The new SSD Thread (Benchmarks, Brands, News and Advice)

Look what I found...maybe it will help!!!

http://forum.notebookreview.com/showthread.php?t=162085

 

HI Gary. Thanks for the reply. Both HD and DVD are IDE and seems to be
on the same cable, like my old desktop. Too bad both device doesn't have
any jumper I can use. So I guess I am out of luck. Unless sony comes out
with new bios that can modify the cable/master/slave system.

 

Thanks. I know the thread. I had it fully memorized in my head.
Read it at least 10 times..

 

http://www.storagesearch.com/simple.html

http://news.softpedia.com/news/512GB-SSD-Drives-From-Zeus-Zeus-IOPS-That-is-52507.shtml

http://www.trustedreviews.com/storage/news/2007/04/20/Record-512GB-256GB-SSDs-Announced/p1

http://dodevice.com/32gb-to-512gb-ssd-cheap-and-fast-by-stec/

http://en.hardspell.com/doc/showcont.asp?news_id=1664

 

Flamenko:

Do you have any idea what could be the Read/Write speeds of the SSD in the Macbook Air ?

Thanks in advance

 

I have been trying to find out what the make and model is on the 64Gb SSD. Its either going to be the Samsung or Sandisk as contracts have not been established with any of the other manufacturers as of yet.

As it is a 64Gb, you will probably see read/write similar to the results of the Samsung 64Gb drive.... 50-80 read and around 40 write. As a matter of fact, the thought process seems to have changed from the best performing to that adequate to the consumer... so I am hearing average read/writes similar to this of 64Gb SSDs in a few brands in my discussions with the manufacturers.

I am curious why they havent advertised the specs of the SSD going into the Air. It may just be because they were targetting two manufacturers and hadn't decided yet.

 

I wrote this in another thread, but it also applies for this one I think, im using a HP 6715b with a 16 GB Mtron SSD, and sometimes when I reboot, the entire harddrive is corrupted, all partitions, chkdsk manages to recover a lot of files and put them in the found.000 folder, but many is lost.

Does anyone have any idea of what could be causing this, could there be some xp + ssd flaw, or something with the chipset?

 

The only two occurrences I know of thi just happen to be with XP, although that may be just a coincidence.

 

I did use Vista before, and had no such problems, although with vista the indexing service refused to work outofthebox, therefore I switched. Aaargh!

 

I dont use indexing and advise to turn it off in the tweak Thread and, as my system is right now...Vista Ultimate, 4Gb RAM, m1330 and the Mtron Pro SSD, I don't know of a faster laptop once tweaked. (Don't get me wrong, I know that there are more powerful but, tweaked...this system is absolutely amazing with the SSD).

 

Well, I depend on windows desktop search (to search in files, calendar etc), and even though there are alternatives, they all integrate comparably poorly into windows :/

With only a 16gb hd, Vista gobbles up a bit much space aswell.

 

On DVnation.com, they have benchmarks of the 1.8" zif hard hard drives.`

On the Samsung, it has an average of 55 read and 23 write.
On the Sandisk, It does have a good average read at 65. But the write starts of well at 23, but after about 15GB's of sequential writing, it drops to around 12. So the overall was about 16.

Here's the really odd thing with Sandisk. They made a new firmware for that 32GB SSD line, but have not released it to the public. So, it is possible that
the 64GB is indeed shipping with it and would have about 30mb/s write, but I have not been able to find any more info on the 64GB model nor on the new firmware.

 

I read on Dell.com that you may be able to get that firmware through Dell.

 

Ok, ive installed VISTA, XP + SSD with my computer simply isnt stable.

 

You will have to elaborate on how it is not stable. I will concede thought that all of these problems seem to be popping up with XP solely and NOT VISTA.

I will try to help but no promises and my last piece of advise will always be the same...upgrade to Vista and follow the Clean Install Thread.

 

Regarding the problem with the Sony TX + SSD, my Sony G11 has only one IDE channel listed under the IDE ATA/ATAPI controllers and has only one device (the HDD - current transfer mode UDMA 5) attached to that channel. The ODD is connected as a USB device so that power can be turned off to extend battery life.

1. What does the TX show when the HDD is installed? One or two devices on the IDE channel.

2. Did you try uninstalling the IDE ATA/ATAPI controller and (if the ODD is a USB device) the USB controllers? Then, when Windows reinstalls the controllers it will also reinstall the devices connected to those controllers.

John

 

Calculus,

Read the last 2 threads here...maybe this could be related. Gotta love Sony. Ide be calling support.

 

I just thought I might say hello, and that I'm having the same problem as Calculus, except I am just replacing a 60gb HDD with a 160gb HDD.
So the issue is not some special SSD-specific thing.

 

What is going to be standard in the near future, 1.8'' or 2.5" SSD? I can imagine that many people buying today's laptops with 2.5'' SATA HDD, will want to switch to 2.5'' SSD when the prices are down enough. Would perhaps not be so good if the 1.8'' is emerging as standard, unless there is an adapter available.

 

I heard that there is no speed difference between 1.8" SSD & 2.5" SSD as there are some difference between 1.8" HDD & 2.5" HDD.

Is that true?

 

I think there is no speed difference because of size.
But, there is much more assortment of 2.5", so for example you can buy a 2.5" Mtron, but not a 1.8" one. Of course, there is a big difference between an Mtron 2.5" and a SuperTalent 1.8", but you can't see a relevant difference if you compare a SanDisk 2.5" to a SandDisk 1.8", from the same model family. I mean, don't compare MLC vs. SLC drives or a newer model to an old one.

 

Sorry if this has already been asked...\

I was wondering flamenko's take on defragmenting an SSD drive. I read an article awhile back from diskkeeper or perfectdisk (can't remember which) that said SSD drives benefit from defragmenting. I can't help to think that they are just struggling to justify there software for when SSDs dominate the market. But maybe there is something to it?

 

Good question Stallen and I know I'm going to write a book here in response. This, for the most part, is a result of a very lengthy conversation I had with the Director of Marketing for Adtron yesterday who left me staggering with new SSD knowledge.

First and foremost, with respect to the defragmenting question, I have yet to have my laptop inform me that it is in need of defragmenting when click 'Defragment Now' in the tools menu of my hard drive properties. I have defragmented, however and never really understood the benefit until my conversation yesterday. The original belief way back was that defragmenting the SSD could actually damage it because of its limited write cycles and so on...this was obviously a great 'wives tale'

There are some differences that we must recognize in order to understand the end result, the first of which is that the hard drive is mechanical whereas the SSD is not. The main reason we defragment a hard drive is to speed up the access time to the information because, defragmenting over time has slowed it. With an SSD, the access time does not change and has far less defragmenting because it is not mechanical in nature, along with the way its advanced wear leveling works.

So, lets say an original access time of a hard drive is 15ms. We must understand that this, technically, is the hard drive bottleneck because that lengthy access time is required for every piece of information retrieved. In layments terms, the hard drive has no memory as to go back and get the rest of the information. With the SSD, there is no spinning and it is, well, I guess it could be compared to a library where everything and its location is perfectly indexed for retrieval. This is why there is only an average seek time of 0.1ms, vice the 15ms of the hard drive which slows significantly when digging through fragmented files.

So now, why does occasional defragmentation help the SSD even though it is not required because of the virtual elimination of fragmentation and lack of mechanical parts? With a hard drive, the installed programs (ie Vista/XP/Office) are placed permanently onto a location on the disk and remain in that spot. This is evident when one finds a powerful optimization/defragmentation program that moves your boot files to the front of the disk. It results in quicker access and quicker start times. With an SSD, due in part to the advanced wear levelling capabilities, a defragmentation will result in all files being reorganized to another place on the SSD. So...your OS, for instance, is actually being moved to a fresh area and the entire disk reorganized.

We can follow this to understand how the SSD has a longer life span than an HD now. If you keep information in the same place on a HD, not only does that information fragment over time, but also, the same 'blocks of storage' are read and written to over and over and over again. In the end, we end up with corrupt sectors on the HD.

With the SSD, all of the 'storage blocks' are utilized equally over time providing for 'wear leveling. Wear leveling is why we can predict a SSD last over a hundred years while the average life span of the HD is just over 10 years. The HD lifespan is referenced here but beware, it is a pdf. I need to give credit to SiliconSystems for convincing me to read this in its entirety; it is very long and enduring.

So now...to expand a bit. Manufacturers believe and can provide calculations that show SSDs will last over a hundred years without question. This is not a concern even being tested anymore because all can say that the SSD will outlive the lifespan of the equipment it is meant for. Thats all we really need to know. As a consumer, you will never kill a SSD through use.

The most interesting thing I learned from Adtron was that the testing that is most important to purchasers, through their business transactions, is the physical testing of the device. If you can picture a Military guy using a Toughbook in the Arctic at 40 Below Zero, you will understand. This can not occur with a hard drive. Conversely, their test methods would make us laugh for the most part, but that sells the product. Picture a paint can shaker with ten times the force shaking a laptop or someone throwing a SSD against a wall as hard as they can; this is reality. This is necessary because it has to be like that in Military fighter jets.

Ok...I thing I need a coffee. For all of you who actually read through all of this, you are well on your way to getting an SSD in your future.

EDIT: Oh and something else I found amusing. Enterprise wants to know that it can destroy SSDs when it needs to. If you can imagine sensitive military information on a SSD, they want to know that they have the ability to blast every bit of information to oblivion immediately and have it proven by the manufacturing testing of this. An example would be a fighter jet going down into enemy territory. The ability of the pilot to completely destroy the flight recorder and all data in the aircraft is a huge plus in mission critical situations such as this. (Post Note: If I am not heard of suddenly for several days after just writing this, can someone inform my wife that she might want to be concerned)

 

Interesting info, flamenko. Again, thanks for doing so much work with the SSD. I'm convinced I want to get one when the prices drop and the capacities increase

 

If the read write speeds on this are accurate, it is one heck of a deal...

http://cgi.ebay.com/64GB-Solid-Stat...ryZ42184QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

I want one!

Dave

 

The write speeds are inaccurate for this drive. I made a note of this awhile back.

Anandtech had done an article last year stating the Samsung had reported these speeds for the 64Gb SSD. The drive for sale is the same as that reviewed and benchmarks provided in the first post above. I have searched very hard to find a Samsung with similar speeds, as reported, but have been unable to do so.

Cooincidentally, I have just been contacted by a Samsung Rep yesterday who was asking about my background. Hopefully, I will be able to exchange some information with him in the near future and see if there is a Samsung drive out with these speeds.

 

oh and ask him if their Samsung is that one usedin MacBook Air

 

Yes, the Samsung 64Gb SSD is that used in the MacAir.

 

Hi Flamenko,
Any idea about the release date of the Mtron 7000 series with 64GB?

 

...not the 7000 series specifically but I will contact them and get you an answer. The 1000 series will be released in the first week or so of February in the 64Gb capacity.

 

Thanks alot,
the only place I can buy them in Europe is ssdisk.eu right?
Or do you know of any other places?

 

Yes...ssdisk.eu but I know of two who have successfully purchased directly from Mtron. I sent an email and should have an answer in a day or so.

 

One thing I have not been able to understand is how advanced the wear-levelling is.

A simple version of wear-levelling is just to wear-level new data over unoccupied memory space. However, this is quite lacking and would make small SSD drives much more prone to failure.

I can imagine more advanced versions of wear levelling that could keep track of how static different data blocks really are and act accordingly (i.e. being able to move exisiting data to another location in order to "equalize" the wear.

However, I have no idea how the wear levelling actually works. Is it simple or advanced? The advice Flamenko received saying that an SSD would benefit from a defragmentation (wear level wise) indicates that wear levelling may not be as advanced.

 

This may help:

What is wear leveling and how does it affect Solid State Disk (SSD) product endurance?

The two dominant NAND technologies are SLC (single-level cell) or binary, and MLC (multi-level cell). SLC technology stores one bit per cell, and MLC stores 2 bits. SLC NAND flash generally specified at 100,000 writes/ erase cycles per block with 1 bit of ECC (error-correction code), whereas MLC is specified at 10,000 cycles with ECC. Wear leveling allows data writes to be evenly distributed over the storage media. More precisely, wear leveling is an algorithm by which the controller in the storage device re-maps logical block addresses to different physical block addresses in the solid-state memory array. The frequency of this re-map, using the algorithm to find the "least worn" area to which to write and any data swapping capabilities are generally considered proprietary intellectual property of the controller vendor.

It is important to note that wear leveling is done in the solid-state memory controller and is independent of the host system. The host system performs its reads and writes to logical block addresses only. So as far as the host is concerned, the data does not move.

For example, if you erased and reprogrammed a block every minute, you would exceed the 10,000 cycling limit in just 7 days. Rather than cycling the same block, wear leveling allows you to distribute the number of blocks that are cycled. An 8Gb MLC device contains 4,096 independent blocks, if you provided perfect wear leveling on a 4,096-block device, you can erase and program a block every minute, every day, for 77 years.

(10,000 x 4,096)/ (60 x 24) = 40,960,000/ 1,440 = 28,444 days = 77.9 years.

It is found here.

 

Thanks for this highly informative thread.

I'm actually looking towards SSDs not just for laptops but for servers as the OS drives in data center environments.

Since data centers usually have all the big storage on external arrays this could potentially remove one of the biggest failure points on servers, the OS drives.

I wouldn't get rid of mirroring (ever) but knowing that the drive is unlikely to fail means I'd consider doing offline mirroring so that I copy out the drive nightly and take it offline after backup.

flamenko:

I'm hoping you have good info on the 1000 series in the very near future.

 

The next section in the link you provided contained the beef on the subject. Evidently there are several wear levelling algorithms in operation.

Dynamic Wear Levelling only applies wear levelling to unused memory space.

Static Wear Levelling applies wear levelling over the entire memory space, thus shuffling around existing data to other locations as needed.

Apparently Ritdata's SSD's use Static Wear Levelling, which is of course much preferred from a durability perspective. Ridata gives the example where an SSD consists of 75% static data and 25% dynamic data. All things equal the SSD employing Static Wear Levelling will have four times the durability compared to the SSD employing Dynamic Wear Levelling.

Of course, an SSD with 90% static data and Static Dynamic Wear Levelling would then be ten times as durable as a corresponding SSD with Dynamic Wear Levelling.

However, when many SSD manufacturers specify the durability in the form of "you can write X GB every day and the drive will last for Y years", they do not specifiy if how much of the data is considered to be static and how much dynamic. Granted, for an SSD employing Static Wear Levelling, the result is the same but for those SSD's employing Dynamic Wear Levelling, the durability can be significantly shorter (10 times less in the 90% static data scenario).

Flamenko, as you are the king of SSDs and have contacts with the major manufacturers, may I make a suggestion for your upcoming reviews as well as updates for your previous ones? If possible, try to find out what wear level algorithm is used by the different SSDs as those employing Dynamic Wear levelling will have significantly shorter lifespans (especially considering that 16GB and 32GB SSDs will likely be filled to the brim with static data).

 

I have put your question to one much more familiar with this than I. If you read the comparison as you were describing though, it is comparing wear leveling to 'no wear leveling' and, the difference of static to dynamic is actually closer to 4 times.

Thanks ahead...

 

Great! It will be very interesting to get more info on which manufacturer has the best/worst wear levelling.

In regards to your other comment, I was referring to static vs dynamic and not static vs no wear levelling.

I believe that the 4-to-1 ratio quoted was attributable to the assumption of 75% static and 25% dynamic data. If this is correct, then the durability ratio of static vs dynamic wear levelling should be 10-to-1 if the SSD has 90% static data, which is not an unthinkable scenario for 16 and 32 GB SSDs.

 

lol.... I know Im out of my league here. It just goes to show the vast array of knowledge floating the sites.

 

Wow ... what a great resource this thread (& site) is!

Just curious, is there a thread that tracks laptops that are available with SSD drives? I'm in the market for a new laptop & am now thinking that I want to restrict my choices to those that offer them as part of the customizing package.

 

Since all Mtron SSDs have a 0.1ms latency time, would getting a 7000 series as opposed to a 3000 series make any difference in terms of OS and application responsiveness?

 

Nope not at all....

Your sole difference will be in write times and if I can pass something on, I don't observe any difference in the two drives. I have had each in my laptops for about a month along with the Memoright in for about a month as well.

 

Flamenko:

Do you think MacBook Air with SSD could match the speed/performance of MacBook Pro with normal harddisk ? I dono what processors but could be 1.8 vs 2.4 and both 2GB RAM

 

jketzetera:

As promised I have found some information that will answer your question. I have realized a great contact in Rick Gimbel, Director of Marketting Programs for Adtron and felt he would be the best to approach as Adtron deals strictly with enterprise and has no claim to consumer sales. Adtron, I believe has the largest SLC SSD to date with a capacity of 160Gb (not to be mixed up with larger, and slower, MLC versions). In any case, here is the reply received:

Hi Les,

The static/dynamic wear leveling scenarios described are correct.

We believe that when vendors state the endurance specifications, they should indicate all of the flash management techniques that they employ. These typically included static and dynamic wear leveling.

The challenge that vendors face is non unusual. The "raw specification" of 10,000 write cycles (MLC) or 100,000 (SLC) write cycles scares people off (the "7-day" example in your blog post). Vendors are very aware of this and use extensive flash management techniques to overcome this technical fact-of-life. However, the explanation of how you get from "7-days" to "77 years" is quite involved and many people do not take the time (or really need to know) how this is accomplished.

As flash technology becomes better understood through market use, this notion that "flash management" delivers "endurance" will be taken for granted. For now, we need publications like your blog to educate those evaluating the technology.

Regards,
Rick

 

Great stuff, Flamenko. 2 questions

1) i wasn't clear if you knew if there was a difference in performance between a 1.8 SSD for an ultra-portable (say, the TZ) and the 2.5 SSD.

2) i know you said you expect some better prices aimed at consumers coming in early feb, but that's only for the 2.5 SSDs, right?

 

Thanks! You truly have access to detailed information regarding SSD's.

I have been done some thinking (yeah I sometimes do that). You have authored a number of articles, reviews and tested number of SSDs. As it stands now, your articles/tests/reviews are spread all over the forums and are not easily accesible to those who did not catch them at the time when they made the front page.

Considering that the storage subsystem is a critical component in determining the speed of a notebook, perhaps notebookreview should open a dedicated section for notebook drives. Either that or that a new section in the forum called Notebook storage (or something similar) is created and your and other relevant threads are moved there.

 

Amen. I completely agree.

 

any idea where we canget the msd 1000 64gb 2.5inch mtron?! they don't seem to have it anywhere yet.. still...

 

A minor disappointment is that the product only carries a 3 year warranty. If the product is as long lived as MTron is touting, wouldn't a 5 year warranty, the same as Seagate on their retail drives, be appropriate?

 

Can anyone of you SSD experts help me on this thread http://forum.notebookreview.com/showthread.php?t=214849 ??

Thank you