Why in the hell would they use a Sata interface to the drive??? PCI-e anyone?
It's much more compatible to use SATA. Think of Disk Utility, partitioning tools, and Bootcamp. As long as they use 6Gbps SATA, they'll get up to 750MB/s throughput. The use of SATA 6G isn't a certainty though as the chipsets only have 2x 6G ports and 4x 3G ports.
but maybe there just isn't room on the stickers. They could be using some patented processes from Anobit to help drive prices down and if this is true, it should be seen in the laptop line too. Cutting prices in half would make them competitive with the cheapest SSD drives that are available right now and would allow them to sell a 2TB SSD for about $1200. I don't think I'd take a $1200 SSD with a 1 year warranty though. Samsung SSDs have a 3 year warranty so I'd just put two 512GB (1TB when they come out) in a 3.5" adaptor.
Sorry, but only a fool chooses PCI Express for the theoretical bandwidth expansion while eating it hard on actual IOPS and costs.
I merely choose PCIe because I thought I might be using the 2nd ODD in the future, and all 4 HDD bays were already in use. Besides, PCIe card weighs less than a SSD ¡
OWC has a 1TB SSD with PCI for $1500 or so. I imagine that AAPL would take this approach and use spinning drives for additional storage. There is a need for compatibility with older storage/optical devices. I push my MacPro pretty hard but I am not a professional, so I need the option of lower cost. My 6-core Xeon can process a BR disk, convert to iPad format, edit pics in PS, and design in AI, without missing a beat. I wouldn't try that with an iMac.
Read... Mac PRO. It's not about the cost, it's about offering high performance at any cost to the pro market. This isn't about appeasing the iPad or iMac user. I'm surprised Apple still cares about their pro market. They have let it sit for far too long.
When I first read this I was thinking "Why the hell wouldn't they just use a bunch of mSATA sockets on the new Pro motherboards?", but if you consider that the largest mSATA capacity is only 256GB, then a 3.5" form factor starts to make more sense. Pro users need large swaths of contiguous space, so a bunch of 256GB drives won't do. Apple really does need to create a single big 2TB SSD to make such users happy.
The only reason why this would make sense is maintaining the Mac Pro pricing policy in a level high enough where desktop users cannot afford it and only a "niche" can buy it. Then call the Mac a niche market.
Last year Apple said the fusion drive was released because SSD was expensive. However, reality is that you can buy 512GB SSD drives at really affordable prices that any home user can afford.
Maybe, but if it's half the average sell price of a PC they won't necessarily go for it. The lowest price on Newegg for a 512GB is $380. The average sell price for a PC is less than $800.
I feel they just realized they cannot keep current prices of the 512GB SSDs for the Mac Pro, so they need to introduce bigger units in order to keep same prices, where margins are high.
Apple doesn't get what's the problem on the Mac line. Margins are a compromise that must be carefully driven. The Mac product line cannot be designed from a bare margins strategy, but from demand.
If you are rehashing your argument for an Apple consumer tower desktop, then you need to argue there's a notable demand for an *Apple* consumer tower desktop. It's also a diminishing market category. Then you have to address the fact that the enthusiast tower market is mostly people that play games, which there wasn't a good Mac game market for decades, and hackers, which don't like Apple's closed nature to any degree.
I don't know about 2TB SSDs. I can see Fusion drives since they are as fast as an SSD for a little more than HDD. 2TB drives would cost over $2,000. Right? 1TB Fusion drives cost about $450.
Updated to adjust for 2TB drives.
+1 figure a 2TB fusion drive for $900 or less (by release date).
Only the biggest of the big data pounders would want performance/price of a true SSD (100% random access across multi-Terabytes of data with ms access).
Now spinning the 1/2 grain of truth a slightly different direction....
.... Large SSDs can decouple highly volatile (many write many read) active cashes from RAMdisk and to the storage fabric for a less complex shared store.
.... As you move in to millions of queries a second over the interwebs for session verifications, having a couple RAIDed nTB SSDs at each cloud instance (for apple a couple per country... which would be a few hundred to near a 1000),
.... I could see a market for Apple to have several thousand 2TB drives as hubs for each iCloud instance tracking all the iOS and AppleID interneted 'call home to momma' calls.
.... Therefore It wouldn't surprise me then if Apple was building nTB SSDs for themselves for their shared iCloud offerings (a multi-terabyte hash table, say of all 'good' Siri queries, or all DRM signatures, session tokens, or whatever).
Think about it... Every time I click on an iCloud item, there are probably 3 if not more signature hashes being verified, and if I'm sharing these to the masses (extend Discretionary Access Control [ACLs] to anyone), each atomic item has to be authenticated, authorized, and decrypted from storage, reencrypted for transport to the receiver. All this key management needs to support millions of transactions per second... seems to me a massive hash of AES keys is the way to go.
Even Siri, where a hash of the voice transmission "Find me the nearest Starbucks" could be computed by the iCloud mothership, which would allow for 'reuse' of past successful queiries (get voice sample/FFT, hash it, compare to my local hash table, if a match, try that, on fail, 'translate' to natural language, hash the translation, look that up on the global 'text' query table. Lots of lookups of lookups of lookups... saving milliseconds will save 10th/s of seconds eventually and my guess is 10ths of seconds speed up equate to large % of 'satisfaction')
Just sayin'... the fact may be that Apple is building massive SSDs... just not for their consumer products.
+1 figure a 2TB fusion drive for $900 or less (by release date).
A 3.1TB Fusion Drive the new iMac are already much cheaper than that. From the 1TB default it's $150 for a 3TB HDD upgrade or $250 — an extra $100 for the 128GB SSD — for the 3.1TB FD.
I fully expect FD to be an option as it does afford those who want a lot of storage capacity the option of also having fast (overall) disk speeds but I also think they will offer large SSD solutions for a lot of money simply because many Mac Pro users care about performance more than anything.
I have a friend who always buys the newest, high-end Mac Pro. He uses it for work so every second saved is money made. I'm not sure if they gets the value from it each time he buys but he thinks he does.
The 27" iMac also has a 768GB SSD option for $750 so I can see a 2TB option being offered either on a card, as a 3.5' option, or both. I wonder how RAID 0 across 5 SSDs would affect speed. I assume it would be the same effect as spreading the load across 5 HDDs.
As long as you keep data on HDD and put the OS + Apps on SSD I'm failing to see the point of a 2TB SSD. But times are changing, and if we 'enter the UHD / 4K video era' we might need larger storage solutions, not fater throughput. The 48 minute docu TimeScapes is 330GB, thus 117MB/s video so current 4TB HDD can only hold 12 of those 4K files.
Apple doesn't use their own hardware for their datacenters. Heck, they don't use their own software; it's all supplied by Microsoft / Amazon / Oracle et cetera
I think you are assuming Apple is software biased when you say 'all.' The beauty of a cloud environment and some truly massive data centers where you own(lease) the rackspace is that you can put your own hardware solutions into particular layers of your service offering (e.g. google).
Apple is not NetFlix (a software company, that sees computation bandwidth and content as 'commodity').
I see Apple building some 'secret sauce' components of their data centers, and the primary components being those that make their core mobile experiences better(faster) and secure.
Whether this rumor is true or not, high capacity solid state drives are the future. Such drives need to become commodity items and I expect Apple's push into the area is focused on that goal. IPad, iPhone, iPod, Apple TV, MBAir are solid state. MBPro will become solid state (and perhaps merge with MBAir?). Time Machine -- not likely -- would make little sense.
Where did this point of view come from? The Mac Pro is the only professionally oriented machine they have. Drop that and respect for the entire Mac Line goes down the tubes. It is no different than Ford and its pickup trucks, volume goes to the F150 but they do have machines, sold at a much lower volume, for professional use.
High performance computing is either deployed through commodity clusters or gpu clusters. This product has fallen in the void between those and a desktop that isn't very large.
Yep. The Next "Mac Pro" is going to be the new Mac Mini with "Optical Thunderbolt". That's it. Build as powerful a super-computer you desire.
I see Apple building some 'secret sauce' components of their data centers, and the primary components being those that make their core mobile experiences better(faster) and secure.
There's simply no way of knowing what Apple uses; I just watch these keynotes* and don't see any Apple hardware. On top of that, there are many articles to be found on what software and hardware Apple uses, although that is merely blog/rumor/analyst kinda stuff.
Yep. The Next "Mac Pro" is going to be the new Mac Mini with "Optical Thunderbolt". That's it. Build as powerful a super-computer you desire.
Huh? Optical Thunderbolt is just Thunderbolt. The connector is electrical, the transceiver is in the cable. The ends are electrical, it gets converted to optical, and back to electrical, a couple centimeters from each end.
Huh? Optical Thunderbolt is just Thunderbolt. The connector is electrical, the transceiver is in the cable. The ends are electrical, it gets converted to optical, and back to electrical, a couple centimeters from each end.
Exactly. The only differences are a 10x improvement in speed and an inability to power devices through it.
Well, it can be, which is the point. Optical is theoretical up to 100Gb/s both ways, yeah?
As long as all of the hardware is updated. Intel could provide Apple with an optical Thunderbolt controller and they'd have to connect it to 16 PCIe2 lanes and that one controller could be split between 4 ports. The GPU would be on 16 lanes so it would only leave 8 lanes unused. The receiving hardware needs to have optical controllers too though.
Comments
It's much more compatible to use SATA. Think of Disk Utility, partitioning tools, and Bootcamp. As long as they use 6Gbps SATA, they'll get up to 750MB/s throughput. The use of SATA 6G isn't a certainty though as the chipsets only have 2x 6G ports and 4x 3G ports.
Apple hasn't branded their custom SSD storage:
http://www.anandtech.com/show/6727/apple-is-using-sandisk-ssds-in-retina-macbook-pro-as-well
but maybe there just isn't room on the stickers. They could be using some patented processes from Anobit to help drive prices down and if this is true, it should be seen in the laptop line too. Cutting prices in half would make them competitive with the cheapest SSD drives that are available right now and would allow them to sell a 2TB SSD for about $1200. I don't think I'd take a $1200 SSD with a 1 year warranty though. Samsung SSDs have a 3 year warranty so I'd just put two 512GB (1TB when they come out) in a 3.5" adaptor.
I merely choose PCIe because I thought I might be using the 2nd ODD in the future, and all 4 HDD bays were already in use. Besides, PCIe card weighs less than a SSD ¡
When I first read this I was thinking "Why the hell wouldn't they just use a bunch of mSATA sockets on the new Pro motherboards?", but if you consider that the largest mSATA capacity is only 256GB, then a 3.5" form factor starts to make more sense. Pro users need large swaths of contiguous space, so a bunch of 256GB drives won't do. Apple really does need to create a single big 2TB SSD to make such users happy.
Maybe, but if it's half the average sell price of a PC they won't necessarily go for it. The lowest price on Newegg for a 512GB is $380. The average sell price for a PC is less than $800.
If you are rehashing your argument for an Apple consumer tower desktop, then you need to argue there's a notable demand for an *Apple* consumer tower desktop. It's also a diminishing market category. Then you have to address the fact that the enthusiast tower market is mostly people that play games, which there wasn't a good Mac game market for decades, and hackers, which don't like Apple's closed nature to any degree.
Quote:
Originally Posted by drblank
I don't know about 2TB SSDs. I can see Fusion drives since they are as fast as an SSD for a little more than HDD. 2TB drives would cost over $2,000. Right? 1TB Fusion drives cost about $450.
Updated to adjust for 2TB drives.
+1 figure a 2TB fusion drive for $900 or less (by release date).
Only the biggest of the big data pounders would want performance/price of a true SSD (100% random access across multi-Terabytes of data with ms access).
Now spinning the 1/2 grain of truth a slightly different direction....
.... Large SSDs can decouple highly volatile (many write many read) active cashes from RAMdisk and to the storage fabric for a less complex shared store.
.... As you move in to millions of queries a second over the interwebs for session verifications, having a couple RAIDed nTB SSDs at each cloud instance (for apple a couple per country... which would be a few hundred to near a 1000),
.... I could see a market for Apple to have several thousand 2TB drives as hubs for each iCloud instance tracking all the iOS and AppleID interneted 'call home to momma' calls.
.... Therefore It wouldn't surprise me then if Apple was building nTB SSDs for themselves for their shared iCloud offerings (a multi-terabyte hash table, say of all 'good' Siri queries, or all DRM signatures, session tokens, or whatever).
Think about it... Every time I click on an iCloud item, there are probably 3 if not more signature hashes being verified, and if I'm sharing these to the masses (extend Discretionary Access Control [ACLs] to anyone), each atomic item has to be authenticated, authorized, and decrypted from storage, reencrypted for transport to the receiver. All this key management needs to support millions of transactions per second... seems to me a massive hash of AES keys is the way to go.
Even Siri, where a hash of the voice transmission "Find me the nearest Starbucks" could be computed by the iCloud mothership, which would allow for 'reuse' of past successful queiries (get voice sample/FFT, hash it, compare to my local hash table, if a match, try that, on fail, 'translate' to natural language, hash the translation, look that up on the global 'text' query table. Lots of lookups of lookups of lookups... saving milliseconds will save 10th/s of seconds eventually and my guess is 10ths of seconds speed up equate to large % of 'satisfaction')
Just sayin'... the fact may be that Apple is building massive SSDs... just not for their consumer products.
A 3.1TB Fusion Drive the new iMac are already much cheaper than that. From the 1TB default it's $150 for a 3TB HDD upgrade or $250 — an extra $100 for the 128GB SSD — for the 3.1TB FD.
I fully expect FD to be an option as it does afford those who want a lot of storage capacity the option of also having fast (overall) disk speeds but I also think they will offer large SSD solutions for a lot of money simply because many Mac Pro users care about performance more than anything.
I have a friend who always buys the newest, high-end Mac Pro. He uses it for work so every second saved is money made. I'm not sure if they gets the value from it each time he buys but he thinks he does.
The 27" iMac also has a 768GB SSD option for $750 so I can see a 2TB option being offered either on a card, as a 3.5' option, or both. I wonder how RAID 0 across 5 SSDs would affect speed. I assume it would be the same effect as spreading the load across 5 HDDs.
Originally Posted by habi
Why in the hell would they use a Sata interface to the drive??? PCI-e anyone?
SATA's a better option.
When's SATA IV supposed to be released? Because SATA Express just won't cut it, will it?
Quote:
Originally Posted by PhilBoogie
As long as you keep data on HDD and put the OS + Apps on SSD I'm failing to see the point of a 2TB SSD. But times are changing, and if we 'enter the UHD / 4K video era' we might need larger storage solutions, not fater throughput. The 48 minute docu TimeScapes is 330GB, thus 117MB/s video so current 4TB HDD can only hold 12 of those 4K files.
Apple doesn't use their own hardware for their datacenters. Heck, they don't use their own software; it's all supplied by Microsoft / Amazon / Oracle et cetera
I think you are assuming Apple is software biased when you say 'all.' The beauty of a cloud environment and some truly massive data centers where you own(lease) the rackspace is that you can put your own hardware solutions into particular layers of your service offering (e.g. google).
Apple is not NetFlix (a software company, that sees computation bandwidth and content as 'commodity').
I see Apple building some 'secret sauce' components of their data centers, and the primary components being those that make their core mobile experiences better(faster) and secure.
Whether this rumor is true or not, high capacity solid state drives are the future. Such drives need to become commodity items and I expect Apple's push into the area is focused on that goal. IPad, iPhone, iPod, Apple TV, MBAir are solid state. MBPro will become solid state (and perhaps merge with MBAir?). Time Machine -- not likely -- would make little sense.
Quote:
Originally Posted by ifij775
Quote:
Originally Posted by wizard69
Where did this point of view come from? The Mac Pro is the only professionally oriented machine they have. Drop that and respect for the entire Mac Line goes down the tubes. It is no different than Ford and its pickup trucks, volume goes to the F150 but they do have machines, sold at a much lower volume, for professional use.
High performance computing is either deployed through commodity clusters or gpu clusters. This product has fallen in the void between those and a desktop that isn't very large.
Yep. The Next "Mac Pro" is going to be the new Mac Mini with "Optical Thunderbolt". That's it. Build as powerful a super-computer you desire.
Quote:
Originally Posted by mdriftmeyer
The SATA III bandwidth is 6Gbits/sec. No SSD drive is ever saturating that bus.
This up-to 1GB/sec PCI Express Card from Intel that is a 400GB SSD Drive:
6Gb/s = 750MB/s which could be considered the theoretical maximum and not likely attainable in the real world.
1GB/s = 1000MB/s which is a little more than 750MB/s.
And that is just the write speed. The Intel 910 series can read at 2000MB/s
Right?
http://www.intel.com/content/www/us/en/solid-state-drives/solid-state-drives-ssd.html
There's simply no way of knowing what Apple uses; I just watch these keynotes* and don't see any Apple hardware. On top of that, there are many articles to be found on what software and hardware Apple uses, although that is merely blog/rumor/analyst kinda stuff.
* 2011 WWDC Keynote screendump
Huh? Optical Thunderbolt is just Thunderbolt. The connector is electrical, the transceiver is in the cable. The ends are electrical, it gets converted to optical, and back to electrical, a couple centimeters from each end.
Originally Posted by JeffDM
Huh? Optical Thunderbolt is just Thunderbolt. The connector is electrical, the transceiver is in the cable. The ends are electrical, it gets converted to optical, and back to electrical, a couple centimeters from each end.
Exactly. The only differences are a 10x improvement in speed and an inability to power devices through it.
10x improvement in distance, but not any faster.
Ditch the optical bays on a current Mac Pro and you can stick in 6 X 4Tb drives, 24Tb is a handy amount of online video content.
PCI does seem like the way to go for a super fast boot drive though, which generally doesn't need to be huge capacity.
Originally Posted by JeffDM
10x improvement in distance, but not any faster.
Well, it can be, which is the point. Optical is theoretical up to 100Gb/s both ways, yeah?
As long as all of the hardware is updated. Intel could provide Apple with an optical Thunderbolt controller and they'd have to connect it to 16 PCIe2 lanes and that one controller could be split between 4 ports. The GPU would be on 16 lanes so it would only leave 8 lanes unused. The receiving hardware needs to have optical controllers too though.