Motherboards Memory Storage Cases/Cooling/PSUs IT Computing Displays Mobile Mac CPUs & Chipsets Video Digital Cameras Linux Gadgets Systems Trade Shows Guides Home Increase Font Size Decrease Font Size Change Page Size
SSD Update - Memoright 32GB on the Desktop
SSD Update - Memoright 32GB on the Desktop
Date: May 21st, 2008
Author: Gary Key
 
 

We will update our recent Samsung/OCZ 64GB SSD article shortly with performance results from the latest Memoright MR25.2-032S GT drive provided by DVNation. In the meantime, we are providing a quick overview of results on the desktop today. This drive is designed for the enterprise or enthusiast user with deep pockets looking for top flight performance, reduced form factor and class leading thermals/acoustics. DVNation provided us with a total of eight drives for a special RAID performance article we are working on with the enterprise user in mind. 

 

The Memoright GT series comes in 32GB, 64GB, and 128GB capacity points with the 32/64GB drives featuring sustained 120/120 MB/s read/write specifications and the 128GB coming in at sustained 100/100 MB/s read/write rates.  Although read/write rates of current high performance SSD drives match or better most desktop SATA drives, the Achilles heel tends to be their random read/write rates.  We have not completed testing yet, but random reads are around 70MB/s and writes coming in around 40MB/s in our benchmarks.

 
Specifications -
 
 

The major differences between this drive and the Samsung/OCZ offerings is reduced seek times, improved sustained transfer rates, longer warranty period, and increased power dissipation numbers along with an MSRP that matches the Samsung based 64GB drive.  Our test bed setup and explanation of the benchmarks can be located in our last article.

 
Quick Results -
 
 

The Memoright drive posted impressive PCVantage scores until we reach the Windows Media Player and Application loading test suites. The Windows Media Player test consists of adding songs to the playlist while playing back music.  This test highly favors drives with fast read rates but the synthetic advantage that Memoright enjoys does not translate to this particular test.  We tried recreating the test and ended up with the Samsung drive barely edging out the Memoright drive by 3%.

In the application loading test, we are baffled (yes, that does happen at times) with the results.  This test utilizes Word 2007, Adobe Photoshop CS2, IE 7, and Outlook 2007.  Our actual application results show the Memoright drive loading Word, and Adobe Photoshop quicker than the Samsung drive while IE7 and Outlook did favor Samsung/OCZ.  We have contacted Futuremark to understand the exact load sequence and procedure as the Memoright drive loaded Photoshop quicker than any drive we have tested to date.  Overall, we must remember that while PCVantage is based on actual application testing, it reflects the pure performance of the drive or controller utilized and not the platform.

In our actual application results, the Memoright drive offers the fastest Game level load and copy speeds. The Nero Recode test was disappointing, but to date, none of the SSD drives have fared well in this test that features sustained write operations of large data blocks in sequential order.  The cache available on the mechanical drives offer buffering which greatly improve their performance.  The Memoright drive did improve upon the Samsung unit by seven seconds in the WinRAR test and almost matched the VelociRaptor results.

 

First Thoughts -

Overall, this drive is very fast in these particular benchmarks and others not yet reported.  From a purely subjective viewpoint, the entire platform seems more responsive with a fast SSD drive. This holds especially true in repeat operations with the same application, as the drive response is nearly instantaneous in most cases.  However, storage capacity and price will continue to be the limiting factors in general acceptance of SSD technology.  Nevertheless, it is nice to know that performance is no longer a question mark and with each successive generation, performance will continue to surpass that of a mechanical drive.


22 Comments
Username:
Password:
Raid Throughput Limits by LTG, 629 days ago
Great info.

It would be helpful for the full article if we could get RAID0 results with the Intel ICHR on board controller.

The reason is very practical - A lot of people doing video use a RAID0 volume for editing and it would be great to know that there were no throughput issues if you save money and use the free Intel controller.

(Obviously in this scenario data is moved off the volume after editing)

Compositing multiple uncompressed HD streams really benefits from this.

Reply
SSD on the desktop by tremelai1, 629 days ago
The largest issue with any SSD is the fact that most desktop class OS's are not tuned for them.
XP in particular is hard coded to expect a ~10-12ms access time.
I don't claim to be a guru on the IO habits of MS OS's but most modern OS's have implemented many kernel code level 'gymnastics' to avoid hard disk latency that is simply not an issue with SSD.
When these algorithms are presented with low latency drives, such as a SSD, they don't compensate for the added performance.

Example:
The Linux elevator CFQ IO algorithm will sweep the disk, perform all requests when the head passes the required location. IO's are investigated and re-ordered based upon several factors, the largest being the head location. With SSD, the head is simulated and instantaneous.
Switching the IO algorithm under Linux to No-op (first-come first-serve with no re-ordering) should allow the SSD to shine.

I found that with large SAN arrays and RAID array's with battery backed-up write cache, disableing IO scheduling optimizers increased performance significantly on database TPC tests and conversely killed performance on JBOD arrays with no cache. It stands to reason that the same effect would hold true on SSD.

I can't find any way to disable these types of IO optimization in Windows like in the example above.

Reply
Alternatives by InternetGeek, 628 days ago
One week ago I installed the Unreal Tournament 2003 Demo in my "Poor Man's SSD", a 4gig OCZ 60X SD Card, just to see how much faster a solid state memory could help my system. The game installed in lest than 5minutes. It loads in less than 30seconds and most maps start in less than 20seconds. This all in a 2gig Core T2400 Laptop (Toshiba P105-S921). It's not the fastest of set ups but I did notice the difference. I can actually see that no more seeking is performed as the system prepares to run the game (CPU loaded but no seeking being carried out).

It's sad though that most current SSD drives seem to suffer from not-so-good random seek times (you seek before reading or writing) performance. If they overcome this caveat I would get a SSD drive for my laptop the moment the storage/price ratio enters SATA territory.



Reply
Alternatives by InternetGeek, 628 days ago
One week ago I installed the Unreal Tournament 2003 Demo in my "Poor Man's SSD", a 4gig OCZ 60X SD Card, just to see how much faster a solid state memory could help my system. The game installed in lest than 5minutes. It loads in less than 30seconds and most maps start in less than 20seconds. This all in a 2gig Core T2400 Laptop (Toshiba P105-S921). It's not the fastest of set ups but I did notice the difference. I can actually see that no more seeking is performed as the system prepares to run the game (CPU loaded but no seeking being carried out).

It's sad though that most current SSD drives seem to suffer from not-so-good random seek times (you seek before reading or writing) performance. If they overcome this caveat I would get a SSD drive for my laptop the moment the storage/price ratio enters SATA territory.



Reply
RE: Alternatives by InternetGeek, 628 days ago
Sorry for the double post.

Reply
What about SSD on intensive gaming like FSX? by Pok3R, 628 days ago
I wonder how a SSD will perform on FSX...that would be a great game test for a SSD; the drive has a non stop job of read/write the terrain data al around while you are flying and would be nice to know how faster than a normal HDD it performs!



Great info congrats and thanks!






Reply
Would it be possible to do a compile benchmark? by puffpio, 628 days ago
Since that kind of test involves reading a bunch of small text files, writing bunch of small object files, and mashing them all together

Maybe a test of compiling the anandtech website source? :D
or some very large project w/ thousands of files and pieces

as a software developer, the idea of a ssd hdd seems awesome especially in a laptop..but i wonder how it would handle project compiles

Reply
RE: Would it be possible to do a compile benchmark? by ravedave, 625 days ago
Why not do a kernel compile? Or perhaps build a whole bunch of popular open source software? It would be great to see how this turns out. At work we build over 1000 individual C# projects a night, they are small so it only takes ~1hr, but it would be interesting to see what sort of improvement could be had. There could be a problem with the compiling being CPU limited, but I'm not sure.


Reply
So by Sharky974, 628 days ago
I'm assuming pretty soon a SSD might be priced well enough to be a viable OS drive to speed things up, kind of a cache drive, while you'll use a second much bigger traditional hard drive for more archival type storage?

Cus that'll be pretty cool.

Reply
what about the new MATRON new drive by Huber, 628 days ago
i saw your matron review back then and i saw there is a newer model - will you be reviewing it soon?

Reply
comment by Visual, 628 days ago
Still quite unimpressive in most real-world situations, but I guess that's because of me buying too much into the SSD hype and having too high expectations.

Price is also gonna stay out of my reach for now.

However, I'm thankful for your reviews - at least now I know I'm not missing too much.

And I'm also looking forward to your RAID SSD tests - they could be a good indication of what manufacturers could accomplish if they made a single drive with internal RAID between its chips.
Perhaps you can also test out a RAID of some cheaper alternatives - namely flash cards + SATA adapters?


Reply
Intel SSD by Chriz, 628 days ago
I really hope the intel SSD drives coming out next month beat the performance of these drives by a large margin with the price being lower. I doubt it, but one can hope.

Reply
It's not Enterprise Class by davegraham, 628 days ago
Sorry to burst your bubble but this drive is in no way, shape, or form ready for anything but SMB and Consumer projects. What type of wear leveling, I/O optimizations are being done? Support? SLAs? seriously, please don't toss out "Enterprise" when discussing these things. I'd happily point you to STEC and their truly ENTERPRISE class SSDs in use in EMC's Symmetrix Storage Arrays but it's in a completely different class of SSDs altogether.

cheers,

Dave

Reply
It'll be a good technology eventually, that's for sure. by gochichi, 628 days ago
These products are at the "subsidize the research and development" stage. This technology will eventually be cheaper to manufacture than harddisks. If you think about it, a hard drive has really elaborate mechanical components. Precise movement at 7200RPM is difficult to produce, that's why there are so few hard drive manufacturers. This new technology is cheap and easy by nature and I expect to see a lot of new or non-storage companies enter this market.

The performance numbers are truly only marginally better than mechanical drives and all of the subjective reassurances in the world that the "system feels more responsive" are great but there are few (computer related) things that I would pay that kind of money for and a 64GB HDD is not one of them.

It shaves off 3 seconds out of 33? That is not much.

The good news however, is that while storage capacities (and the demand for them) are ever increasing, the core level components and most important software fits quite well in 128GB or less and so this technology should be competitive soon. A 128GB SSD and a 1TB HDD sounds about right on a desktop, and on a laptop you can do away with the 1TB HDD.

This reminds me of the move to LCD quite a bit, those early untis where so pricey and had a lot of problems and yet in terms of raw materials, shipping etc. it was a cheap technology. It is so much cheaper now to buy a 24" LCD that it ever was to get 24" CRT (There was like one model, the Sony $1800.00 one). I see no reason to be an early adopter, the industry will necessarily move in this direction, I don't have to promote it with my dollars. I am pretty sure that it will get to the point where the basic harddrives all use this cheaper technology.

Reply
Overall by ravedave, 628 days ago
It would be great if Anandtech could dig into why these drive's theoretical benefits don't materialize. Is it the operating system not taking advantage? Is it the limit of SATA? Do SSDs need cache like HardDrives?

Reply
RE: Overall by James5mith, 628 days ago
One of the problems is that yes, most SSD's do not have cache memory.

Although the Memoright does have 16MB of cache memory. This cache memory is what allows it to "smooth" out writes and nearly match Read/Write speeds. (whereas drives without cache have much lower write speeds than read speeds, I.e. the MTRON with 120MBps Read/90MBps Write.)

The other problem is that the Windows OS is coded to actually expect and work around the latency of traditional hdd's. Under Linux, you can actually disable this behavior (see SSD on the desktop by tremelai1, 22 hours ago) above.

Reply
RE: Overall by ravedave, 625 days ago
If that's true then I'd like to see Anandtech run some tests on Linux with the settings changed. I personally am not convinced that there is a fix that could be done to windows to make it faster with SSDs. Vista SP1 is recent enough that the engineers should have taken SSDs into account. I think the real gains to be had are improvements in the drive controllers. SSDs are made up of many chips, why not make each chip it's own "internal drive" and have an "internal raid 0"? Using a method like that you should be able to saturate SATA pretty quickly.

Reply
Smaller/less expensive SSD models might work well for many, including developers by lda, 627 days ago
I think the SSD vendors are missing the potential for a sizeable market. Speaking for myself, I would be interested in a much less expensive 4gb or 8gb model.

My primary use of a computer is software development. With the correct configuration, you can alter the location of many, if not most of the components of the operating system and compiler/ide.

For example, a registry entry is available to change "Program Files" from the default of
C:\Program Files
to what you specify. This applies to many other directories.

The idea would be identify those files that are actually used to do the specific tasks (such as software development) and get those on the SSD. In my case, the computer tends to stay on for days at a time, so "launch" of the o/s is less of a priority.

I think there are utilities from (the former) SysInternals and WinInternals that identify which files are actually used. The processs of getting the correct "working set" of files onto the SSD could possibly be semi-automated ... a'la Bart-PE

With tools like Visual Studio 98, 2003, 2005, 2008, etc., there are options to specify a "cascade" for the Include files, Lib files, etc. so that the first in the priority list could be on the SSD.




Reply
Could it be the HDD controller? by Targon, 624 days ago
Many people blame the drives for the extra latency we are seeing at certain tasks. Considering that the drive is connected via a hard drive controller of some kind, I would strongly suspect the problem is caused by that. Doing tests with the same drive on different chipsets with different hard drive controllers would be a good way to isolate this possibility. Even if you are looking at a 2.5 inch laptop hard drive, you can connect these drives to a normal desktop machine to do the tests and narrow down the sources of latency.

Drivers also can come into it since we have seen bad drivers before. Could it just be a function of drive controllers not really being designed with SSDs in mind?

Reply
Cheap alternative? USB Thumbdrive RAID??? by mooshi, 623 days ago
Since new motherboards come out with 8 or more USB ports (and if i'm not mistaken, each is its own dedicated channel, what's to stop one from hooking up a USB flash thumbdrive to each and setup a RAID? Thumbdrives are cheap, and if you get the right ones, they are pretty fast too.

I bought a Corsair GT a couple years ago (got it because of the read/write rates of 10+MB/sec), and while I paid good money for it, there've gotta be cheap, fast, non-trivially sized (2GB+) memory sticks available now.

Setup 8 x 4G thumbdrives shouldn't be too hard, just might have to find those extra USB headers on the motherboard and breakout to USB ports. It won't cost as much as an SSD, and if you ever get bored of it, you can always break the RAID and use the drives for other stuff.

The following is an old article (10/05) by Anand, showing a particular thing that should be kept in mind--the file sizes involved play a big role in the apparent transfer rates. Bigger is faster, so in setting up the RAID, it seems a large stripe size would be most beneficial:

http://www.anandtech.com/memory/showdoc.aspx?i=2549&p=14&cp=4

I figure someone has tried this, but I haven't found a good article on it yet...I figure 100+MB/sec read/writes have got to be possible with this.

Reply
RE: Cheap alternative? USB Thumbdrive RAID??? by Visual, 622 days ago
I'm not convinced USB ports on a mobo are really independent. But yes, there are ways around that anyway - several PCI cards with more USB ports, SATA to flash-card adapters, etc...
I really would love to see a review of some such configuration.

Reply
is this drive slc or mlc? by danieldsk, 622 days ago
i don't see that info anywhere...

thanks.

Reply
Comments Page 1 of 1





AnandTech.com Blog Categories
All categories
Anand's Macdates
Anand's Theater Construction
Anand's Updates
Cases and Power Supplies
CeBIT 2008
CES 2008
Computex 2009
Derek Decanted
Eddie's Got Game
Gary's First Looks
IT Computing general
Jarred's Musings
Kris's Corner
Raja's Ramblings
Rob's Experiences...
Ryan's Ramblings
Virtualization
What's New with Wes
Blank
Blank

Blank

Latest news by
DailyTech

 February 9, 2010

Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank
Blank

 February 8, 2010

Blank


more Blogs Discussions



pipeboost
Copyright © 1997-2010 AnandTech, Inc. All rights reserved. Terms, Conditions and Privacy Information.
Click Here for Advertising Information