OCZ SSD sales increase 265%
- Thread starter Grall
- Start date
the next gen sandforce drives will be faster. Seems like the g3 is more of a sideways upgrade than a foward one.
Anand has a piece on each of them , but if you want to max out sata 6.0 it looks like its a sandforce drive.
My 120 gig vertex 2 just came in from newegg. $189.99 after MIR isn't bad to me. I paid that for my 60 gig june 2009.
So I'm going to keep the vertex 1 as the boot drive and put my games on the vertex 2. Hopefully when i go bulldozer in teh summer a sata 6.0 90 to 120 gig drive is under $150 and I will make that the new boot drive.
aaronspink
Veteran
Couldn't be any slower in the real world. SF drives so far are benchmark specials that break down after they roll out of the parking lot.
Doubtful outside of zeros.
I dunno. My vertex 2 is very very quick and isn't causing trouble for me at all so far. My 60 gig vertex performs better than my friend who has a second gen intel ssd (80 gig) also.
http://www.anandtech.com/show/3971/...troller-sf2000-capable-of-500mbs-and-60k-iops
http://www.anandtech.com/show/3965/intels-3rd-generation-x25m-ssd-specs-revealed
The current sandforce drives do quite well with sata 6.0
http://www.anandtech.com/show/3812/the-ssd-diaries-crucials-realssd-c300/2
They lead intel in almost every test out there.
I think that "Intel Inside" logo on your business card might be tainting the color of your glasses
In all seriousness, my OCZ Vertex 2 Extended 120Gb drive has been stupidly fast since four months ago when I bought it. That's all the patches and whatnot on a Win7 Home Premium x64 platform, six Steam games, a multitude of VMWare 7 VM's being created and deleted, several ripped DVD's that were later moved to an SD card (ie, deleted from the HDD), same thing with a handful of CD's to MP3's, and four months of Outlook 2010 email linked to my corporate account and associated office doc creation and modification.
Cold-booting to a usable OS desktop is still under 25 seconds, opening Outlook is still around two seconds with my huge PST and OST files, opening a 30-page powerpoint presentation with a big pile of graphics and animations is still a single second or less, etc. For all intents and purposes, it is every bit as fast today as the day I installed it.
So if this is a "benchmark only" drive, then it certainly has me fooled
Kyyla
Veteran
I'm waiting for those too. I'm hoping we'll see 160GB for 200$.
I'm not sure we will see those prices. I'm expecting 160GB from intel at around $250 or more for G3 drives.
Next gen sand force will most likely have the 120 gig drives down to $150 and 240 gig drives at the $350 mark.
I wonder if we will see G2 intel drives with smaller nand hit thier lower price tiers.
aaronspink
Veteran
I think that "Intel Inside" logo on your business card might be tainting the color of your glasses
No not really. The SF controller is known to have a lot of issues.
http://www.xtremesystems.org/forums/showpost.php?p=4581793&postcount=11
http://www.bit-tech.net/hardware/storage/2010/06/25/sandforce-ssd-test/9
As soon as you do actual file copy tests or use real data the performance degrades. In addition, once the drive has been in use for a while, performance also degrades. This is a known issue with the way that the SF controllers support TRIM and GC.
One of the more humorous things with SF controller is that simply reading the same file over and over results in degraded performance!
what uncompressable data am I going to put on the drive ? I use mine for games and other programs. Not video and pictures. I have a cheap western digital caviar black drive thats 2TB for that
aaronspink
Veteran
The compression they are using seems to work for very simple to compress data and not much else. So if it isn't basically all zeros or ones it appears to degrade heavily.
That's interesting. Am I reading those graphs right? The original vertex performs better in the long run than the other SSDs?
edit: for uncompressed sequential writes anyway... Although it seems the differences between the other SSDs and the vertex are miniscule enough to not be that noticeable in practice.
edit: for uncompressed sequential writes anyway... Although it seems the differences between the other SSDs and the vertex are miniscule enough to not be that noticeable in practice.
Last edited by a moderator:
I see three 'issues', first is massive file copies resulting in slower performance: Your first link says that if I absolutely hammer the drive (what would be considered 'seven days worth of writes over the span of a few hours' that the firmware will purposefully degrade write performance as a protective measure. That write performance will eventually be allowed to come back. Since I have no reason to do this, and neither does any other "standard" user, then I find it a non-issue. But let's think bigger: this is exactly contrary to what you tried to tell us earlier about these drives being "benchmark only" that fail when they 'leave the parking lot.' Really, if this drive was going to be a stellar benchmark horse, this intentional degradation should make it look FAR worse than it really is. And I don't see that happening... So you need to decide which of your issues are really true: is this a benchmark winner and yet a loser in real life? Or is it a benchmark loser (because of this flaw) but still a winner in real life? It's looking a bit like neither to me...
The second issue: Uncompressable file performance. As the above, I fail to see how this would apply to me. ZIP files and MP3's and my DVD rips are bandwidth-unintensive or archive data, and thus are NOT stored on my SSD drive. Stuff that IS stored on my SSD drive would be things like Windows DLL's and binaries and other such applications -- things that, I have found, are very easily compressed. Further, they share a lot of 'common' data elements, and so write amplification can be considerably reduced at a block level (ie, subcomponents of individual files, similar to Microsoft's ImageX compression methods.)
Finally, TRIM performance degredation. Sure, it was an issue on older firmware, just as it was for the original OCZ Vertex drives with 1.3 firmware. When they upgraded to 1.5 firmware, all that issue went away (which was also covered in your link.) Same thing happened for OCZ Vertex 2 drives
So, I still see Intel's offerings as sub-standard to the SF-1200 and SF-1500 offerings. That's my opinion, and I really don't see anything you've presented that has much chance of changing my mind. Got anything else?
Last edited by a moderator:
I'm on my third refresh of SSDs...
I started out with :
60GB Vertex - Windows
240GB Vertex - Data drive
then :
80GB x25-M G2 - Windows
160GB x25-M G2 - Data drive
now :
60GB Verted 2E - Windows
240GB Vertex 2E Bigfoot - Data drive
In my day-to-day usage I have found performance wise that :
Vertex < X25M-G2 < Vertex 2E
One key thing to remember about Sandforce drives, is they drop in sequental write performance (write throttling) on purpose as they value NAND longevity over maximum speed. Sandforce controllers seem to have very intellegent data placement, static data rotation and TRIM/GC to keep the lifetime as long as possible.
I started out with :
60GB Vertex - Windows
240GB Vertex - Data drive
then :
80GB x25-M G2 - Windows
160GB x25-M G2 - Data drive
now :
60GB Verted 2E - Windows
240GB Vertex 2E Bigfoot - Data drive
In my day-to-day usage I have found performance wise that :
Vertex < X25M-G2 < Vertex 2E
One key thing to remember about Sandforce drives, is they drop in sequental write performance (write throttling) on purpose as they value NAND longevity over maximum speed. Sandforce controllers seem to have very intellegent data placement, static data rotation and TRIM/GC to keep the lifetime as long as possible.
in what period of time did you go through 3 sets of drives, and was that because of performance degradation or NAND longevity?
that would suggest writing speed rather than volume damages the drive
that would suggest writing speed rather than volume damages the drive
aaronspink
Veteran
And yet neither Sandforce nor any of their partners will release actual endurance data nor endurance/usage guidelines/specifications. So I'm willing to bet it is primarily hot air.
I've pretty much gotten to a point that unless an SSD vendor is willing to publish data then they are basically full of BS.
In contrast, Intel and a couple other vendors have extensive data published with full details and assumptions of testing along with both endurance/usage guidelines and specifications.
aaronspink
Veteran
Cycles is all that matters. The issue with the SF drives is their algorithms, etc, start to break as the drive has to deal with fragmentation and internal data movement.
Actually, I had some bad luck with motherboards dying on me after only a year or so and each time I rebuilt my PC I decided to re-pourpose the SSDs into other laptops/desktops.
The Vertex drives have about 80% of they life left, and the X-25Ms and Vertex 2's are still on 99% on their respective media wareout SMART values.
The more aggresive background Garbage Collection and lack of TRIM (initally at least) has helped to use up many more NAND cycles on the Bearfoot based Vertex drives.
