Ok I have been testing a single Intel X25-E vs a VelociRaptor 74Gb, to find out which is best.

well looks like im glad i tested it for myself as most if not all reviews and benchmarks on this SSD have been misleading which gave me high hopes. (same thing happened when i checked out a raid card and found out for myslef)

Simply put the single VelociRaptor 74Gb is as fast at loading photoshop cs3 3dSmax sony vegas etc, i was expecting huge differences, but it was a joke, if the SSD drives were much cheaper and in raid then it should be a clear winner. They even loaded windows the same, not enough diff to make me give a shi

From now on I will try to Ignore all the crappy benchmarks i see online. anyway the VelociRaptor in HD Tach got a burst of 270.3 mb/s, which would put it in the lead compared to the pics on this site.

So looks like in a few years+ they will be much better, but as it is they are so similar. A good pro of the SSD was formatting completed fast like within 1-2 minutes, funny thing i found is with the Adaptec 5805 raid card it made a normal 250GBsata format so much quicker too as if it stores info and skips it, but this is after 1 full long time usual format.

how come no reviews etc of the VelociRaptor 74GB?

Now im still looking for that amazing difference in loading again!
and sending back this rip off Intel SSD.
Go and check them out for yourself and stand up and dont keep it and fool yourself if you dont like it.

Price has gone up since this review.

Sooo Dissonance, the burning question is, what was the personal experience like?

I know it's subjective but I am curious.

(Addtional info for posts #79 and #80)
Here are my thoroughly unscientific bench tests of an SSD vs conventional drive on Linux.

64GB SSD MCCOE64G5MPP-0VA 2.5in
vs.
60GB Hitatchi HTS721060G9S 7200rpm 2.5in

The system:
Dell Latitude D620
2.0Ghz Core 2 T7200
2GB RAM
Linux Kernel 2.6.27.6 vanilla
Ext3 file system (ordered, noatime)

The test: Best of three; execute OpenOffice 3.0.0 Writer software under the four different I/O schedulers the Linux kernel offers. (reboots in between all tests)

command line: time oowriter

(lower is better)

Hitatchi 60GB 7200 rpm 8mb cache:

noop (no optimizations, aka FIFO)
real 0m7.712s
user 0m0.049s
sys 0m0.123s

deadline
real 0m7.601s
user 0m0.046s
sys 0m0.115s

anticipatory
real 0m6.725s
user 0m0.048s
sys 0m0.146s

CFQ (Completely Fair Queuing)
real 0m7.711s
user 0m0.042s
sys 0m0.137s

64GB Samsung SSD:

noop (no optimizations, aka FIFO)
real 0m2.369s
user 0m0.029s
sys 0m0.111s

deadline
real 0m2.380s
user 0m0.030s
sys 0m0.128s

anticapitory
real 0m2.444s
user 0m0.036s
sys 0m0.098s

CFQ (Completely Fair Queuing)
real 0m2.668s
user 0m0.038s
sys 0m0.112s

This data shows not only the obvious speed increase the SSD has over a conventional drive but is also shows that the anticipatory/CFQ IO schedulers can help conventional drives and may hurt SSD drives.

I'd love to buy one of these or even 2 but Intel will have to cut the price in half for it to happen.

80gb's for $800 is just too much..... I'm sure their is a market but I wouldn't ignore the enthusiast segment at the same time and with that price Intel is totally ignoring it.

in reality I'd buy 2 of these at 1 / 4 the current price if they offered 100mb/s read and 80mb/s write because the latency times are gone in going with one.... or two which is more than enough reason to buy.

but again Intel apparently has no interest in the enthusiast atm in this market.

Indeed, this may be the first enterprise-class product to bring Extreme branding into rack servers. Dude, that's totally where the X25-E's power savings will add up, as stacks of drives are combined in RAID arrays where every watt saved will also lower cooling costs for the rack.

This is the most annoying sentence I've ever read in a tech report review.

I'm not usually a critic of the editor's choice awards, but how can this even begin to be a candidate when no one will buy one cause they're so ungodly expensive and even the enterprise sector they're aimed at would be leery of such costs as well.

another quick note. WIndows 7 will detect the hard drive type and if it finds a SSD, it will change its io elevators.

I/o elevators assume that your hard drive has only one head.
SSD's don't have a head thus these optimizations, designed for mechanical hard drives are obsolete on SSD's!

I would like to see a benchmark using Windows Server 2008 vs Vista or Server 2003 vs XP with user applications.

Vista and XP both have, what kernel geeks call, disk I/O elevators. Drive elevators are i/o optimization strategies surrounding the physical head position of the drive. A good elevator will re-order I/O requests to transfer blocks near the last head position to the front of the queue, and re-order subsequent i/o's in close neighborhood chunks. This is much faster than FIFO when you have a multi millisecond head movement cost. On Vista/XP, both OS's will pause approximately 10ms between major head movements. This wait is designed to anticipate and satisfy an new i/o requests in the heads local neighborhood.

These optimization are a boon in performance for 5400 and 7200 rpm hard drives but they will hold down the performance of RAID arrays with battery backed-up cache, SAN's with large NVRAM cache and SSDs.

Windows Server 2003/2008 are optimized for SAN's and RAID arrays large nvram caches. These two OS's use i/o elevators that expect mulit-head arrays and are tilted toward FIFO scheduling with deadline, out of order algorithms.

Linux gives you four io elevators. CFQ or completely fair queuing, Anticipatory (most analogous to XP/vista's), Deadline and FIFO.

My tests have shown deadline and FIFO fly with SAN's and SSD's.

SSD for most apps, HDs for backup and storage.

I just showed my SQL servers the benches and white stuff started oozing out of the drive bays. icky.

The goo is there to secure and protect the BGA underneath the chips from oxidation, thermal expansion, and the microfractures that would result.

Lead free solder is a PITA for the entire industry.

http://www.techreport.com/articles.x/15931/5

This type of inconsistency I continue to see from SSD articles is precisely why I won't even consider one of the things.

I fail to understand how a drive which should technically be faster simply isn't.
I realise IT is a quirky thing, I've been a hardware enthusiast since my 286.
None the less based on what we're initially told and the specs for the device itself, it doesn't make sense to me, it's like Intel releasing a C2D @ 3ghz then one at 3.8ghz but the newer one is slower?

I really want to see consistency from a product when it comes to performance, if it's going to be that much more expensive for the 32gb E model, over the M model, because it's so much better, I expect it to win outright, period.

So, I shall still wait yet another year for a reasonably priced SSD to come out, which clearly and distinctly wins in benchmarks.

Apologies for raining on the party. Using any SSD as the ONLY drive in a PC without an accompanying hard-disk is potentially courting a disaster consisting of flaky operation and lost files. Exactly when that disaster happens is a time-event that is strongly specific-usage-dependent. The fuller the SSD becomes with static program and static data, the faster the disaster will happen. Consider a laptop PC with a single 80Gbyte SSD and with, say, 75Gbytes full of OS, applications and static data. Still perfectly usable. However the remaining 5 Gbytes will be occupied by dynamic data, especially the Windows virtual memory. So now the wear leveling life-time has to be computed for just this 5Gbytes, instead of the original 80Gbytes !!! And the fuller the disk with static data, the faster the remaining "freely-accessible" data-cells will wear out....Any volunteers to compute the number of cycles of Windows virtual memory in a typical day in a laptop with the minimum RAM recommended for the chosen operating system....???

And please find me the published read-lifetime of a high-density SLC flash-cell after 100,000 cycles. I doubt it is 10 years.... more likely a few weeks, or less. The Tech Report would do us all a great favor if they could extract from Intel's Reliabity Test department the graphical profile of write-cycles vs guaranteed data-retention time for the SLC flash-cells in the latest generation of the Intel SSDs. That graph seems to be a closely guarded secret.... or maybe I have been looking in the wrong places for it.... Intel's website, Google etc... Such information has been openly published in the past as reliability information on conventional flash memory -- 2-4 weeks of data-retention after 100,000 cycles comes from reliability data on conventional flash-memory of much bigger storage cell-size than current SSDs.

A SSD is an exceptionally valuable device for rapidly storing and retrieving data on a NON-CYCLIC basis...such as booting-in the OS
and storing/loading applications and their stored data. Anything involving short-term rapidly-cycled data (e.g: Windows virtual memory) must use a conventional hard-disk for storing/retrieving such data, if a SSD cell-wear catastrophe is to be averted as long as possible.. It is instructive to note that one of the latest Toshiba lap-tops that includes a SSD also mandatorily includes a hard-disk... I wonder why??

Some other differences with hard-disks:-

SSD cells will leak data while just sitting there with power applied.

Data-failure on a hard-disk is generally a one-way event with the failure easily verified by a hard-disk test-program. A marginal SSD cell can remain in a flaky data-state for a long period of time. The flaky nature can result in system-failures with symptoms that are very similar to failing RAM, motherboard or CPU. It may be difficult to pin-point the problem to the SSD... unless the wear-leveling algorithms in the SSD are supplemented by physical hardware-sensing of each cell's voltage margin above threshold and which can flag the user before the cell-data fails.

Maybe I'm missing something here. It's fast and efficient. And 32GB. So how is it $750?

I get that there's a lot of great technology in it... but really, is there so much fancy silicon on there that it is justifiably priced at more than ten times the cost of a 32GB USB flash drive?

This drive is the holy-grail for workstation and server crowd.

For everybody else, just stick with a mainstream 7200RPM HDDs. They are much cheaper and deliver almost the same performance at non-workstation/server workloads.

Man I cannot wait until these become affordable. I currently have a OCZ 60GB SSD as my OS and apps drive and it's pretty awesome but the write speeds are nothing to write home about. I would like to see these in a RAID array pitted against the Fusion-IO drive.

Too bad about the storage size and associated price tag.

My only surprise is that it didn't win in absolutely all the test (apart from CPU usage). Why didn't it boot the quickest for example? What could be the reason for this?

A request: since these drives will fit quite nicely in a notebook, it might be a good idea to include at least one standard notebook drive in future comparisons.

I expect if you bought one of those SSDs today and a bunch of hard drives of various sizes and ran them in a machine,the hard drives would be cold, dead mechanical door stoppers before the SSD would fail.

ICH7? Does this present any potential bottlenecks that we should be aware of?

Oh yeah one other thing, honest question.
Is the test rig beginning to get a little old? Maybe it's time for a new beast of sorts for testing, that one is fairly old, if I recall.

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

by using the PCI-e bus directly it blows away Sata1/2 SSDs

Obligatory:

But, can it play Crysis?

I'm curious about the idle & load ratings from Intel, and why they're so wildly different from TR's results. After all, the idle rating was off by an order of magnitude... what tests did they run to get that 0.06W number?

And not to sound like a total bumpkin, I know that company ratings are normally the very top and bottom they expect from the drive, with actual operating values somewhere in between. I just expected them to be somewhat closer.

I'd REALLY like to see this in comparison with 3.5" & 2.5" 15K RPM SAS drives. Especially given that it's as expensive, if not more expensive than SAS. SATA comparisons are neat and all, but there nowhere in the same class. Why not make a more useful comparison?

Kudos on using the 1 ohm resistor, it's a pretty easy method to measure current draw, not to mention more accurate and useful than measuring system power consumption as a whole. Was this a first in TR reviews? I can't remember. On a side note, look at that power consumption of the Hitachi drive under load... 15 watts is lot for a 7K drive!

'Extreme' branding has been making inroads in the enterprise for a while. Witness HP EXTREME storage: http://h71028.www7.hp.com/enterprise/cache/592778-0-0-225-121.html

Picked this up somewhere, maybe DailyTech -- it's part of notes I'm keeping for my new build next year--

"The street price for the 2.5" 80GB X25-M will drop from $600 to $525 on November 30.

Moving over to the single-level cell (SLC) parts, Intel will introduce its high performance 32GB X25-E and the price will drop to $575 on December 28. A 64GB part will be introduced in the first half of 2009 for $990."

that drive is going to run out of write-erase cycles much quicker than…

I see that and think it will be a cold day in hell before I jump on the SSD bandwagon.