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

Just found this via Google:

http://techlime.com/storage-media-drives-devices/intel-x25-m-ssd-reviews

Another positive review of the Intel MLC, but I'm posting it here because it begins with a great introduction for non-engineers to Flash tech and MLC vs. SLC; worth a read.

 

Haven't really been able to test it that way yet. I went to install Windows XP Pro on it yesterday, but a lack of drivers for my ASUS motherboard led to a comedy of errors that results in a non-bootable XP that thinks it's got a RAID drive, but doesn't.

Sigh.

I'll be trying again tonight.

 

Incredible read speed, 230MB/s constant w/ 0.0ms latency and write speed 75MB/s w/ 0.1ms latency. Real life performance is pretty bad though, about the same as OCZ and WD Raptor for start up/shut down/reboot/Crysis load times.

 

This is the drive for me...

http://www.dvnation.com/Fusion-IO-IODrive-SSD-Solid-State-Disk-Drive-Review.html

Any questions?

I thought not

 

Ohhh !@#*&^%...I don't believe what i'm seeing!

Edit: Okay, now seriously...i'm all for performance but regular consumers aren't ready for that kind of technology! Nor will they need that much speed...that's more suited for lunatics! And the cost....OMG!

 

Yep, got to give that DV Nation credit for being right on the ball! Finally No BS about slow writes or whatever. Just balls out PERFORMANCE. Price? Whatever they want to charge. THEY OWN THE MARKET.

Worried about price? Then I guess this is not the drive for you Me, I'll take 2, as soon as VISA raise my credit limit

 

I do. This is something I've been wondering about for a while -- how much is the SATA / SAS / IDE interface slowing down an SSD?

Way pricey though -- but imagine integrating something like this onto a motherboard.

 

I would love to be an investor in FusionIO. Cannot help but wonder what the boot speed will be

 

WoW

WoW

WoW

This is some awesome stuff...There's a ton of info. about the drive just by doing a google search. The pricing seems sketchy (or that most likely is a typo in the 2nd link w/ the GB) but if that second link is for real w/ pricing I might consider getting this thing.

 

What boot? Bios to desktop in one second

 

Well, I will believe that when I see it. Windows can slow down ANYTHING Price is 2695 on DV Nation, available now.

 

I know. That's quite pricey and I'd really prefer something for my laptop next. I'll probably end up with something for my storage server instead.

 

wow WoW WOW!! ****, that thing is awesome fast. I'd love to be an investor

 

At that price, you could just buy a ramdisk or something. Maybe even two or three and raid them.

 

well u wont find out boot speed for a while since its not bootable

and the benchmarks are inconsistent with their claimed performance. eg it says like 96,000 random write 4k iops, but it only does ~38k sequential iops in atto (and random cant be faster). still 38k is ... (words dont describe)

 

4th quarter. That could be anywhere from 10/1 to 12/31 of this year.

 

**** thats not for human beings right now... Its for lunatics... With that price I would get a car (well just a ****ty one). Really amazing .

 

Same thing I said!

 

Interesting indeed that they finally have a product actually available. Kinda sad that they feel justified in charging that much of a premium. It's a good flash controller chip, but it's no better than what Intel has - it's just using twice as many flash chips and a better interface. Granted they have a bit of a window before the SLC Intel drives are available, but once they are then they'll have to contend with raid arrays of such offering better performance and more capacity for lesser cost. Not to mention, it's getting to the point now where IO intensive applications can be done entirely in system memory for an equally reasonable cost.

 

There's something to note in that benchmark. It's interesting I think:

Fusion-IO ioDrive and Memoryright GT results:
http://www.dvnation.com/Fusion-IO-IODrive-Review-p3.html

Intel X25-M 80GB SSD:
http://www.legitreviews.com/article/788/8/

SSD64GB Patriot V2:
http://forum.notebookreview.com/attachment.php?attachmentid=24409&d=1221605816

OCZ Core Series SATA-II
http://benchmarkreviews.com/index.p...sk=view&id=200&Itemid=47&limit=1&limitstart=7

WD Velociraptor 300GB: http://www.desktopreview.com/shared/picture.asp?f=323

Look at the Atto disk benchmark results.

At 512byte, compared to 8MB for write(The percentages are compared to the 8KB results. 10% means at 512byte, it achieves 10% of the 8MB results)

Fusion-IO ioDrive: 3.11%
MemorightGT 64GB: 11.4%
Intel X25-M 80GB SSD: 16.3%
SSD64GB Patriot V2: 5.56%
OCZ Core Series SATA-II: 4.69%
Velociraptor 300GB: 10.9%

4KB results for write(important for general desktop usage):
Fusion-IO ioDrive: 22.2%
MemorightGT 64GB: 56.3%
Intel X25-M 80GB SSD: 80.3%
SSD64GB Patriot V2: 36%
OCZ Core Series SATA-II: 32.4%
WD Velociraptor 300GB: 56.9%

The Fusion-IO ioDrive degrades faster in write than all other SSDs at 512 byte and 8KB data sizes. Its worth noting its still the fastest however.

Still, considering the max write speed for the X25-M is only 70MB, its interesting it can equal the speed at 512 byte and 4KB data sizes with the MemorightGT which has a higher max write speed of 120MB.

The SLC version of the Intel drive with 170MB write might be able to equal/exceed the Fusion IO at 512 byte and 4KB sizes if the small data size write performance improves as much as it does with max write throughput.

For those with truly unlimited money, there are always the DRAM-based storage devices...

 

Hi! I'm an SSD newbie, think this thread is a great resource, and apologize if I commit any thread faux pas.

My question is general... About how many GB can be written to a 64 GB SSD before performance drops significantly (let's say to the level of an average 5x00 RPM conventional HD)? My background thinking (or lack thereof!) follows...

We often read an HD should be left 10 to 25% free to allow it to operate efficiently (allows for defragmentation etc.). The average of 10 & 25% is 18%. If I left my 80 GB HD 18% free (14 GB), I'd be able to use 66 GB of it for OS/programs/data. I've read that with an SSD, page filing can be turned off to gain about 4 GB of space for OS/programs/permanent storage. I'm going to count that as another 4 GB of space that can't be used on the conventional HD (though not being tech savvy, for all I know that 4 GB can be within the 14 GB not written to to gain overall efficiency). With those assumptions, my 80 GB HD gives me 80 GB - 14 - 4, or 62 GB that I can use for OS/programs/storage.

I've read that defragging SSDs is unnecessary. So, presumably much of the 18% usually left free for efficiency does not need to be left free. And, as noted before, apparently page filing can be turned off with SSDs thus gaining 4 GB of space that could not otherwise be used for OS/programs/permanent storage.

So, it seems to me that I can fill the 64 GB SSD almost completely up with OS/programs/data with no performance decline such that the 64 GB SSD actually has more OS/programs/data storage space on it than my 80 GB HD which, can only fill up to about 62 GB before performance drops. Is that roughly correct?

Les has written "Further, I noticed that the graph line in the SSD chart was steady across from beginning to end, whereas the hdd graph line eventually faded from an average of 40mb/sec to a low at the end of 23 mb/sec. It was a steady decline in the HDD." Not being very familiar with HD testing, I'm interpreting this to mean the HD slowed as it filled up (rate drops as capacity is neared) but the SSD did not. If this interpretation is correct (but maybe the x-axis wasn't even how many GB had been written - I did not see the graph, just the quote), it would bolster the idea above that a 64 GB SSD offers effectively the same storage as an 80 GB conventional HD. Thank you for your thoughts.

 

But an Intel X25-M with 30% drop in PC Mark Vantage is still very fast. Faster than a WD Velociraptor.

Article is here: http://www.legitreviews.com/article/788/11/

 

Gordy....

Its nice to know I am still read lol. My interpretation of the downward curve of the hd is not so much the filing of the hd but rather, its decline due to the stress caused by the larger files being read. All hds seem great initially but there is always the decline, whereas the ssd does not show that.

Can this be strictly because there is no high speed disk or arm moving to, not only find, but also, retrieve the data??? Not sure as I still would not claim an more of the status of a laymen who enjoys dabbling in ssds.

Next... I am curious of your impression of the ssd in a 64bit Vista environment? I am going to bet that you have probably tweaked somewhat so we are pretty much privy to the same experience. Personally, I have never been so...content...with a laptop as I am now. No crashes, no freezing, stuttering, loss of info and no slowing of the drive as is normal with a hd that fragments...its a very comfortable place to be I think.

 

SSDs also show a decline in performance as capacity fills. There are articles online about it. It's not just the Intel drive but others as well. The (few)sites I have read about that indicate its because of the write levelling algorithm.

The thing about the performance decline is that since SSDs have uniform performance throughout the drive, once the drive fills up, the entire drive's performance degrades with it.

Yes, you do not need to ever use defrag. The reason is because throughout the ENTIRE drive the performance is the same. In fact, they recommend you to NOT use defragmentation because you can lower the lifespan of the drive.

I'm not sure if you would want to disable "page file". What you say sounds like the amount of space used for write levelling algorithms, in attempt to increase the lifespan of the SSD.

 

I got my Patriot Warp 64GB SSD today. Since I also got a factory image on my replacement hard disk, I just imaged that over and ran some tests. HDTune and Vista performance index are attached.

Boot times on a factory direct Dell Vista Business 32-bit:
* Toshiba 7200RPM: 62 Seconds
* Patriot SSD: 42 Seconds

I'm pretty impressed so far - havent had any hiccups. As I get my OS back to the way it was, I'll update to see if any issues arise.

 

Disabling pagefile is strictly a performance increase but it will get you 4Gb back. I would recommend it only to those with 4Gb ram as I have my system speeding along void of pagefile and have yet to experience a prob in over a year....and I beat the heck out of my system trust me.

Just to explain, even if you reduce your page file and have 4Gb, your system still utilizes the pagefile giving it priority over your precious 4Gb ram. If you monitor your ram through a sidebar gadget you will see. Shutting it down completely forces the use of your 'much speedier' ram and, well... I definitely notice a visible difference in laptop system activity performance.

For me... ssd, 4Gb ram, clean install, tweaking and shutting down pagefile has my system running very fast and VERY stable.

Okok naysayers hit me eheheheh.

 

for a HDD, the graph slows down towards the end because the diameter of the disk on the inside is less than on the outside. on the outside, for a given revolution of the disk, more data passes under the head, it can therefore read more data in a given time period.

while a HDD slows down when it fills up due to less chance to find continuous blocks to write to (and the end result when it cant is fragmentation), most likely a SSD slows down because it is looking for blocks that havent been written to too much already. whenever it has to write to blocks that arent physically continuous, the performance worsens. however, im not sure of why (or if) the read speed would slow down if no writes are being performed.

ive been with a pagefile turned off ever since i got 2GB ram a few years ago (i keep an eye on ram usage), its probably been one of the bigger differences thats happened, going from hdd churning every alt-tab to being instant.

 

Yeah I'm really considering the Patriot v2, but there doesn't seem to be a level of consistency that I'm happy with amongst users. Some are reporting the same stutters (makes sense since it's still using the Jmicron controller) while others don't.

I sure don't want to shell out a few hundred to go backwards in overall happiness.

 

Well, I'm going to try out my 32GB Patriot V.2 with Vista tomorrow likely - can check to see if it's just something odd with XP that's causing the issues I'm seeing. Yes, it still has dead pathetic random write IOPS, but I haven't really seen any odd pauses in either OSX or an Ubuntu install, definitely not anything like what I can make it do in XP. Either way, so long as the drive isn't having to do any writes really, then it performs great, even under XP.

 

I want to ask urs the Patriot V.2 is worth to buy or not
thx

 

Yes please let us know if you experience any hickups/stutters.

 

i'm using a super talent mx 120gb ssd.. and it works really well.. no stutters or anything.. but startup is slower (0.5ms seek time? :x) -compared to the mtron 32gb ...

 

HELP!

I have been trying to wrap my head around the performance of SSDs but have not been able to resolve a number of contradictions. I simply cannot reconcile the performance numbers churned out by the benchmarks in the reviews with the actual real world effect (in a workstation usage scenario with Windows XP/Vista) of SSDs.

For example, in the case of the OCZ Core series (and other MLC-drives using the Jmicron controller) we learned that fantastic ATTO/HD Tune/HD Tach graphs can have little relevance to real-world performance, as evidenced by the OCZ Core series dismal random write performance, which results in OS “stuttering” and system unresponsiveness.

Ok, so in the case of the OCZ Core-series (and other MLC-drives using the Jmicron controller) the disconnect between the expected and actual performance is mainly attributable to the huge performance difference between sequential and random write performance. So I think I do understand the reason why those drives perform very well in most benchmarks but not as great in real-world Windows workstation usage.

So, suddenly Intel steps into the market and launches an MLC-based SSD with specifications that completely blows away all other current SLC and MLC competitors. The Intel X25-M has 2x – 3x faster sequential read and write speed, and something like 10x – 20x faster random read and write speed at small data blocks.

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3376&p=2

In theory, the Intel drive should leave the competition lying in the dust and utterly crush everything performance-wise. So, along comes Anand and performs an in-depth review of the Intel drive and compares it to a number of competitors. The PCmark Vantage benchmarks put the Intel clearly on top by a wide margin in most tests. However, the lead over e.g. the OCZ / Samsung SLC drive in the PCmark Vantage benchmarks is more in the range of 20% - 50% rather than the 2x -3x (or 10x – 20x) advantage in the drive specs.

However, towards the end of the review, Anand starts to do more “real-world” oriented tests. He measures application start-up time, average fps running time-demos, time to load a game level, time to extract a large archive while starting an application etc.

Suddenly, the performance difference between e.g. the OCZ / Samsung SLC-drive and the Intel drive are significantly reduced and in some instances the roles are reversed.

According to the drive specs and the synthetic benchmarks the Intel drive should significantly outperform the OCZ/Samsung SLC-drive by virtue of having 2x – 3x higher sequential read and write performance and being capable of 100x higher random write IOPS at 4kb data size!!

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3403&p=8

What gives?

In the “real-world” tests the difference between the OCZ / Samsung SLC drive and the Intel drive are minimal. Even more confusing is how the OCZ / Samsung SLC drive manages to beat the Intel drive significantly in the application-start/archive-extraction combo test. To my understanding, the Intel drive should completely blow away the OCZ / Samsung SLC drive, based on its far superior specs.

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3403&p=15

It would seem that for Windows workstation usage, synthetic benchmarks simply fail to be a reliable indicator of actual performance. I also do not understand how the Intel drive with such incredible specs (and incredible benchmark results) does not perfom significantly better when it come to the more “real-world” oriented tests (application load time, average fps during time demo in Crysis, time to extract/create large compressed archives etc…

I am confused (as usual )

 

the intel can't deliver more performance at any point where disk is not bottleneck. in a game environment like in crysis, disk should NEVER be bottleneck for fps. it just means it loads the textures faster into ram, so you don't have that blurry pictures while the textures are not loaded. but it should never harm fps as the fps are independent of disk-accesses.

similar for other cases. and an slc is always faster than an mlc just by itself. so while the intel is the way fastest mlc around, the slc can sometimes win just because it is a different basic technology.

 

Yes, but for some reason the Crysis benchmark is touted by a number of sources (that is why I referenced it).

Yes, the Intel is an MLC and MLC is slower than SLC. However, if Intel specs says that their drive is 2x - 3x and 10x - 100x times faster in sequential and random reads and writes than e.g. the Samsung, I cannot understand how it can lose in any test and how it can lose by a such large margin in the combo test.

 

the only explanation i can give for the intel ssd's not being as good in all areas, is when read/write are mixed. i think that when you're purely doing only random writes, then the controller does a good job at the random->linear remapping (i.e. writing files linearly which is where MLC is ok, and then remapping the subsequent reads to wherever the files ended up), but if this is interrupted by reads, then performance is affected.

e.g. one benchmark was filecopy, which presumably is a lot of reads and writes together, and the drive performance was fairly low, despite it having both better individual read and writes than the drives it was level/below in the copy test.

can IOmeter do interleaved read/writes? ie not just 50% distribution between random read/random write at say 4k, but actually do write then read then write then read, etc? if so this may confirm this hypothesis.

 

That makes sense. If the Intel drives are showing a similar tech to what you describe, it might be common to ALL the future SSD drives.

There are 2 common things reported for Intel drives:
-Boot times are not too fast
-Despite that, it does not show that for application launch times that its slow

I can see some other reason for it though. Intel has said that while its the fast read/write speeds that makes SSDs boot fast and launch applications fast, its really the latency of the drive that affects such factors. But look at this: http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3403&p=2

So because the SLC technology is inherently faster than MLC, the fundamental foundations can't be changed. Latency of the SLC tech by nature is lower than MLC drive. While Intel did much to bridge the gap, it can't escape the origin, and which is why its not dominating in everything. To mitigate being slower MLC tech, they had to implement something that won't sacrifice write performance.

Intel SSD drives are also said to have an algorithm in the controller that over time, it'll try to find out usage patterns and optimize the performance. It has been said in some reviews when the usage patterns are GREATLY changed, the performance of the drive drops before it goes back up.

Eventually, we still have one conclusion: You get what you pay for. You DO need to pay much higher costs to get the response time you get with an SSD drive(like OCZ Core), and even more to get an SSD without any quirks(X25-M).

For ultimate performance, some with much higher budgets may want to look for the X25-E which is SLC technology(2x cost per GB over MLC device so we are looking at $500 for 32GB version here).

True. And even with "real world" benchmarks that does pure application launch times and file transfer times, it doesn't show that the bad JMicron controller based MLC drives suffer stuttering either.

 

The weird thing is the Intel drive beats the other SSD drives, even the Samsung in the following test.

File copy/zip-unzip test

http://www.laptopmag.com/review/storage/intel-x25-m.aspx?page=4

Slowest boot time:

http://www.laptopmag.com/review/storage/intel-x25-m.aspx?page=5

...yet fastest in hibernate/wake test

 

I tell ya, I am almost beginning to think that these SSD's are going to be as much of a pain as spinning drives, and maybe faster maybe not. I always forget to factor in the INCREDIBLE hype like printer "pages per minute" although I must admit that printers seem to have gotten a tad better.

I mean you cannot even get a darn Intel drive and there is already so much conflcting info I can hardly stand it! And I have a hard time believeing that an SSD cannot BOOT faster than a regular drive.

I mean if we have come all the way around to a darn $500 drive for only 32GB as being the only drive worth getting, this is precisely where we were a year ago with an Mtron 3000 although they were $649, so I guess there has been maybe a little progress, sort of.

Sunday night ramblings from a very frustrated potential SSD user. Dave

 

I really want the X25-M so I called NCIX(Canada). They said the ETA is October 10th.

 

To All,

Thanks for the replies commenting on how full an SSD can get before performance drops. As PhilFlow pointed out, a 30% reduction in the Intel X25-M's performance due to its being 80% full (Phil pointed us to http://www.legitreviews.com/article/788/11/ for this info) means it's still fast, just not as fast as it was.

So, at what "fullness" does it degrade to a point it's not acceptable anymore (let's use 4200 RPM HDs as the lowest performance we'd accept just for sake of argument)? Does it sound safe to ASSUME filling up an SSD to 90% will degrade its performance, but that it will still be faster than a half-filled and optimally performing 4200 RPM conventional HD? Has anyone filled their SSD up to 90 or 95% or have other links to data on performance vs. how full the HD is? I think this is a worthwhile inquiry as we're all going to be filling these things up and it would be nice to know how far we can go.

Les, I should have been more clear. I'm such an SSD newbie that I don't even have one yet! So, no tweaking yet.

 

That's true with all SSDs implementing write-levelling algorithm. Which will be all of them in the future.

 

i cant actually find the file copy benchmark where the intel was slow (i remember it was around the 30MB/s mark and probably one of the pcmark/vantage or similar mum and dad benchmarks).

 

Could it be that because of the lower write endurance of MLC NAND chips, the wear-level algorithms of MLC-drives have to be more aggressive than those for SLC that-drives (if they are both to have the same lifespan)?

Also, Intel clearly states (brags) that its wear-level algorithms use all memory cells almost equal (4% dispersion), while some competitors have 3x maximum wear, compared to average wear.

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3403&p=3

This could imply that once the Intel drive gets full, the drive will constantly shuffle around the data on already written cells, as you are writing new data to the free space on the drive, in order to ensure equal wear of all cells. This of course would result in reduced performance, as the SATA-interface IOs are sharing the drives bandwidth with the internal wear-level algorithm in action.

So the hypotheses are:

1. MLC-based drives might have wear-level algorithms that are more aggressive than SLC-based drives. More aggressive wear-level algorithms should have a negative impact on performance.

2. The Intel SSD is very good at utilizing all cells almost equally, compared to some other SSD-drives. This increases the life-time of the drive but could do so at a bigger performance penalty, compared to SSDs that are not re-shuffling blocks as frequently and thus not achieving as equal wear of the cells.

 

Actually, even the SLC devices had the similar performance loss.

And not just that, the X25-M optimizes the drive depending on the usage pattern and the performance loss was supposed to be rather significant for the time being until the drive gets used to the pattern again.

The thing is that not all sites show all low results or all high results. Some sites show low file copy results and some show high results.

 

Very interesting. Do you have a link you can share?

 

While I haven't found that link yet, I have been able to find the reasons for variety for the X25-M drive in performance:
http://www.hardware.fr/art/imprimer/731/

SSDs all have what is known as an “Indirection System” – aka an LBA allocation table (similar to an OS file allocation table). LBAs are not typically stored in the same physical location each time they are written. If you write LBA 0, it may go to physical location 0, but if you write it again later, it may go to physical location 50, or 8.567 million, or wherever. Because of this, all SSDs performance will vary over time and settle to some steady state value. Our SSD dynamically adjusts to the incoming workload to get the optimum performance for the workload. This takes time. Other lower performing SSDs take less time as they have less complicated systems. HDDs take no time at all because their systems are fixed logical to physical systems, so their performance is immediately deterministic for any workload IOMeter throws at them.

The Intel ® Performance MLC SSD is architected to provide the optimal user experience for client PC applications, however, the performance SSD will adapt and optimize the SSD’s data location tables to obtain the best performance for any specific workload. This is done to provide the ultimate in a user experience, however provides occasional challenges in obtaining consistent benchmark testing results when changing from one specific benchmark to another, or in benchmark tests not running with sufficient time to allow stabilization. If any benchmark is run for sufficient time, the benchmark scores will eventually approach a steady state value, however, the time to reach such a steady state is heavily dependant on the previous usage case. Specifically, highly random heavy write workloads or periodic hot spot heavy write workloads (which appear random to the SSD) will condition the SSD into a state which is uncharacteristic of a client PC usage, and require longer usages in characteristic workloads before adapting to provide the expected performance.

When following a benchmark test or IOMeter workload that has put the drive into this state which is uncharacteristic of client usage, it will take significant usage time under the new workload conditions for the drive to adapt to the new workload, and therefore provide inconsistent (and likely low) benchmark results for that and possibly subsequent tests, and can occasionally cause extremely long latencies. The old HDD concept of defragmentation applies but in new ways. Standard windows defragmentation tools will not work.

SSD devices are not aware of the files written within, but are rather only aware of the Logical Block Addresses (LBAs) which contain valid data. Once data is written to a Logical Block Address (LBA), the SSD must now treat that data as valid user content and never throw it away, even after the host “deletes” the associated file. Today, there is no ATA protocol available to tell the SSDs that the LBAs from deleted files are no longer valid data. This fact, coupled with highly random write testing, leaves the drive in an extremely fragmented state which is optimized to provide the best performance possible for that random workload. Unfortunately, this state will not immediately result in characteristic user performance in client benchmarks such as PCMark Vantage, etc. without significant usage (writing) in typical client applications allowing the drive to adapt (defragment) back to a typical client usage condition.

In order to reset the state of the drive to a known state that will quickly adapt to new workloads for best performance, the SSD’s unused content needs to be defragmented. There are two methods which can accomplish this task.

One method is to use IOMeter to sequentially write content to the entire drive. This can be done by configuring IOMeter to perform a 1 second long sequential read test on the SSD drive with a blank NTFS partition installed on it. In this case, IOMeter will “Prepare” the drive for the read test by first filling all of the available space sequentially with an IOBW.tst file, before running the 1 second long read test. This is the most “user-like” method to accomplish the defragmentation process, as it fills all SSD LBAs with “valid user data” and causes the drive to quickly adapt for a typical client user workload.

An alternative method (faster) is to use a tool to perform a SECURE ERASE command on the drive. This command will release all of the user LBA locations internally in the drive and result in all of the NAND locations being reset to an erased state. This is equivalent to resetting the drive to the factory shipped condition, and will provide the optimum performance.

Soooo....

defragmentation is back. But not in the way we know it.

EDIT: Ohhh, I get it now. So the performance degradation has to do with the optimization algorithm Intel has on their controller. And as you use it, the performance will go up again, considering the user has similar usage patterns.

Here's the link, I misunderstood it however. All SSD devices must have optimization going on depending on usage pattern.
http://www.hwupgrade.it/articoli/storage/2044/intel-x-25m-ssd-in-test-con-molte-sorprese_10.html

 

So these are the reasons for such varying benchmarks with review sites all over the web.

1. The SSD is trying to optimize the performance, but according to some sites/Intel, the benchmark suits are random enough that it takes longer to get used to them.
2. Usually as a reviewer they might run the synthetic benchmarks like PCMark/IOmeter before running loading benchmark using applications. Because of that, the drive is attempting to optimize the performance and taking a performance hit until its back up to par.

That could be the reason that different sites show different loading time results.

 

SAMSUNG 128 GB SSD - COMES WITH DELL XPS 1530M BENCHMARK

I will post the csv files later once im finished. I was just so curious as to what the performance of the drive would be like so its the first thing i tested out of the box. (all bloatware and whatnot is still on the drive)
Conclusion: Samsung doesnt dissappoint, If you looking for an SSD at decent cost and size this is it.

from iometer:
Ran for 4 minutes (performance slowly decreases and results saturate ~3minutes with increasing time)

Total I/O per sec = 709.57
Total MBs per sec = 2.77
Average I/O response time = 1.4078 ms
Maximum I/O response time = 25.6365
%CPU Utilization (total) = 4.11%
Total Error Count = 0

wow not bad at all for a ~$400 128GB SSD

The transfer rates from Atto and whatnot i didnt bother to run
they are most likely ~published specs 90MB/s Read and 70MB/s write.

All i can say is i think the drive everyone here will want is not the Intel SSD
(unless the 160GB is miraculously cheap). But the 256GB Samsung SSD coming soon. It should have atleast similar I/O performance which is decent and ~200MB/s Read and 120MB/s Write (more throughtput than Intel XM-25)
The cost will be the deciding factor, but if 128GB is a 450$ upgrade... 256 probably will be launched around $675?

 

Dseo...big question is...is there any pausing at all? I am sure you will keep us up to date if that be the case.