backtop

Print E-mail del.icio.us 53 comment(s) - last by therealnickdan.. on Aug 20 at 4:16 PM

Intel's 2.5" and 1.8" SSDs  (Source: Intel)

Intel SSD Roadmap  (Source: Expreview)
Intel SSDs will have a read speed of 240MB/sec and a write speed of 170MB/s

Today the SSD market is not that big. Many consumers aren't willing to spend the significant price premium over a standard hard drive to get the benefits of an SSD. Some businesses, however, realize the potential of the SSD in an enterprise computing environment.

Expreview reports that it has an Intel roadmap that shows that three new SSDs are going to be unveiled by Intel. The SSDs will be in the X-series and are said to have a read speed of 240MB per second and write speed of 170MB per second. The SSDs will come in 2.5-inch (X25) and 1.8-inch (X18) form factors and will come in two models -- E and M. The E model is reported to be the top-of-the-line product with the highest performance, while the M model is configured for power savings.

Expreview reports that the X18-M will only consume 0.25 W of power in active mode. Capacities for the X25-E model are reported to be 32GB, 64GB, and 160GB with availability in the last quarter of 2008. An 80GB X25-M and an 80GB X18-M are reported to be launching in Q3 2008. A 160GB version of the X18-M is expected to hit the market around Q1 2009.

Micron and Intel have been working together to develop what the companies say will be the world's fastest NAND flash memory. DailyTech reported in early August that Micron had unleashed its own SSDs that offered read speeds even faster than what the Intel roadmap is claiming. Micron's SSDs claim to reach a read speed of 250MB per second.

DailyTech reported in April that Intel's SSD prototypes were going through in-house testing; odds are these are the same SSDs that are now showing up on the roadmap.

Comments     Threshold


This article is over a month old, voting and posting comments is disabled
But what about stability
By hlper on 8/15/2008 1:04:35 PM , Rating: 2
I notice that in the most recent SSD articles there has been no mention of the limited read-write capacity of SSD drives. Even in the discussion, no one brings it up. I am bringing it up because I am wondering if no one thinks this is still a problem? Are the latest generation of SSD's good enough for your precious data?

Personally, I think speed is great, but I want reliability before I give up my traditional hard drive. I will also admit am not an expert on the subject of these drives, so feel free to educate me.




RE: But what about stability
By Chadder007 on 8/15/2008 1:14:29 PM , Rating: 2
Thats what I thought about too...


RE: But what about stability
By rippleyaliens on 8/15/2008 3:45:39 PM , Rating: 5
If you rely even on a raid controller for your data.. then you are still not in the game.
SSD=low power, somewhat performance, but IO- it the Main advantage of SSD.
100mb in/out, who cares. BUT >>30k Disk IO now that is what i want. I have 2x 300SAS 15k drives, probably the fastest on the market. Raid controller. I stripe the drives.. If i loose 1, i wait on the warranty, if the raid gets hosed. then i just Re - Acronis the system back. takes me 15min to do a restore. Complete restore. Now for my critical DATA, that is on my server, running raid6. HORRIBLE write speeds. compared to my sas drive. BUT it is on 1) second server, 2) Mission critical stuff, is burned to DVD. and also backup to tape.

Stability, these drives , even if they last just 2 years of HEAVY usage, its worth it. For in 2 years, there will be a 2x maybe 3x in capacity, and possibly 2x in performance. So stability is mute. I still have usb thumb drives 3 years old, still rocking. Hard drive's still rockin...

WAYYYY back when 1.6gb WD drives hit, they were indeed the fastest.. but in 3 years when 20gb drives hit, (fall 1999), would you still want the 1.6 wd 5400 rpm, or the 20gb 7200 drive.. DUH.. Its not like you will keep a hard drive forever.. I would hope you would replace them . .. right???

CPR you only get 2 out of 3
C= Cheap
P= Performance
R= Reliability
if you want cheap but good performance you dont get reliability
If you want Cheap and Reliable- you dont get performance
If you want Performance and Reliability, it wont be cheap..


RE: But what about stability
By retrospooty on 8/15/2008 9:42:08 PM , Rating: 5
"if you want cheap but good performance you dont get reliability
If you want Cheap and Reliable- you dont get performance
If you want Performance and Reliability, it wont be cheap.. "

Or you could buy a Maxtor and get only one... cheap.


RE: But what about stability
By JackBurton on 8/15/2008 1:16:30 PM , Rating: 2
2 million hours MTBF. Is that good enough for you? :)


RE: But what about stability
By Icelight on 8/15/2008 1:20:51 PM , Rating: 2
Because of how they come to that number: No. :)


RE: But what about stability
By TomZ on 8/15/2008 1:44:53 PM , Rating: 5
MTBF doesn't model the wear-out mechanism inherent in SSD drives; it is just for "random" premature failures.

SSD wear-out can be modeled by something along the lines of a maximum amount of data that can be written to the drive before the minimum guaranteed number of flash erase/write cycles starts to be exceeded.

For example, a particular Toshiba MLC SSD drive is rated for 40TB of write activity before the minimum cycle limit begins to be hit. If you assume a 5-year desired lifespan, this works out to 44GB of writing per day. Toshiba estimates that average users write 1.4GB per day, and power users 5.2GB. Based on this model, Toshiba expects this drive to easily last more than 5 years in typical usage (with margin).

ref. http://www.toshiba.com/taec/adinfo/ssdfacts/, see SSD Myth #5.

Also, for comparison, this is for a Toshiba MLC drive. SLC drives have much higher guaranteed number of erase/write cycles, and therefore, would have a much longer expected life compared to the 5-year example above.

I don't know whether the Intel drive is SLC or MLC. Maybe the high-end one is SLC and the low-end one MLC?


RE: But what about stability
By icanhascpu on 8/15/08, Rating: 0
RE: But what about stability
By leexgx on 8/15/2008 10:36:35 PM , Rating: 2
one ssd i have seen think it was that super insane fast PCI-e 4x card that had 500gb of space and needed no power that did 900MB/s,

that drive had auto bad sector remove from the drive space pool so when an part became unwriteable it would just mark it as bad and make the drive smaller (some driver is installed to make the changes in windows auto or on next reboot)

the life of SSD are more likey to be more reliable then an spinning disk, guess i have to wait 1-2 years before i get SSD get the price to an point of an raptor (£150 or less) 200MB/s plus 0.1ms access time never need to use raid 0 agane for my use any way

on the points of dieing SSD first thing you should do on vista is turn off the auto defrag or any defrag tool (on Vista mite not be an bad idea to turn off system restore as well) as that can reduce the life of the SSD by alot SSD does not need defraging all parts of the disk go at the same speed fragmented or not (system restore makes restore points in 400MB in sizes or the system restore volume shaddow service can use as much space thats in your user folder space can bog the pc down for 30-60 mins before it gives up due to out of space or it passes 1 hr)


RE: But what about stability
By Baov on 8/16/2008 11:10:47 AM , Rating: 4
I have a EWF write filter on top of my system partition and from what it tells me, i usually don't write anything even close to 500meg a day. Given, i've optimized my system, but i don't feel like i'm missing anything.


RE: But what about stability
By Digobick on 8/15/2008 1:22:25 PM , Rating: 4
SSDs are also more reliable than HDDs. Using wear-leveling technology, SSDs offer mean time before failure (MTBF) ratings of approximately two million hours compared to typical HDD MTBF ratings of between 300,000 to 500,000 hours.


RE: But what about stability
By 306maxi on 8/15/2008 1:29:09 PM , Rating: 2
The only problem is - and correct me if I'm wrong - that there is currently no easy way to get data off a failed SSD. At least if you PC gets hit by a power surge which blows your hard drive you can get the board swapped out and get your data off. With an SSD you might as well not bother.


RE: But what about stability
By TomZ on 8/15/2008 1:48:07 PM , Rating: 5
Rather than worry about whether SSD and/or HDD can be recovered, it makes more sense to just back them up. Problem solved.


RE: But what about stability
By 306maxi on 8/15/2008 2:20:32 PM , Rating: 2
I agree. But most people like myself don't have enough money to buy a terrabyte of SSD storage in the first place. So for most people with a desktop that buy an SSD, an SSD is going to be a thing they have their OS on and they'll store all of their photos, videos and so on which don't need the high read write speeds on a traditional hard drive.


RE: But what about stability
By RU482 on 8/15/2008 3:34:56 PM , Rating: 1
But but but...what if your system gets zapped during a backup?? Then you have a Fubar'd main drive, and a Fubar'd backup. How are you going to get your data then??


RE: But what about stability
By TomZ on 8/15/2008 3:37:01 PM , Rating: 4
RE: But what about stability
By GaryJohnson on 8/15/2008 10:38:46 PM , Rating: 4
If your files are that critical you should be backing them up offsite.


RE: But what about stability
By 306maxi on 8/15/2008 1:25:33 PM , Rating: 3
Storing your holiday snaps or baby videos on these would be dumb. This is more a drive for you OS and other applications which will make be helped by the fast reads and writes. For large amounts of data magnetic storage will still be the best choice for vital data for reasons of cost and reliability.

I'll certainly be buying an SSD when prices get a bit lower. But I'll still keep my important stuff saved on a traditional magnetic storage hard drive.


RE: But what about stability
By Spectator on 8/16/2008 2:25:28 AM , Rating: 2
It would make sence to have a 2-4 gig dram partition to place the swap file on.

That would ease any fears about wearing things out quickly yes?


RE: But what about stability
By winterspan on 8/16/2008 4:50:47 AM , Rating: 2
The reason why is because it is not an issue anymore for decent SSDs. I have seen multiple articles that went into the issue, and basically they found that modern SSDs will easily last many multiples of a PCs intended lifetime. I will try to dig up some links..


wow
By Natfly on 8/15/2008 12:32:18 PM , Rating: 2
Pretty soon we'll need something faster than sata 2.




RE: wow
By paydirt on 8/15/2008 12:43:43 PM , Rating: 2
I thought SATA2 could handle 3GB/s, or maybe that's 3 Gb/s which would be about 360 MB/s?


RE: wow
By Natfly on 8/15/2008 12:49:10 PM , Rating: 3
3Gbit/s
3000MHz
1bit/clock
after encoding comes out to 300MB/s

http://en.wikipedia.org/wiki/Serial_ATA#SATA_3.0_G...


RE: wow
By rudolphna on 8/15/2008 9:43:29 PM , Rating: 2
but there is of course system overhead. Remember, USB 2.0 throughput is supposed to be 60MB/s. With just the external 3.5" HDD connected (WD BTW) maximum read speads are only 32MB/s. I have an identical drive in my computer (500GB WDC 16MB) I ger reads of 90MB/s.. USB overhead. That is why firewire is faster, the theoretical throughput is lower, but there is less bus overhead.


RE: wow
By leexgx on 8/15/2008 11:20:11 PM , Rating: 2
its 300MB/s norm


RE: wow
By therealnickdanger on 8/15/2008 12:51:04 PM , Rating: 2
Well, SATA will have to be bumped up to 6Gbps soon, but there will have to be a new interface eventually. SSDs are just getting started right now and are already reaching the 3Gbps barrier. It will not take very long at all to hit the 6Gbps barrier either. I don't think most folks realize just how fast SSD is going to get. Basically, in the very near future, we're looking at local storage operating at speeds very close to RAM as we know it. It will escalate quickly IMO.


RE: wow
By Natfly on 8/15/2008 1:11:37 PM , Rating: 3
I'm not sure how soon SATA 6Gb/s will actually become available. It seems like it is still largely a work in progress. Draft specifications just came out last month, I don't know when the final revision is expected. Not to mention the time it takes for chipset makers implement the controllers, and motherboard makers releasing new boards. I wouldn't hold my breath.


RE: wow
By therealnickdanger on 8/15/2008 2:27:33 PM , Rating: 4
It hasn't been a priority because mechanical drives have not even been able to saturate the 3Gbps bandwidth. Not to sound like a broken record, but by this time next year, SSDs will easily be operating at the limit of SATA/3Gbps for reads and writes. The industry will demand a faster interface and the "work in progress" will be put on the fast-track. Sure, it may be 2010 before we see any actual hardware for SATA/6Gbps, but it is likely it will take even less time for a replacement of that.

I think of the Gigabyte i-RAM. It operates on a PCI interface with a high degree of speed. I'd imagine a device could be made to take advantage of the PCI-E 2.0 interface in order to satiate the demand of faster and faster SSDs.


RE: wow
By therealnickdanger on 8/20/2008 4:16:28 PM , Rating: 2
Not SATA
By iwod on 8/15/2008 12:37:04 PM , Rating: 2
I am not even sure if SATA was suppose to be for SSD. Would PCI Express be much better fit?




RE: Not SATA
By TomZ on 8/15/2008 12:45:13 PM , Rating: 2
No, SATA is better, for a few reasons:

1. SSDs can reuse the long-used ATA command set, which SATA uses, but PCIe doesn't. This means that existing software drivers can be used without modification (except for possible SSD optimization).

2. SSDs can reuse the existing SATA connectors and defined mechanical envelopes established for HDDs. HDDs and SSDs can therefore be supported and interchanged in the same slot, and zero development is required on the part of desktop/laptop OEMs to support SSD.

SATA is a perfect fit for this transitional product.

But I will add that, connecting the flash directly to the CPU (on the CPU data bus), and allowing the main CPU to do the work in place of the embedded controller in the SATA SSD would lead to at least two benefits - much better performance (especially reading) and potentially lower cost.


RE: Not SATA
By William Gaatjes on 8/15/2008 1:42:08 PM , Rating: 2
quote:
But I will add that, connecting the flash directly to the CPU (on the CPU data bus), and allowing the main CPU to do the work in place of the embedded controller in the SATA SSD would lead to at least two benefits - much better performance (especially reading) and potentially lower cost.


In order to reach similair performance, flash memory has to be read and written at the speeds of at least current ddr2 technologies. I don't think flash is that fast yet or ever will be. What you propose is that the flash memory replaces the dram memory and that is not going to happen. Write and read speeds are way too low for that. If flash ever starts to match current dram speeds, you would still want a seperate controller and port aside the main memory bus with then newer and faster dram technologies.

Anyway i like your idea, maybe a flash drive with controller and directly connected to the cpu through a hypertransport link for example. That would not be bad at all. But that flashdrive has first to reach 1GB/sec barrier at least. It would save the latency going from cpu > hypertransport > pci-ex. Imagine the speed such a pc would have...


RE: Not SATA
By TomZ on 8/15/2008 1:53:26 PM , Rating: 2
quote:
In order to reach similair performance, flash memory has to be read and written at the speeds of at least current ddr2 technologies.

I don't see why this has to be the case. Nor am I suggesting that flash replace DRAM at all.

In fact, the BIOS, which is flash-based right now and much slower than DDR2, is already mapped into the CPU's memory space. That already works today.

The differences I would see would be that the flash would be NAND instead of NOR (what the BIOS uses now), and that a device driver, BIOS, or some other software piece would handle the wear-leveling and sector re-mapping required for NAND.


RE: Not SATA
By William Gaatjes on 8/15/2008 2:44:07 PM , Rating: 2
quote:
In fact, the BIOS, which is flash-based right now and much slower than DDR2, is already mapped into the CPU's memory space. That already works today.


If you mean that the bios routines are copied into fast dram memory and executed from that dram memory your right. If you mean it is mapped in the adress space and execute from the bios flash chip, well that is not what is happening. The wait states for the cpu would be in millions.

quote:
The differences I would see would be that the flash would be NAND instead of NOR (what the BIOS uses now), and that a device driver, BIOS, or some other software piece would handle the wear-leveling and sector re-mapping required for NAND.


With these sort of things dedicated hardware under tight software control is much faster then software alone at the moment. But with many cores to come and a dedicated connection like hypertransport it could actually work.
However, i think you still need some dedicated mmu like unit to do all the housekeeping and remapping of memory addresses for wear leveling on the cpu. It would be like moving all the intelligence on the cpu die in software and hardware. Very much possible. The SSD could be nothing more then a flash memory connected to a simple controller with a HT link. The controller would just turn the commands and data from the ht link to data and commands the flashchips understand.


RE: Not SATA
By JKflipflop98 on 8/16/2008 9:24:29 PM , Rating: 2
You're referring to phase-change memory. Dram speeds with flash-like data retention. Good stuff.


RE: Not SATA
By William Gaatjes on 8/17/2008 4:43:27 AM , Rating: 2
I read about that. Very promising.
It has faster write speeds, it is much more durable then flash. But it is temperature sensitive and in a pc that is a bad thing. The temperature range has to be increased first. I don't know how far progression has come at this moment.

There is also Feram and Mram. But seems like a dead end.

But maybe this technology will make it if they don't drown it in patents : The memristors. It seems very promising.

http://www.edn.com/blog/980000298/post/1790025979....


RE: Not SATA
By RU482 on 8/15/2008 3:57:28 PM , Rating: 2
I know Sandisk makes PCI-e minicard SSDs
There is some type of 3rd party driver involved


And they'll cost you...
By HaZaRd2K6 on 8/15/2008 12:27:13 PM , Rating: 1
Your left arm, both your kidneys, and you right leg.

I mean I'm all for stupidly fast speeds, but they can't possibly be anywhere near inexpensive.




RE: And they'll cost you...
By 306maxi on 8/15/2008 12:39:42 PM , Rating: 3
I would say that with Intel getting into the business that this will help to push prices down quicker.


RE: And they'll cost you...
By ImSpartacus on 8/15/2008 1:02:03 PM , Rating: 2
Yeah, Intel will play price games. They have deep enough pockets. They aren't going to throw out $3000 SSD's just so two dozen websites can bench them and have no one else want them. They will be reasonable.


RE: And they'll cost you...
By paydirt on 8/15/2008 12:41:26 PM , Rating: 2
It will be interesting to see how Intel prices these... If they were to match the price of the Acer (or whatnot, cheap SSDs) with superior performance on the M (mainstream) line, I think they will sell like hotcakes. Or even if at a slight premium...


RE: And they'll cost you...
By Khato on 8/16/2008 6:33:45 PM , Rating: 2
I'm expecting them to be around the price points of the OCZ core series for the MLC versions (lower write speed), then maybe twice as much for the SLC...

Intel's point in this venture is to create a profitable market for all the flash capacity they have. Most of the reason for their flash division losing so much money last quarter is that there simply isn't enough demand.


Sizes
By BansheeX on 8/15/2008 12:43:08 PM , Rating: 2
I think the interesting aspect of all this SSD news is that 3.5" is practically being abandoned. I have long desired for desktop parts to become more like they're smaller notebook brethren. There's really no reason we can't reduce the PC footprint with 2.5 hard drives and slimline optical drives. Ultimately, there will be material savings, cost savings, and power savings in pursuing this. It used to be that notebooks were a niche product, and thus their parts were way more expensive, but that's changed. Notebooks are big sellers these days. I just bought a 320gb 2.5 WD drive for $100. It's completely silent and faster than my old 7200rpm 3.5 drive due to its higher density and larger cache. I couldn't have foreseen this years ago, but it's great.




RE: Sizes
By PAPutzback on 8/15/2008 1:23:26 PM , Rating: 2
In order for this "small footprint" to happen we need Blu-ray capacity to double and then sell 2.5 inch blue Ray discs and drives. Then we will be on our way to a very small factor unit. I am all up for it.


RE: Sizes
By natanojlui on 8/15/2008 9:18:40 PM , Rating: 2
Maybe Intel is making the drives in 2.5 and 1.8 inches because it is for their 'Centrino' Brand. Since, Centrino is so successful, it would allow them to sell more SSD. Plus, SSD are always marketed to be "low-power", which appeals to the mobile users.


RE: Sizes
By rudolphna on 8/15/2008 9:46:23 PM , Rating: 2
HD4870, Geforce 9800GX2, and (presumably) Intel Larrabee. Those are BIG cards. I dont think they will fit well in a smaller system.


Don't forget about...
By pauldovi on 8/15/2008 1:44:37 PM , Rating: 2
Random writes, which are horrible on SSD's.




RE: Don't forget about...
By A Mad Pole on 8/15/2008 2:20:28 PM , Rating: 2
Are those slow random write speeds caused by the Flash memory characteristics or the Flash controller shortcommings?


RE: Don't forget about...
By Hypernova on 8/15/2008 6:15:50 PM , Rating: 2
Flash memory characteristics IIRC

Data stored on flash are in 64B clusters so even if you just want to change one byte you need to erase the entire cluster and rewrite in the same 63B and 1B new data. This increases wear and slows things down when write data size is less then cluster size.


bring it on
By kattanna on 8/15/2008 12:23:32 PM , Rating: 2
as more players enter the field, prices will drop and performance will increase.

one of the truly great things about the computer industry




RE: bring it on
By ksherman on 8/15/2008 1:49:26 PM , Rating: 3
one of the truly great things about capitalism.

Fixed for you... ;)


What about RAID 0
By Nijje on 8/16/2008 12:46:50 PM , Rating: 2
Am I correct in assuming MTBF time is rather pointless when comparing them to regular platter drives since total memory block degradading is far more likely to occour prior to reaching a threshold for drive failure when doing 20Gb+ writes a day?
Doesn´t that also mean that larger drives will be a safe bet to extend the drive life?
If so then running an RAID 0 array of 4x 32gb should fail at the same time as a 128gb, but obviously preform much faster.
When adding regular platter drives in a raid array the MTBF time is often recalculated by dividing the drive´s specc MBTF with the number of drives since the probability of any one failing rises with the amount of drives added. (of course running large arrays without failure for years is very common)
But does that really apply in the same way with an array of SSD´s ?
From a corporate perspective it probably makes a huge difference when doing the numbers.




By Cullinaire on 8/16/2008 1:10:18 PM , Rating: 2
*Gushing* Oh, I'm SO getting one of these when the 64GB version is FAR!




"When an individual makes a copy of a song for himself, I suppose we can say he stole a song." -- Sony BMG attorney Jennifer Pariser

DailyTech Poll
Which web browser do you use on your primary personal machine? 






44 Comments









botimage
Copyright 2009 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki