Apple SSD?
I wonder why Apple aren't making SSD themselves. May be they want to wait for the technology to mature a bit more?
Since they are already the largest Flash buyer and gets favorable pricing from Flash manufacturers. Snapping Flash with a SSD Controller surely is simple enough to bring the performance of Mac to a new level.
Since they are already the largest Flash buyer and gets favorable pricing from Flash manufacturers. Snapping Flash with a SSD Controller surely is simple enough to bring the performance of Mac to a new level.
Comments
I wonder why Apple aren't making SSD themselves. ...
Absolutely not. Apple has neither the interest, the expertise, the facilities, nor the volume to make manufacturing SSDs a profitable enterprise.
Designing a high performance controller is another beast altogether different.
Why bother when intel, samsung and others are all working feverishly to make high performance SSDs a commodity product like high performance HDDs?
If I'd play devil's advocate I'd say:
Why bother creating a new battery controller and high capacity battery if dozens of companies work feverishly to develop longer lasting laptop batteries?
Considering that Apple probably gets the best price on Flash chips in the whole industry, they could increase their profits per computer if they were to offer their own SSDs at the same price point as they currently ask for the Toshiba SSDs.
It also allows for all-new form factors.
One thing I always suggested was to mount the SSD chips behind the screen. Screens need a certain minimal thickness or they flex too much. Quite impossible to have a 1mm thick lid, even if OLEDs would allow for it. So if lids have to be 3-5mm thick anyway for stability reasons, how about filling that space with Flash chips?
I can certainly see a MacBook Air model that's even thinner without an HD in the bottom half.
IMHO it would make sense for Apple to develop their own SSD controller, offering their own SSDs.
If I'd play devil's advocate I'd say:
Why bother creating a new battery controller and high capacity battery if dozens of companies work feverishly to develop longer lasting laptop batteries?
Considering that Apple probably gets the best price on Flash chips in the whole industry, they could increase their profits per computer if they were to offer their own SSDs at the same price point as they currently ask for the Toshiba SSDs.
2 things:
1) Toshiba gets the best price on flash chips...from Toshiba. Likewise Samsung gets the best flash price on flash chips from itself. They aren't selling flash to Apple at a loss.
2) The controller can be on the chip anyway. The flash that Apple buys from Samsung for their iPods has the controller on chip.
It also allows for all-new form factors.
One thing I always suggested was to mount the SSD chips behind the screen. Screens need a certain minimal thickness or they flex too much. Quite impossible to have a 1mm thick lid, even if OLEDs would allow for it. So if lids have to be 3-5mm thick anyway for stability reasons, how about filling that space with Flash chips?
They could but then lose the ability to change drives. This is what they do with iPods.
I can certainly see a MacBook Air model that's even thinner without an HD in the bottom half.
The MBA drive is fairly small anyway and the screen is thinner than it would be if you stuck a pcb with flash chips behind it. I don't think there's really THAT much empty space there.
IMHO it would make sense for Apple to develop their own SSD controller, offering their own SSDs.
A SSD "controller" at this point is simply a PATA or SATA controller chip. No need to make one. To control the actual flash chip, all that stuff is often on the chip itself. All you're talking about is spreading the guts of a SSD across the motherboard somewhere. At most you save a teeny bit of space at the expense of soldering all the flash in place.
Samsung gets the best price for their own flash chip. That is like SJ gets the best price for ANY apple products.
The point is SSD today have a relatively high profit margin. That is why EVERYONE are jumping on broad and making SSD.
And no. A SSD controller is not just a PATA SATA controller chip. The controller is the brain of the SSD. It is what separate Intel's SSD from the rest. Intel 's SSD is currently the fastest on the market. The make and optimize their own controller chip. While others rely on either Samsung Contorller, Jmicro ( extremely poor ) or the new Indilinx.
So in reality there are really only 4 different SSD depending on controllers. Speed will vary slightly depending on quality of flash chip used and customization of firmware.
In theory, ( if Apple and Intel are close enough ), Intel could sell the controller chip to Apple and Apple just gather the rest. It would cost apple less then $100 to make an 40GB SSD that is one of the fastest SSD on the market. And even with Indilinx it would be less then $80 for a 30GB.
I have always believed HDD is the biggest bottleneck in the whole compter system. As Anand has also repeatedly stated in his newest SSD article.
If Apple pays at least $40- $70 for the HDD anyway. An extra $10 - $30 dollars would make the Mac experience unmatched by any other PC manufacture.
...
I have always believed HDD is the biggest bottleneck in the whole compter system. ...
What you believe is not nearly as important as what is. There are many bottlenecks. That is the nature of digital electronic circuitry.
What you believe is not nearly as important as what is. There are many bottlenecks. That is the nature of digital electronic circuitry.
Actually he's correct.
When you look at the storage subsystems you see a very vast hierarchy.
Chip speed L1 and L2 caches.
RAM in the Gigabytes now that runs at 1066Mhz and throughputs GBps.
Then you get to your HDD.
Pitifully slow access times and horrible latency (4ms on the avg) as compared to
DRAM and NAND technology.
Your computer is literally sitting there waiting for the pokey HDD to retrieve the information. Nehalem is a fast architecture but it's avg speed improvement is still only %15-20 % on the avg over Penryn. A fast SSD can deliver performance delta over HDD that dwarfs the advantage of Nehalem.
If I had a Quad or Octo core I'd be spending the extra cash to put my Boot drive on an Intel SSD. It's the equivalent of adding more Ghz or more cores.
Actually he's correct.
.... It's the equivalent of adding more Ghz or more cores.
You illuminated the flaw in the logic. The goal is [presumably] to increase system performance. Adding 1 GHz to a core will increase system performance, but not as much as the increased clock speed would suggest to the naïve. A rule of thumb is that actual performance scales as the square root of the core clock speed. Going from 2 GHz to 3 GHz can be expected to improve overall performance by roughly 22%. To do better, many other engineering enhancements are required. Switching to SSD increases system performance, but again not nearly as much are the the faster component might naïvely suggest. Benchmarks suggest that SSDs in Macs roughly double overall system performance. Real world performance improvements are probably less. If the 100% performance improvement is real, then it is welcome. My rule of thumb, however, is that performance must triple before the improvement becomes obvious.
Arh...... Allow me to explain
Samsung gets the best price for their own flash chip. That is like SJ gets the best price for ANY apple products.
The point is SSD today have a relatively high profit margin. That is why EVERYONE are jumping on broad and making SSD.
And no. A SSD controller is not just a PATA SATA controller chip.
The flash itself can have one integrated (MMC anyway for MP3 players...Samsung has one). The intel one is a MLC/SATA flash controller. The Jmicron MLC/SATA controller is a particularly bad one but this still isn't rocket science...and you don't know which part Jmicron screwed up it's implementation on (could be the SATA implementation, their wear levelling, etc...controlling the NAND itself isn't the hard part...heck Intel's wear levelling is causing some oddball problems now).
The Samsung SLC/SATA controller is fine. It's not as if Samsung wont build a fast MLC/SATA controller.
Yah really think that Apple will have some magic pixie dust advantage over Samsung or Intel? And there are Indilinx based SSDs already on the market and Apple can just buy Indilinx controllers if they wanted rather than building and the fabbing their own. Sure they have those PA-Semi folks but I'm sure their busy NOT building something for a commodity market.
I have always believed HDD is the biggest bottleneck in the whole compter system. As Anand has also repeatedly stated in his newest SSD article.
If Apple pays at least $40- $70 for the HDD anyway. An extra $10 - $30 dollars would make the Mac experience unmatched by any other PC manufacture.
Flash prices are not low enough to make a SSD for $30 more than a similar sized HDD. In a couple years but not quite yet. Parity a couple years after that.
And it's not $30. It's all the R&D to fab a controller and write the wear levelling and other algorithms. Easier just to wait another year or two.
Just as the Anand's article point out, which i share most of his opinion, and i quote
Although we're still looking at SSDs in their infancy, as a boot/application drive I still believe it's the single best upgrade you can do to your machine today
The SSD, although double the speed of of fastest consumer HDD and 10 - 20x faster in random write. Doesn't feel much faster when you first use it. And i quote again
you don’t think they’re fast, until I take one away from you.
We have gone past the era of Mhz and CPU performance. Netbook, as many pundit pointed out are showing how small and comparatively slow computers are enough for everyday computer usage.
We have most of our Mhz and CPU speed we need, we then had all our memory capacity filled with cheap DDR2 Ram. Both component will continue to increase in speed and capacity as well as lowering cost following Moore's law.
HDD, on the other hand, haven't had all the speed improvement compare to other parts of the system. While their capacity have seriously increased over the past years. ( Especially over the past 2 years, jumping from 500Gb to 2TB ) Their sequential speed may only have > 50% improvement. And Random read and write will be much less then that as well.
@ Vinea
I never said Apple should build their own controller chip. All i said was buy some controller chip and snap some Flash to make a SSD. This way apple uterlise their cheap Flash contract price.If Samsung has a decent controller chip the may as well buy the whole package from Samsung anyway. Since they are already buying all their Flash from Samsung.
@Mr Me. : I understand and agree with your point. A 2Ghz system doesn't mean 100% improvement over a 1Ghz. But Performance Matrix are just numbers and benchmarks, it is the user experience that counts.
There has been a lot of rubbish in this thread about clock speed, for a given architecture and application you cAn get very good speed ups by increasing clock rate! It is a question of context. Well that and the damage that Intel has inflicted by progressively introducing slower Pentium architectures in the past causing people to believe that performance increase drops as clock rate increases.
Just as the Anand's article point out, which i share most of his opinion, and i quote
The SSD, although double the speed of of fastest consumer HDD and 10 - 20x faster in random write. Doesn't feel much faster when you first use it. And i quote again
There are many reasons to consider SSD, speed is just one of them. For many the durability in portable applications is more important than speed.
The reality here is that speed is significant to many and SSD have already outstripped the speed of common interfaces(SATA). The industry needs some leadership at the moment and this is where Apple can have considerable impact. They need to either define or adopt a storage format that is PCI Express based and on a compact PC card. This to get rid of the legacy interface and more importantly the mechanical form factor.
We have gone past the era of Mhz and CPU performance. Netbook, as many pundit pointed out are showing how small and comparatively slow computers are enough for everyday computer usage.
We have most of our Mhz and CPU speed we need, we then had all our memory capacity filled with cheap DDR2 Ram. Both component will continue to increase in speed and capacity as well as lowering cost following Moore's law.
For the most part all of the above is crap. The speed of modern PCs is not good enough, nor is the physical size or power disapation. Also don't assume that we can continue to benefit from Moores law with regards to DRAM, we are very much reaching real physical limits with current DRAM tech.
HDD, on the other hand, haven't had all the speed improvement compare to other parts of the system. While their capacity have seriously increased over the past years. ( Especially over the past 2 years, jumping from 500Gb to 2TB ) Their sequential speed may only have > 50% improvement. And Random read and write will be much less then that as well.
Yes but mechanical hard drives will be around longer than Flash based drives. The reality is I don't see much of a future in Flash as there is much more interesting and suitable tech. Right now, with flash using leading edge semiconductor processes, they are barely economical for storage devices with one quater the capacity of magnetic drives. There is already concerns in the industry that flash has hit the wall.
@ Vinea
I never said Apple should build their own controller chip. All i said was buy some controller chip and snap some Flash to make a SSD. This way apple uterlise their cheap Flash contract price.If Samsung has a decent controller chip the may as well buy the whole package from Samsung anyway. Since they are already buying all their Flash from Samsung.
The best thing Apple can do is to define a new electrical interface to a new format storage "board". Done right such a board standard might not be much thicker than four millimeters. This would be a big packaging advantage. Implement the latest PCI interface and it might have a chance in keeping up with the coming tech.
It is not an issue of cost but rather Apple needs to shake up the industry with respect to secondary storage. Now should Apple produce their own cards, probably not. What they need to do is to partner with the storage industry to move significantly beyond SATA and the old mechanical formats.
Dave
... There are many reasons to consider SSD, speed is just one of them. For many the durability in portable applications is more important than speed.
The reality here is that speed is significant to many and SSD have already outstripped the speed of common interfaces(SATA). The industry needs some leadership at the moment and this is where Apple can have considerable impact. ...
For the most part all of the above is crap. The speed of modern PCs is not good enough, nor is the physical size or power disapation. ...
Yes but mechanical hard drives will be around longer than Flash based drives. ...
The best thing Apple can do is to define a new electrical interface to a new format storage "board". ...
It is not an issue of cost but rather Apple needs to shake up the industry with respect to secondary storage. ...
You appear to be defending no technology that we can have included in any computer that we may purchase today or that we can buy from a third-party tomorrow. The rest of us, unfortunately, are limited to products that are actually on sale.
For now and for some time to come, the SSDs that we can buy are Flash drives. As such they have all of the limitations of Flash drives--expensive storage with questionable lifespan. For this reason, when I purchased my 17" MacBook Pro, I went with the 320 GB HDD over the 250 GB SSD. I was not going to spend an additional $1000 for 70 GB less storage, especially when I have no idea low long the SSD drive will last.
It bears remembering that Flash drives were not the first SSDs. SSDs have been around for nearly 30 years. Flash is just the first SSD technology to achieve significant market penetration. It is disingenuous to bash Flash while hyping SSD because there is no competing SSD technology on the horizon to take Flash's place.
You appear to be defending no technology that we can have included in any computer that we may purchase today or that we can buy from a third-party tomorrow. The rest of us, unfortunately, are limited to products that are actually on sale.
For now and for some time to come, the SSDs that we can buy are Flash drives. As such they have all of the limitations of Flash drives--expensive storage with questionable lifespan. For this reason, when I purchased my 17" MacBook Pro, I went with the 320 GB HDD over the 250 GB SSD. I was not going to spend an additional $1000 for 70 GB less storage, especially when I have no idea low long the SSD drive will last.
It bears remembering that Flash drives were not the first SSDs. SSDs have been around for nearly 30 years. Flash is just the first SSD technology to achieve significant market penetration. It is disingenuous to bash Flash while hyping SSD because there is no competing SSD technology on the horizon to take Flash's place.
I'm not sure how you came to the conclusion that I was bashing flash drives. Even after rereading my post I don't see that element in my post. I do however detect a bit of bashing in your post with respect to cost. I tried to actually show that flash drives actually have advantages beyond speed.
In anyevent it is important to take all of flashes negatives and positives into account before dismissing the technology. Especially in laptops where many magnetic disks seem to have an especially short life. I don't expect flash drives to become cheaper than magnetic drives on a per bit basis anytime soon, depending on the usage though they may end up being very cost effective.
Also it is not bashing flash to say that you expect flash based SSD to have a short life time on the market relative to magnetic disk drives. It is a reality of the technology and nothing more. Yeah I would like to see Samsung or somebody make gigantic strides in capacity and performance but they are already using bleeding edge production technology and still can't deliver cheap SSD. That simply leaves the door open to other tech. The fact that you can't buy that tech now means nothing if you are not willing to pay for flash drives at current prices & capacities.
Dave
I'm not sure how you came to the conclusion that I was bashing flash drives.
That's now too hard to find:
... The reality is I don't see much of a future in Flash as there is much more interesting and suitable tech. Right now, with flash using leading edge semiconductor processes, they are barely economical for storage devices with one quater the capacity of magnetic drives. There is already concerns in the industry that flash has hit the wall.
...
...
Also it is not bashing flash to say that you expect flash based SSD to have a short life time on the market relative to magnetic disk drives. ...
You are being too cute by half. There is nothing profound in saying that technology changes. Technology changes. Everybody expects technology to change. What next? The stunning insight that the sky is blue?
You are not the first person who tries to take two sides of the same issue. My position on Flash drives is unambiguous:
You illuminated the flaw in the logic. The goal is [presumably] to increase system performance. Adding 1 GHz to a core will increase system performance, but not as much as the increased clock speed would suggest to the naïve. A rule of thumb is that actual performance scales as the square root of the core clock speed. Going from 2 GHz to 3 GHz can be expected to improve overall performance by roughly 22%. To do better, many other engineering enhancements are required. Switching to SSD increases system performance, but again not nearly as much are the the faster component might naïvely suggest. Benchmarks suggest that SSDs in Macs roughly double overall system performance. Real world performance improvements are probably less. If the 100% performance improvement is real, then it is welcome. My rule of thumb, however, is that performance must triple before the improvement becomes obvious.
Sounds like you know a little about how audio perception works.
That's now too hard to find:
Again I don't know where you see this bashing of flash. I'm here actually supporting the idea that Apple needs to get involved in this business. Not so much to go into production but rather to define new mechanical standards and interface standards for SSD. The big problem with current SSD is that they sit in the same mechanical format as the old magnetic drives.
However I'm being forward looking here when I suggest that flash won't be around as long as magnetic drives. There is no bashing of flash there, rather it just acknowledges reality. Ideally any storage card format defined today would still be usable five years down the road when the alternatives are on the market.
You are being too cute by half. There is nothing profound in saying that technology changes. Technology changes. Everybody expects technology to change. What next? The stunning insight that the sky is blue?
Well hopefully something more insightful than the above.
In the context of this thread recognizing change is one of the reasons I'm putting forth that would justify Apple involmemt in the production of SSD. I'm not impling making the drives but rather defining new formats that reflect that we are dealling with solid state electronics here and not old rotaing media. It is pretty obvious that the storage industry isn't acknowledging the transition so it is up to Apple to do so.
Why you may ask. Simple really to acknowledge that their product line up is changing and that the need new solutions for incorporation into their products. The AIR is a good example of a machine that could use a thinner higher capacity drive. Possibly the new tablet and iPod devices too. The old format "drives" simply constrain design possibilities in unacceptable ways.
I'm more concerned that Apple keep change moving forward in positive ways.
You are not the first person who tries to take two sides of the same issue. My position on Flash drives is unambiguous:
Your position is rediculous if you ask me. I'm not taking both sides to argue but rather to show that not everybody has the same perceptions as you do or even myself.
- That is from your point of view. Many would see a flash drive as a bargain for any number of reasons.
Quote:
- Flash has a finite usable lifespan.
Quote:
- Its faster speeds, lighter weight, and lower power requirements do out overcome its negatives.
In the end Apple needs deeper involment in SSD standards. I don't think they need to get inti production themselves though if they can get enough manufactures on board. The problem here is the lack of change, something Apple can have a positive impact on.True, but for many users it means absolutely nothing. Further flash based drives will outlast mechanical drives in certain applications due simply to the lack of mechanical parts.
They most certainly do for some people or the sales of these devices would be tanking.
Dave
I wish the hybrid drives were cheaper. Dunno why they're so danged expensive.
Well, one thing he's wrong on is that the cost will remain high. The Flash folks are talking HDD parity sometime soonish. ....
It is not clear whom you are talking about, but I can't find an instance where I said Flash would remain high--until now. It bears reminding everyone that HDDs have seen significant price drops in recent years. I have no idea what competitive means. For Flash to reach the price point of even a super premium HDD, its price would have to drop by least 75% while HDD held steady. Could it happen? Maybe. Is it likely to happen? No.