Intel's SSD has been bested. New Sheriff in town
Anand tests preproduction OCZ Vertex 2 with Sandforce controller
Sandforce is a Silicon Valley company that doesn't make SSD but focuses their effort on making the best controller. They have some upcoming controllers (SF-1200, SF-1500) that not only beat the competition in sequential reads/writes but even beats Intel's MLC product handily in random writes by a healthy margin and just edges Intel's SLC Extreme series (Sandforce 50MBps versus Intel's SLC at 48MBps)
Sequential read/write are over 250MBps and basically limited by SATA 3Gbps speed.
Sandforce has been able to design a controller that delivers great performance without the need for using as large a DRAM cache as well.
What's interesting is their RAISE technology which is akin to offering RAID5 like features for data integrity. True to form Anand said the preproduction unit was stable unlike many of the preproduction SSD coming across his bench so this bodes well for SSD stability.
Prices are not out yet but clearly the Vertex 2 Pro is going to command premium pricing and will actually be worth it for some people.
For this type of performance I could certainly see going with a smaller drive for Boot and moving larger data to high density storage.
I imagine in a couple of years the landscape of computers will be on a new path. Optical disc and HDD will be out at the client and HDD will become secondary mass storage with Cloud becoming our tertiary storage.
If Light Peak takes off you have more of a reason to move to networked storage (because of it's flexibility with multiple protocols and speed) for your mass storage requirements.
I think how we view storage is going to become less about how much we get per Gigabyte and rather what kind of performance is it offering along with its size.
Sandforce is a Silicon Valley company that doesn't make SSD but focuses their effort on making the best controller. They have some upcoming controllers (SF-1200, SF-1500) that not only beat the competition in sequential reads/writes but even beats Intel's MLC product handily in random writes by a healthy margin and just edges Intel's SLC Extreme series (Sandforce 50MBps versus Intel's SLC at 48MBps)
Sequential read/write are over 250MBps and basically limited by SATA 3Gbps speed.
Sandforce has been able to design a controller that delivers great performance without the need for using as large a DRAM cache as well.
What's interesting is their RAISE technology which is akin to offering RAID5 like features for data integrity. True to form Anand said the preproduction unit was stable unlike many of the preproduction SSD coming across his bench so this bodes well for SSD stability.
Prices are not out yet but clearly the Vertex 2 Pro is going to command premium pricing and will actually be worth it for some people.
For this type of performance I could certainly see going with a smaller drive for Boot and moving larger data to high density storage.
I imagine in a couple of years the landscape of computers will be on a new path. Optical disc and HDD will be out at the client and HDD will become secondary mass storage with Cloud becoming our tertiary storage.
If Light Peak takes off you have more of a reason to move to networked storage (because of it's flexibility with multiple protocols and speed) for your mass storage requirements.
I think how we view storage is going to become less about how much we get per Gigabyte and rather what kind of performance is it offering along with its size.
Comments
Sequential read/write are over 250MBps and basically limited by SATA 3Gbps speed.
Yeah, they are very high performance. Here's a test on SATA 3 - must be an older drive but the same series:
http://www.nordichardware.com/news,8596.html
internal quad-RAID design allows it to get up to 480MB/s read and 550MB/s write.
Interesting notes on this page about the new drive:
http://www.nordichardware.com/news,10481.html
Says it won't arrive until March.
Prices are not out yet but clearly the Vertex 2 Pro is going to command premium pricing and will actually be worth it for some people.
I had hoped they'd reach a certain performance point and then drive the price down by now. For the consumer market, there's little point in going above SATA 3 so that's a good place to stay as the sustained write saturates that.
I think how we view storage is going to become less about how much we get per Gigabyte and rather what kind of performance is it offering along with its size.
I agree. I think there's a limit to how much storage people really need in their machines. There will be the odd few who want to store their 1,000 favorite movies in 1080p but even at that, you're talking about 10GB per movie = 10TB. You could almost get that on a single Drobo or similar. Hook that to a network and have your 250GB SSD internal and you can stream the data.
You'll never need to take it all with you so just sync what you need when you go.
Optical discs work for distribution but network streaming is just so much more convenient. If we get to 100Mbit cellular networks on 4G soon enough:
http://news.bbc.co.uk/1/hi/technology/8412035.stm
Blu-Ray could easily vanish because an entire 10GB 1080p H264 film will download in under 20 minutes and very easily stream down in real-time. A 50MBit upload is great for online backups too.
http://news.bbc.co.uk/1/hi/technology/8412035.stm
Why do I get the feeling this "ten times faster" will end up being actually two-to-three times faster when using a real product.
Why do I get the feeling this "ten times faster" will end up being actually two-to-three times faster when using a real product.
That always seems to be the case doesn't it?
Anyways, it's interesting that this topic has come up because along with my planned purchase of the Apple tablet I'm also budgeting for a upgrade to my PC. And that budget will include a top of the line SSD so I appreciate this forum topic.
I'm getting an SSD, 8 gigs of ram, and yes folks......windows 7. Dun dun dahhhhh!
I know on my MBP I was up to around 70GB just installing apps on the new machine befor any significant amount of data was created or loaded. That excluding data considered to be part of the installation of a software package, but things like XCode, neoOffice and the like take up lots of room. It is also likely to be a very long time before those apps are successfully replaced by the cloud. In the end it is the mainframe wearing a new skirt but the same old ugliness resides under that skirt.
Moving away from the cloud, the excessive focus on transfer rates really is misguided and takes the eye off the lack of the lack of storage space on the SSD. People should be ashamed considering how easily and quickly the marketing departments where able to pull the wool over your heads. Faster boot times are nice but not truely significant with respect to overall run times on a PC. As to app performance that is dependent upon what the app does. We don't need however extremely fast transfer rates to watch a movie, relatively transfer speed isn't important anymore for that usage. Lots of storage for those GB sized files is however. Start editing or transcoding those files and your storage needs easily quadruple.
As to that so called RAID like internal performance that is all well and good but what we really need is for that internal RAID to protect from device failures. Like it or not Flash is not that reliable, especially in the higher density versions. So what we really need is for true internal RAID like behaviour that actually increases the reliability of flash not just to make it faster.
The sad reality is that Flash at one time could be easily counted on to retain data for ten or more years. With the latest process nodes you are lucky to get two years of data reliabilty. Don't believe me the look it up on line. In anyevent this is a stark signal that Flash has it a dead end.
I do agree with the indicated state of optical drives, they need to be deleted from laptops now. However my reasoning is different in that I want that space for more internal storage. That could be a hybrid SSD HDD system or a notebook set up with an array of SSD. The goal is to address performance some what but to also massively increase internal storage capacity. Apple should be able to squeeze in three SSDs, of conventional design, into a 15" laptop. Drop the legacy mechanicals and interface and the might gave room for four SSDs on Mini PCI Express cards.
Whatever, all I know is that we are stuck in a world of legacy secondary storage. Apple is in a position to break out from the pack here. They should too as the mechanicals of a rotating disk are archaic. For something like the rumored tablet SSD on a PC board is the only option to soldered in flash.
In the end this new SSD, while impressive performance wise, has far to many feet in the grave. The industry needs something better than SATA3 connecting to a kegacy mechanical form factor.
Dave
That always seems to be the case doesn't it?
Anyways, it's interesting that this topic has come up because along with my planned purchase of the Apple tablet I'm also budgeting for a upgrade to my PC. And that budget will include a top of the line SSD so I appreciate this forum topic.
I'm getting an SSD, 8 gigs of ram, and yes folks......windows 7. Dun dun dahhhhh!
interesting but nothing mentioned of the processor or GPU each ofvwhich has a bigger overall impact than the disk drive for desktop use. Especially elwhen all the goodies in Windows are turned on.
I don't want to discount SSD performance just that you seem to elevate it's importance way to much. Atleast in my opinion. I sit here looking at my old 2008 MBP and I think gee that would be much snappier with an SSD installed. But i already know that such a drive would not give me the room i need. Especially considering I'm already using an external drive. Maybe by the time I replace that MBP phase change or nano mechanical storage will have replaced Flash and other legacy tech.
Dave
Why do I get the feeling this "ten times faster" will end up being actually two-to-three times faster when using a real product.
That's still a great margin.
Anand tests preproduction OCZ Vertex 2 with Sandforce controller
Sandforce is a Silicon Valley company that doesn't make SSD but focuses their effort on making the best controller. They have some upcoming controllers (SF-1200, SF-1500) that not only beat the competition in sequential reads/writes but even beats Intel's MLC product handily in random writes by a healthy margin and just edges Intel's SLC Extreme series (Sandforce 50MBps versus Intel's SLC at 48MBps)
Sequential read/write are over 250MBps and basically limited by SATA 3Gbps speed.
Sandforce has been able to design a controller that delivers great performance without the need for using as large a DRAM cache as well.
What's interesting is their RAISE technology which is akin to offering RAID5 like features for data integrity. True to form Anand said the preproduction unit was stable unlike many of the preproduction SSD coming across his bench so this bodes well for SSD stability.
Prices are not out yet but clearly the Vertex 2 Pro is going to command premium pricing and will actually be worth it for some people.
For this type of performance I could certainly see going with a smaller drive for Boot and moving larger data to high density storage.
I imagine in a couple of years the landscape of computers will be on a new path. Optical disc and HDD will be out at the client and HDD will become secondary mass storage with Cloud becoming our tertiary storage.
If Light Peak takes off you have more of a reason to move to networked storage (because of it's flexibility with multiple protocols and speed) for your mass storage requirements.
I think how we view storage is going to become less about how much we get per Gigabyte and rather what kind of performance is it offering along with its size.
In three years, Cloud will be dead and replaced with another promised clone.
Moving away from the cloud, the excessive focus on transfer rates really is misguided and takes the eye off the lack of the lack of storage space on the SSD. People should be ashamed considering how easily and quickly the marketing departments where able to pull the wool over your heads. Faster boot times are nice but not truely significant with respect to overall run times on a PC. As to app performance that is dependent upon what the app does. We don't need however extremely fast transfer rates to watch a movie, relatively transfer speed isn't important anymore for that usage. Lots of storage for those GB sized files is however. Start editing or transcoding those files and your storage needs easily quadruple.
The HDD is the slowest component in your system. There's a reason that folks have been pretty excited over SSD performance. Not everyone stores movies on their machines and frankly those are easily kept on a NAS which has acceptable performance for watching movies.
As to that so called RAID like internal performance that is all well and good but what we really need is for that internal RAID to protect from device failures. Like it or not Flash is not that reliable, especially in the higher density versions. So what we really need is for true internal RAID like behaviour that actually increases the reliability of flash not just to make it faster.
Like it or not HDDs are less reliable than flash.
The sad reality is that Flash at one time could be easily counted on to retain data for ten or more years. With the latest process nodes you are lucky to get two years of data reliabilty. Don't believe me the look it up on line. In anyevent this is a stark signal that Flash has it a dead end.
Manufacturer data sheets still say 10 years.
If you want to use flash for archival media pick SLC flash. Even then, you're probably better off with multiple archival quality gold BRDs. For HDDs I've read as low as 5 years sitting on a shelf. For pretty much every technology you need to refresh every 5 years anyway. Tapes, HDDs need to get rewritten. Optical media needs to get reburned to fresh media.
I do agree with the indicated state of optical drives, they need to be deleted from laptops now. However my reasoning is different in that I want that space for more internal storage. That could be a hybrid SSD HDD system or a notebook set up with an array of SSD. The goal is to address performance some what but to also massively increase internal storage capacity. Apple should be able to squeeze in three SSDs, of conventional design, into a 15" laptop. Drop the legacy mechanicals and interface and the might gave room for four SSDs on Mini PCI Express cards.
500GB local storage on a laptop seems decent enough for me now. I can always supplement with removable media and external drives.
Whatever, all I know is that we are stuck in a world of legacy secondary storage. Apple is in a position to break out from the pack here. They should too as the mechanicals of a rotating disk are archaic. For something like the rumored tablet SSD on a PC board is the only option to soldered in flash.
Soldered in flash is stupid for the tablet given it should be large enough for a 1.8" drive. It removes any opportunity to upgrade to larger space.
In the end this new SSD, while impressive performance wise, has far to many feet in the grave. The industry needs something better than SATA3 connecting to a kegacy mechanical form factor.
Right.
Why do I get the feeling this "ten times faster" will end up being actually two-to-three times faster when using a real product.
The real-world examples of commercial services show it equaling or exceeding home broadband speeds. You can get this type of service in some parts of the US:
http://www.clearwire.com/
3-6Mbps is certainly a lot lower than 100Mbps but still fast enough to stream HD video.
First and most irritating is the idea that people will use the cloud for primary storage.
Nobody mentioned that. Tertiary storage is what was said.
SSD = primary storage with boot drive, apps and working data
home backup solution = secondary storage with backups and larger archives available for sync
cloud = tertiary storage for more backups, streaming media content and wireless syncing
Faster boot times are nice but not truely significant with respect to overall run times on a PC.
It's not just boot times but file opening and saving times and program load times. Designers who have to work on files that are hundreds of MB in size would rather have them open and save in 1 second than 10 seconds.
As to that so called RAID like internal performance that is all well and good but what we really need is for that internal RAID to protect from device failures.
The drive mentioned has something like internal RAID 5 so there's some in-built data protection.
That could be a hybrid SSD HDD system or a notebook set up with an array of SSD. The goal is to address performance some what but to also massively increase internal storage capacity.
The SSD in question should come in capacities up to 400GB but the price is what will drive a hybrid solution first. I agree that Apple would do well to offer configurations where you get an SSD as a boot drive and 1x or 2x 2.5" mechanical drives for additional storage and get the ball rolling on the removal of optical drives once and for all.