large streams of data, which almost never happens with any real game or major application.
I really like Anandtech, and I think the reviews are usually great, but there is a reason most of those in the know go elsewhere for hard drive reviews...
Chadder007 - Thursday, December 27, 2007 - link
Now just to wait for them to get significantly cheaper. :)