How the hell do Apple expect to get this patent?! I mean the specific lightning connector diagram fair enough but the rest is such a generic concept and design. Intel and others must have so much prior art with regard to this. I wish Apple would stop getting silly patents just for courts to invalidate them later.
What would be the benefit of doing that when the internal NAND is likely to do 25MB/s? Bragging rights?
So how is it that an SSD will deliver files so fast I get them before I ask for them, but the storage on an iPhone can't even manage USB2 speeds? Aren't both using NAND?
So how is it that an SSD will deliver files so fast I get them before I ask for them, but the storage on an iPhone can't even manage USB2 speeds? Aren't both using NAND?
Do you know how RAIDed HDDs work? Do you know how, say, RAID 0 works? That RAID offers no redundancy but does offer speed improvement by writing/reading to/from multiple disks. To put it in simple terms that's what the SSD controller does. It controls the data being written and read across multiple NAND chips, thus increasing the relative performance.
There is no SSD controller in the iPhone. It's just the NAND right on the board, and I think it's just the one chip.
So how is it that an SSD will deliver files so fast I get them before I ask for them, but the storage on an iPhone can't even manage USB2 speeds? Aren't both using NAND?
Digging up a post I made back in February (I started designing with Flash memory back in the early 1990s) ....
Apple uses the least expensive type of NAND. Which makes sense. Smartphone or tablet storage hardly requires enterprise grade speed or robustness, and users do not read/write data that often.
(Which is why the Apple chips' limit of around 5,000 erase-write cycles is plenty... and now you also know why Safari doesn't save its cache to disk (last I looked). It only uses available RAM, which is why the info in a tab often has to be re-downloaded.)
The type of flash memory used in the iPhone could have anywhere from about 20 to 60 MB / sec max read speed per die. There are file speed tests out on the 'net.
SSDs, on the other hand often have:
better memory that's been tested with faster (and/or more) write cycles.
multiple memory chips for faster parallel access.
better flash controller chips with delayed writes, fancy wear leveling, etc.
RAM buffers so that slow writes can take place later.
Supposedly Apple recently started using a custom flash controller related to their acquisition of an Israeli engineering group. This controller can theoretically do 600MB/s reads... if the device had enough parallel memory chips, that is. AFAIK, there's no such setup in an iOS device yet.
This controller can theoretically do 600MB/s reads... if the device had enough parallel memory chips, that is. AFAIK, there's no such setup in an iOS device yet.
There is cost, power and size issues to deal with for a regular SSD design but I can see many paths for Apple creating a specialized controller that would best be described as a Flash/SSD-hydrid for iDevices.
I hope something is done soon because like HDDs in PCs these NAND chips are performance bottlenecks.
For the record, I invented a hybrid optical/inductive coupling system - in 1991, and I published it through a document proposing a universal storage media standard. Being I am now a company director, for legal reasons, I cannot say more, other than the fact, I look forward to challenging Apple's patent on the grounds of prior art.
There is cost, power and size issues to deal with for a regular SSD design but I can see many paths for Apple creating a specialized controller that would best be described as a Flash/SSD-hydrid for iDevices.
As you probably know, larger storage NAND Flash chips are internally made up of multiple stacked smaller memory die.
For example, a 64GB chip might have four 16GB die inside it.
Apple could eventually combine their purchased Flash controller technology and some RAM into a chip package along with internal parallel access to multiple memory slices, and end up with some very SSD-like performance increases.
The likely downside would be greatly increased power usage, which is probably what's holding them back.
I tell ya, I can't wait for someone to make a real battery breakthrough. It would open up all sorts of possibilities.
Apple could eventually combine their purchased Flash controller technology and some RAM into a chip package along with internal parallel access to multiple memory slices, and end up with some very SSD-like performance increases.
Good point. Is there any evidence that there's just one controller in the iPhone/iPad? I can't find it at Ars, AT, Wiki et cetera...
As you probably know, larger storage NAND Flash chips are internally made up of multiple stacked smaller memory die.
For example, a 64GB chip might have four 16GB die inside it.
Apple could eventually combine their purchased Flash controller technology and some RAM into a chip package along with internal parallel access to multiple memory slices, and end up with some very SSD-like performance increases.
The likely downside would be greatly increased power usage, which is probably what's holding them back.
I tell ya, I can't wait for someone to make a real battery breakthrough. It would open up all sorts of possibilities.
That's what I'm hoping thinking. Apple could build a PoP-like design that is essentially an SSD card built into a single unified chip. This would be more costly than simply having, say, 64GB NAND but the benefits of smaller 4x16GB, with a requisite controller and RAM cache would likely be easily folded into the device. Considering Apple's efficiency along the entire part of the chain the benefits would appear to outweigh the cons by a considerably margin.
This then opens up Lightning to be able to read/write much faster than it currently does over USB.
Too bad this discussion is over. An optical signal out of an iPod or iPhone is clearly needed for Apple's dual-channel (left/right) laser-projection stereo video glasses. They could end up using two 4K video streams, though I always thought it would be 2K per channel.
If anybody cares, I'll look up their ancient patent filings for these glasses.
Comments
Quote:
Originally Posted by SolipsismX
What would be the benefit of doing that when the internal NAND is likely to do 25MB/s? Bragging rights?
So how is it that an SSD will deliver files so fast I get them before I ask for them, but the storage on an iPhone can't even manage USB2 speeds? Aren't both using NAND?
Do you know how RAIDed HDDs work? Do you know how, say, RAID 0 works? That RAID offers no redundancy but does offer speed improvement by writing/reading to/from multiple disks. To put it in simple terms that's what the SSD controller does. It controls the data being written and read across multiple NAND chips, thus increasing the relative performance.
There is no SSD controller in the iPhone. It's just the NAND right on the board, and I think it's just the one chip.
Quote:
Originally Posted by v5v
So how is it that an SSD will deliver files so fast I get them before I ask for them, but the storage on an iPhone can't even manage USB2 speeds? Aren't both using NAND?
Digging up a post I made back in February (I started designing with Flash memory back in the early 1990s) ....
Apple uses the least expensive type of NAND. Which makes sense. Smartphone or tablet storage hardly requires enterprise grade speed or robustness, and users do not read/write data that often.
(Which is why the Apple chips' limit of around 5,000 erase-write cycles is plenty... and now you also know why Safari doesn't save its cache to disk (last I looked). It only uses available RAM, which is why the info in a tab often has to be re-downloaded.)
The type of flash memory used in the iPhone could have anywhere from about 20 to 60 MB / sec max read speed per die. There are file speed tests out on the 'net.
SSDs, on the other hand often have:
better memory that's been tested with faster (and/or more) write cycles.
multiple memory chips for faster parallel access.
better flash controller chips with delayed writes, fancy wear leveling, etc.
RAM buffers so that slow writes can take place later.
Supposedly Apple recently started using a custom flash controller related to their acquisition of an Israeli engineering group. This controller can theoretically do 600MB/s reads... if the device had enough parallel memory chips, that is. AFAIK, there's no such setup in an iOS device yet.
Quote:
Originally Posted by Urahara
Could have just used Thunderbolt connection.
Disappointing that they made an update to USB connection last year but didn't do a faster Thunderbolt connection.
Intel just announced a 20 megabit thunderbolt upgrade Last week with the introduction of a new control processor. It will be available this fall.
http://www.engadget.com/2013/04/08/intel-announces-next-gen-thunderbolt-20-gbps-throughput/
Specifically it will be able to handle 4k video streams.
Quote:
Originally Posted by SolipsismX
Do you know how RAIDed HDDs work? [snip]
Gotcha. Thanks!
Quote:
Originally Posted by KDarling
Digging up a post I made back in February (I started designing with Flash memory back in the early 1990s) ....
[Useful and interesting info snipped]
Cool beans. Interesting stuff about cache writes. Thanks!
There is cost, power and size issues to deal with for a regular SSD design but I can see many paths for Apple creating a specialized controller that would best be described as a Flash/SSD-hydrid for iDevices.
I hope something is done soon because like HDDs in PCs these NAND chips are performance bottlenecks.
That would be 20 gigabit, just like your link says.
Quote:
Originally Posted by SolipsismX
There is cost, power and size issues to deal with for a regular SSD design but I can see many paths for Apple creating a specialized controller that would best be described as a Flash/SSD-hydrid for iDevices.
As you probably know, larger storage NAND Flash chips are internally made up of multiple stacked smaller memory die.
For example, a 64GB chip might have four 16GB die inside it.
Apple could eventually combine their purchased Flash controller technology and some RAM into a chip package along with internal parallel access to multiple memory slices, and end up with some very SSD-like performance increases.
The likely downside would be greatly increased power usage, which is probably what's holding them back.
I tell ya, I can't wait for someone to make a real battery breakthrough. It would open up all sorts of possibilities.
Good point. Is there any evidence that there's just one controller in the iPhone/iPad? I can't find it at Ars, AT, Wiki et cetera...
That's what I'm
hopingthinking. Apple could build a PoP-like design that is essentially an SSD card built into a single unified chip. This would be more costly than simply having, say, 64GB NAND but the benefits of smaller 4x16GB, with a requisite controller and RAM cache would likely be easily folded into the device. Considering Apple's efficiency along the entire part of the chain the benefits would appear to outweigh the cons by a considerably margin.This then opens up Lightning to be able to read/write much faster than it currently does over USB.
The naked single NAND chip on the board and reported speeds prove the simplicity of the setup.
If anybody cares, I'll look up their ancient patent filings for these glasses.