A new process node doesn't automatically mean cheaper prices. While you get more GBs out of each wafer, the cost of all the R&D for the new process node is huge. Moreover, 24nm process is now a mature process and available in high supply, whereas the yields of 19nm process are poorer and supply is limited.
As I've said before, it seems that SSD/NAND prices drop linearly, i.e. a new process node doesn't introduce a significant price cut.
It's a new process and yield is much lower than 24nm. Did you expect them to sell it to you at a huge loss? And it's a new controller, why would a new controller that hasn't been sold in volume be cheaper anyway?
- The Vertex 4 with new firmware has been the best consumer drive on the market for database-type sustained mixed random I/O workloads. - The M3 Pro was much better than its peers at this same type of workload.
If this uses the new controller that the Vertex 4 uses but continues Plextor's success at optimizing for industrial use cases this could be the drive of choice for serious database use.
Hardware makers know most enthusiasts who spring for a high-priced SSD or other hardware product is unlikely to keep it for five years if technology makes it pretty much obsolete or less cool than the trick-of-the week - for some filks. Thus you can't use a 5 year warranty to conclude that one SSD is any more reliable than another. The extended warranty is usually just a marketing game to justify the price increase. A hundred year warranty doesn't prevent the loss of valuable data however...
As noted I'd wait 6-12 months on any new SSD to see how it works out for reliability and compatibility unless you kile being a guinea pig. By then it should have had 3-10 firmware updates, a few recalls, a lot of RMAs and almost be ready for Prime Time. It's amazing that hardware makers just crank half-baked goods out the door weekly with little concern for proper validation.
As Anand has said himself: "SSDs are immature tech". Each new iteration contributes to the learning curve for the manufacturer but does not necessarily increase reliability nor compatibility which can only come from proper validation testing.
@Kristian, being as you have a review sample on the way do you have a rough idea when you'll have the review up? I'd like to install Win 8 on some new SSD's around mid August...
240GB SanDisk Extreme (SF-2281 with Toggle NAND) review is coming up.
As for the M5 Pro review, FedEx has not delivered my review sample yet so there will be a delay. It was scheduled for Thursday delivery, but the tracking page has not been updated since Wednesday (last update was departed from Paris). I'll call them on Monday unless something happens during the weekend.
There's another SSD review going live this weekend, though :-)
BTW, anyone who thinks SSD prices are too expensive should check them out again. There has been a dramatic price drop even in ridiculously overpriced Australia. In about the last 2.5 months local prices for a Samsung 830 128GB SSD have dropped from ~$170 to $115. And of course Americans are even better off.
Samsung is doing a great job of getting their drive down in pricing so that it's a no brainer at times. Their rep is actually rather good all thing considered with regards to SSD's. Certainly, not the OCZ "down in the toilet" low.
And when the performance is solid, they don't seem to have the problems to flash out that other brands always tend to have, they offer solid tools to monitor their drives, and THEN they start offering stellar pricing...
It just makes it easy to recommend Samsung and never think about it again. Until they release a new drive, that is.
Totally agree with you too. I see Samsung 830 128GB SSD going for as low as $80 to $90 from very reputable e-tailers, big recommendation to all my friends. 128GB are the best price/performance ratio, and it has plenty space to put all software on it.
They will need to add logic to detect the power cut, to reroute power input from the capacitor, to write the DRAM content to SRAM quickly and efficiently. It is adding yet another possible point of failure that needs thorough testing and validation, and I suspect even adds some degree of legal responsibility to them. For normal drives they clearly state data loss is possible and you shouldn't complain if it happens, but for drives that claim to implement data loss prevention failure to meet that claim in even a small fraction of cases could be a serious problem.
They are not enough. What if a power supply fails, rather than utility power? What if a UPS fails? The data in the DRAM cache that has not yet been written to Flash is lost, and that is quite likely to corrupt database files, etc. This is why caching RAID controllers generally have on-board battery power.
It seems that some of the increased performance in recent months on SandForce based SSDs may be at the expense of the TRIM feature - which can result in a big loss of performance over time.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
23 Comments
Back to Article
gevorg - Tuesday, August 7, 2012 - link
Nice to see a 5 year warranty, but I will still wait at least 6 months before considering any newly released SSD.Shouldn't the prices be lower than M3 Pro, since there was a 24nm=>19nm process shrink?
Chaser - Tuesday, August 7, 2012 - link
Thanks for sharing with all of us your purchase consideration policies.Anyway, sounds like a great drive. It's good to see Plextor jump into the SSD market so aggressively. A very trusted name for sure.
gevorg - Tuesday, August 7, 2012 - link
Thanks for being a guinea pig for us!Kristian Vättö - Tuesday, August 7, 2012 - link
A new process node doesn't automatically mean cheaper prices. While you get more GBs out of each wafer, the cost of all the R&D for the new process node is huge. Moreover, 24nm process is now a mature process and available in high supply, whereas the yields of 19nm process are poorer and supply is limited.As I've said before, it seems that SSD/NAND prices drop linearly, i.e. a new process node doesn't introduce a significant price cut.
chrnochime - Tuesday, August 7, 2012 - link
It's a new process and yield is much lower than 24nm. Did you expect them to sell it to you at a huge loss? And it's a new controller, why would a new controller that hasn't been sold in volume be cheaper anyway?dave_rosenthal - Tuesday, August 7, 2012 - link
- The Vertex 4 with new firmware has been the best consumer drive on the market for database-type sustained mixed random I/O workloads.- The M3 Pro was much better than its peers at this same type of workload.
If this uses the new controller that the Vertex 4 uses but continues Plextor's success at optimizing for industrial use cases this could be the drive of choice for serious database use.
RaistlinZ - Tuesday, August 7, 2012 - link
I look forward to how this performs. Might finally be time to upgrade my Vertex 2.Beenthere - Tuesday, August 7, 2012 - link
Hardware makers know most enthusiasts who spring for a high-priced SSD or other hardware product is unlikely to keep it for five years if technology makes it pretty much obsolete or less cool than the trick-of-the week - for some filks. Thus you can't use a 5 year warranty to conclude that one SSD is any more reliable than another. The extended warranty is usually just a marketing game to justify the price increase. A hundred year warranty doesn't prevent the loss of valuable data however...As noted I'd wait 6-12 months on any new SSD to see how it works out for reliability and compatibility unless you kile being a guinea pig. By then it should have had 3-10 firmware updates, a few recalls, a lot of RMAs and almost be ready for Prime Time. It's amazing that hardware makers just crank half-baked goods out the door weekly with little concern for proper validation.
As Anand has said himself: "SSDs are immature tech". Each new iteration contributes to the learning curve for the manufacturer but does not necessarily increase reliability nor compatibility which can only come from proper validation testing.
B3an - Tuesday, August 7, 2012 - link
@Kristian, being as you have a review sample on the way do you have a rough idea when you'll have the review up? I'd like to install Win 8 on some new SSD's around mid August...Kristian Vättö - Tuesday, August 7, 2012 - link
Should be up this weekend unless something unexpected comes up.Daggarhawk - Friday, August 10, 2012 - link
Is there a review in the works for the SanDisk SSDs? seems like a major gaping hole in the bench database... thanks for all your excellent work.Kristian Vättö - Saturday, August 11, 2012 - link
240GB SanDisk Extreme (SF-2281 with Toggle NAND) review is coming up.As for the M5 Pro review, FedEx has not delivered my review sample yet so there will be a delay. It was scheduled for Thursday delivery, but the tracking page has not been updated since Wednesday (last update was departed from Paris). I'll call them on Monday unless something happens during the weekend.
There's another SSD review going live this weekend, though :-)
lunadesign - Wednesday, August 15, 2012 - link
@Kristian - Any update on the timeframe for your review? I'm anxiously waiting to see what your testing uncovers.Shambolic - Tuesday, August 7, 2012 - link
BTW, anyone who thinks SSD prices are too expensive should check them out again. There has been a dramatic price drop even in ridiculously overpriced Australia. In about the last 2.5 months local prices for a Samsung 830 128GB SSD have dropped from ~$170 to $115. And of course Americans are even better off.HisDivineOrder - Wednesday, August 8, 2012 - link
Samsung is doing a great job of getting their drive down in pricing so that it's a no brainer at times. Their rep is actually rather good all thing considered with regards to SSD's. Certainly, not the OCZ "down in the toilet" low.And when the performance is solid, they don't seem to have the problems to flash out that other brands always tend to have, they offer solid tools to monitor their drives, and THEN they start offering stellar pricing...
It just makes it easy to recommend Samsung and never think about it again. Until they release a new drive, that is.
amdwilliam1985 - Wednesday, August 8, 2012 - link
Totally agree with you too.I see Samsung 830 128GB SSD going for as low as $80 to $90 from very reputable e-tailers, big recommendation to all my friends. 128GB are the best price/performance ratio, and it has plenty space to put all software on it.
FunBunny2 - Tuesday, August 7, 2012 - link
Yet another attempt at "enterprisey" SSD, and no on board power. Unless I missed it. I mean, how much can a cap actually cost?Visual - Wednesday, August 8, 2012 - link
Well, it's not as simple as "put a cap".They will need to add logic to detect the power cut, to reroute power input from the capacitor, to write the DRAM content to SRAM quickly and efficiently. It is adding yet another possible point of failure that needs thorough testing and validation, and I suspect even adds some degree of legal responsibility to them. For normal drives they clearly state data loss is possible and you shouldn't complain if it happens, but for drives that claim to implement data loss prevention failure to meet that claim in even a small fraction of cases could be a serious problem.
Death666Angel - Wednesday, August 8, 2012 - link
Shouldn't data sensitive enterprise users use UPS in any case and wouldn't that work for any SSD/HDD? Or are they not enough?Jaybus - Wednesday, August 8, 2012 - link
They are not enough. What if a power supply fails, rather than utility power? What if a UPS fails? The data in the DRAM cache that has not yet been written to Flash is lost, and that is quite likely to corrupt database files, etc. This is why caching RAID controllers generally have on-board battery power.Beenthere - Saturday, August 11, 2012 - link
It seems that some of the increased performance in recent months on SandForce based SSDs may be at the expense of the TRIM feature - which can result in a big loss of performance over time.http://www.tweaktown.com/articles/4870/lsi_sandfor...
lunadesign - Friday, August 17, 2012 - link
Is this drive going to be reviewed by AnandTech soon?scbdpa - Tuesday, August 28, 2012 - link
i guess this one won't be reviewed?