Unlike Thunderbolt, USB 3 will be on every netbook, notebook, desktop, and server in the coming months. Thunderbolt will only be available on higher end products.
USB 3 will become the standard and Thunderbolt will become the next FireWire...
TB will be on virtually every computer with an Intel CPU. Apparently you don't understand what native support means, all a manufacturer has to do to add it is provide the (royalty free) plug.
TB will be on virtually every computer with an Intel CPU. Apparently you don't understand what native support means, all a manufacturer has to do to add it is provide the (royalty free) plug.
I don't believe it, but I guess we will have to wait and see the outcome.
1) Thunderbolt will supposedly be more than twice as fast as USB 3.0. USB, because of its design, apparently, never comes close to its theoretical throughput. This was a big deal in the FW/USB2.0 comparisons, where the USB2.0 peak was higher, but FW400 provided a much faster sustained connection.
2) TB does the capability to provide backwards compatibility with USB 2.0, but the rumor was that the USB groups asked them not to add this in. That would have been the nail in USB3.0's coffin.
However, USB has a pretty strong brand name, and that might be enough to carry it through, and TB might get relegated to Professional uses, like FW was. Lets see if Intel is able to push it enough.
The USB standard was mean to replace the RS-232 and the parallels port in the mid 90, It was a build on simple controler-host topology without any DMA. While the protocol is really simple to boost by speed up the signalling, because of the lack of DMA on USB, reading or writing on USB device make great impact on the CPU. More you boost the bandwidth on USB more you taxes your CPU, making USB inapt for high bandwidth and processing solution like video editing. This is the main reason that explain why FW always been superior to USB for storage.
The new ThunderBolt protocol is a PCI Express 4x external bus, since it pass PCI packet and Intel controller is acting like an PCIExpress-TB bridge with repeater for daisy-chaining device, you virtually can create any kind of external TB adapter like USB, SCSI, Fiber Channel, Ethernet based on existing PCI Express card. ThunderBolt doesn't need to be backward compatible with any port because by design you can adapt any type of port on a PCI bus like Thunderbolt is.
i'm sick of buying adapters and not being able to plug in old devices when they're needed - that's why TB will take a long time to gain marketshare. the displayport is still hardly anywhere, but the USB port is everywhere. i don't think we'll kill USB until 802.11x and NFC are fast enough for everything to forgo cables with different connectors.
Can't wait to boot up my external SSD drive using TB as my boot up hard drive and just use my internal hard drive for storage. Oh shoot. That means I have to buy the upcoming iMac with TB. Who wants a 27inch iMac, 2010 model?
The USB standard was mean to replace the RS-232 and the parallels port in the mid 90, It was a build on simple controler-host topology without any DMA. While the protocol is really simple to boost by speed up the signalling, because of the lack of DMA on USB, reading or writing on USB device make great impact on the CPU. More you boost the bandwidth on USB more you taxes your CPU, making USB inapt for high bandwidth and processing solution like video editing. This is the main reason that explain why FW always been superior to USB for storage.
Wherever did you get the idea that USB doesn't support DMA? Having worked on many USB chips and device drivers, I can assure you that even the cheapest USB chips found in PCs support DMA.
That is not to say that there isn't a *tremendous* overhead in the USB stack. But this is a software issue, not a result of the lack of DMA.
Quote:
The new ThunderBolt protocol is a PCI Express 4x external bus, since it pass PCI packet and Intel controller is acting like an PCIExpress-TB bridge with repeater for daisy-chaining device, you virtually can create any kind of external TB adapter like USB, SCSI, Fiber Channel, Ethernet based on existing PCI Express card. ThunderBolt doesn't need to be backward compatible with any port because by design you can adapt any type of port on a PCI bus like Thunderbolt is.
It will be interesting to see what the software overhead is for Thunderbolt. Not saying it will be any different from PCIe -- I'm saying I don't know and we'll have to see before making major proclamations about its performance.
USB does pretty good with one device attached. When start attaching a half dozen, the software stack really starts to chew up time. Hopefully, this won't be true for Thunderbolt.
"Though Intel has said it plans to support USB 3.0 with its next-generation chipsets, some in the PC market believe Apple's adoption of Thunderbolt could have a major impact."
Why? Nobody pays attention to what Apple does. Nobody copies what Apple is doing. Nobody takes their lead from Apple... oh wait.
i'm sick of buying adapters and not being able to plug in old devices when they're needed - that's why TB will take a long time to gain marketshare. the displayport is still hardly anywhere, but the USB port is everywhere. i don't think we'll kill USB until 802.11x and NFC are fast enough for everything to forgo cables with different connectors.
I'm sick of all those adapters too. On a brightest side, the TB port is really a great opportunity for docking station like monitor, I predict the next Apple standalone display will have some port replicator (Ethernet, USB, FW, Sound) connect to the mac thru one TB connection. But beside adaptor and port replicator, by the way of using the PCI protocol on TB, it's really easy for hardware developer to adapt their existing stuff on TB like video grabbing PCI card or SAS Raid unit.
It's ridiculous to even talk about this. As I've said from the beginning, Apple will support USB 3.0 when Intel supports it and we now know that will happen with IvyBridge.
What many don't realize here or can't seem to grasp is that every shipping system with a USB 3 port to date has been shipping with a PCI Express to USB bridge chip to give yoiu those USB ports. Generally those chips support 2-3 USB ports over a single PCI Express lane thus there is no way to ever get full performance across all of the USB 3 ports implemented!
Quote:
Macs don't have USB 3.0 because Apple would have to add an additional chip to support it. Once Intel natively supports USB 3.0, Apple would have to go through a similar level of effort to avoid supporting it. There's no reason for Apple to do that.
Nothing to do with "acting monopolistic".
Exactly. They would have to use chips to implement USB 3 that up until late last year where power hungery and buggy. Beyond that the bridge chips could never reach USB 3's communications rates across all ports. So really there has been very little incentive for Apple to implement what has been beta hardware to be charitiable.
There is also nothing monopolistic about waiting for technology to mature or stabilize.
As for all the glowing reports from people with their new USB 3 drivesd, that is all fine and good but disk drives, at least magnetic ones, won't be saturating USB3 anyways. It is easy to see great results when your basis for comparison is USB 2, but people need to realize is that TB is a forward looking port. Nobody concerned with TB gives a rats a$$ about how fast a legacy notebook drive dongle works on TB. Yeah the port is faster but that isn't why the port exists by itself, rather it is a forward looking way to handle external communications for future products.
Thus when is see these discussions about USB vs TB the only thing I can think of is man do these people not get it. TB is not a replacement for USB, so any discussion that revolves around TB replacing USB is misplaced at best.
Unlike Thunderbolt, USB 3 will be on every netbook, notebook, desktop, and server in the coming months. Thunderbolt will only be available on higher end products.
USB 3 will become the standard and Thunderbolt will become the next FireWire...
I keep seeing this sentiment but I don?t get it. It sounds like people are drawing comparison based on the speed of the port, but that isn?t a fair comparison. FW was a market failure compared to USB for factors that do not affect TB.
For starters it?s supported by Intel. Second, it not encumbered with the same excessive licensing fees FW was. Third, the mDP port now has more of a reason to catch on as it?s a svelte, free option that can be used for fast data and display.
I see no reason why USB3.0 and TB will not be widely popular.
Wherever did you get the idea that USB doesn't support DMA? Having worked on many USB chips and device drivers, I can assure you that even the cheapest USB chips found in PCs support DMA.
That is not to say that there isn't a *tremendous* overhead in the USB stack. But this is a software issue, not a result of the lack of DMA.
It will be interesting to see what the software overhead is for Thunderbolt. Not saying it will be any different from PCIe -- I'm saying I don't know and we'll have to see before making major proclamations about its performance.
USB does pretty good with one device attached. When start attaching a half dozen, the software stack really starts to chew up time. Hopefully, this won't be true for Thunderbolt.
You're right, I was wrong the USB controller card got DMA, but the USB device doesn't have access directly to the ram to push or read data directly like the FW. It's like TCP/IP, the software driver control each block making a lot of CPU overhead.
The main appeal of TB it's all hardware base, their is any software part of the communication process and doesn't add any layer on top of PCI. But it's still wise of you to wait and see in the real world with real product was is mean in term of real performance.
I'm assuming TB will allow bus-powered drives and charging for future iOS devices when sync'ed, though not sure I've seen this for sure. Seems necessary for it to be successfully in a broad way.
I'm assuming TB will allow bus-powered drives and charging for future iOS devices when sync'ed, though not sure I've seen this for sure. Seems necessary for it to be successfully in a broad way.
Yes, 10 Watts. More than USB but less than Firewire.
Quote:
Originally Posted by Wikipedia
Standard USB hub ports can provide from the typical 500 mA [2.5 W] of current, only 100 mA from non-hub ports. USB 3.0 and USB On-The-Go supply 1800 mA [9.0 W] (for dedicated battery charging, 1500 mA [7.5 W] Full bandwidth or 900 mA [4.5 W] High Bandwidth), while FireWire can in theory supply up to 60 watts of power, although 10 to 20 watts is more typical.
I think, but I?m not sure, Thunderbolt?s 10W is constant regardless of the connection type. Is this enough to connect a external 3.5? HDD without a supplemental power source? I sure hope so.
It's ridiculous to even talk about this. As I've said from the beginning, Apple will support USB 3.0 when Intel supports it and we now know that will happen with IvyBridge.
Macs don't have USB 3.0 because Apple would have to add an additional chip to support it. Once Intel natively supports USB 3.0, Apple would have to go through a similar level of effort to avoid supporting it. There's no reason for Apple to do that.
Nothing to do with "acting monopolistic".
Well, Sony, Dell, toshiba and many others have added the chip with their new sandy bridge 2 computers to allow USB 3.
I stand by my thoughts that apple can and ought to support USB 3 now. Will those iMacs coming in via boats from china right now have USB 3? Probably not.
So if they aren't monopolistic based on hardware ... Jobs penchant for hating blu-ray, hating adobe's flash. Or as evidenced by the recent closed wall approach to iPad, iPhone and now the mac app store. Where they are demanding $ from any company selling services/content on iOS devices... And blocking those who won't pay the vig fee...
Apple loves to keep their overall costs down. It gives them higher margins. Then they love to close their os and charge others 30% to do business there.
I love apple products, but kind miss the old apple which was open more and... Well even if the products cost more... Offer more. You know... Offer USB 3 and blu-ray when others are
Sorry but I ain't the only person noticing the recent years less than nice and cuddly approach of apple.
Thunderbolt is already twice as fast as USB 3.0. TB has a theoretical 10Gb/s throughput, while USB 3.0 maxes out at 5.0 Gb/s. But that's just the beginning for Thunderbolt.
When costs come down enough, it will be possible to use optical buses for another 10x boost in speed, or up to 100Gb/s. And optical Thunderbolt should allow cable runs of up to 100 meters. I doubt any copper-based technology could get anywhere near that performance level. Any future USB 4.0 (if there is one) will need to go optical to compete.
Thunderbolt is already twice as fast as USB 3.0. TB has a theoretical 10Gb/s throughput, while USB 3.0 maxes out at 5.0 Gb/s. But that's just the beginning for Thunderbolt.
Isn’t TB 10Gbps in each direction while USB3.0 is 5Gbps total (theoretical speeds for both)?
Comments
Unlike Thunderbolt, USB 3 will be on every netbook, notebook, desktop, and server in the coming months. Thunderbolt will only be available on higher end products.
USB 3 will become the standard and Thunderbolt will become the next FireWire...
TB will be on virtually every computer with an Intel CPU. Apparently you don't understand what native support means, all a manufacturer has to do to add it is provide the (royalty free) plug.
TB will be on virtually every computer with an Intel CPU. Apparently you don't understand what native support means, all a manufacturer has to do to add it is provide the (royalty free) plug.
I don't believe it, but I guess we will have to wait and see the outcome.
Some interesting aspects.
1) Thunderbolt will supposedly be more than twice as fast as USB 3.0. USB, because of its design, apparently, never comes close to its theoretical throughput. This was a big deal in the FW/USB2.0 comparisons, where the USB2.0 peak was higher, but FW400 provided a much faster sustained connection.
2) TB does the capability to provide backwards compatibility with USB 2.0, but the rumor was that the USB groups asked them not to add this in. That would have been the nail in USB3.0's coffin.
However, USB has a pretty strong brand name, and that might be enough to carry it through, and TB might get relegated to Professional uses, like FW was. Lets see if Intel is able to push it enough.
The USB standard was mean to replace the RS-232 and the parallels port in the mid 90, It was a build on simple controler-host topology without any DMA. While the protocol is really simple to boost by speed up the signalling, because of the lack of DMA on USB, reading or writing on USB device make great impact on the CPU. More you boost the bandwidth on USB more you taxes your CPU, making USB inapt for high bandwidth and processing solution like video editing. This is the main reason that explain why FW always been superior to USB for storage.
The new ThunderBolt protocol is a PCI Express 4x external bus, since it pass PCI packet and Intel controller is acting like an PCIExpress-TB bridge with repeater for daisy-chaining device, you virtually can create any kind of external TB adapter like USB, SCSI, Fiber Channel, Ethernet based on existing PCI Express card. ThunderBolt doesn't need to be backward compatible with any port because by design you can adapt any type of port on a PCI bus like Thunderbolt is.
The USB standard was mean to replace the RS-232 and the parallels port in the mid 90, It was a build on simple controler-host topology without any DMA. While the protocol is really simple to boost by speed up the signalling, because of the lack of DMA on USB, reading or writing on USB device make great impact on the CPU. More you boost the bandwidth on USB more you taxes your CPU, making USB inapt for high bandwidth and processing solution like video editing. This is the main reason that explain why FW always been superior to USB for storage.
Wherever did you get the idea that USB doesn't support DMA? Having worked on many USB chips and device drivers, I can assure you that even the cheapest USB chips found in PCs support DMA.
That is not to say that there isn't a *tremendous* overhead in the USB stack. But this is a software issue, not a result of the lack of DMA.
The new ThunderBolt protocol is a PCI Express 4x external bus, since it pass PCI packet and Intel controller is acting like an PCIExpress-TB bridge with repeater for daisy-chaining device, you virtually can create any kind of external TB adapter like USB, SCSI, Fiber Channel, Ethernet based on existing PCI Express card. ThunderBolt doesn't need to be backward compatible with any port because by design you can adapt any type of port on a PCI bus like Thunderbolt is.
It will be interesting to see what the software overhead is for Thunderbolt. Not saying it will be any different from PCIe -- I'm saying I don't know and we'll have to see before making major proclamations about its performance.
USB does pretty good with one device attached. When start attaching a half dozen, the software stack really starts to chew up time. Hopefully, this won't be true for Thunderbolt.
Why? Nobody pays attention to what Apple does. Nobody copies what Apple is doing. Nobody takes their lead from Apple... oh wait.
i'm sick of buying adapters and not being able to plug in old devices when they're needed - that's why TB will take a long time to gain marketshare. the displayport is still hardly anywhere, but the USB port is everywhere. i don't think we'll kill USB until 802.11x and NFC are fast enough for everything to forgo cables with different connectors.
I'm sick of all those adapters too. On a brightest side, the TB port is really a great opportunity for docking station like monitor, I predict the next Apple standalone display will have some port replicator (Ethernet, USB, FW, Sound) connect to the mac thru one TB connection. But beside adaptor and port replicator, by the way of using the PCI protocol on TB, it's really easy for hardware developer to adapt their existing stuff on TB like video grabbing PCI card or SAS Raid unit.
It's ridiculous to even talk about this. As I've said from the beginning, Apple will support USB 3.0 when Intel supports it and we now know that will happen with IvyBridge.
What many don't realize here or can't seem to grasp is that every shipping system with a USB 3 port to date has been shipping with a PCI Express to USB bridge chip to give yoiu those USB ports. Generally those chips support 2-3 USB ports over a single PCI Express lane thus there is no way to ever get full performance across all of the USB 3 ports implemented!
Macs don't have USB 3.0 because Apple would have to add an additional chip to support it. Once Intel natively supports USB 3.0, Apple would have to go through a similar level of effort to avoid supporting it. There's no reason for Apple to do that.
Nothing to do with "acting monopolistic".
Exactly. They would have to use chips to implement USB 3 that up until late last year where power hungery and buggy. Beyond that the bridge chips could never reach USB 3's communications rates across all ports. So really there has been very little incentive for Apple to implement what has been beta hardware to be charitiable.
There is also nothing monopolistic about waiting for technology to mature or stabilize.
As for all the glowing reports from people with their new USB 3 drivesd, that is all fine and good but disk drives, at least magnetic ones, won't be saturating USB3 anyways. It is easy to see great results when your basis for comparison is USB 2, but people need to realize is that TB is a forward looking port. Nobody concerned with TB gives a rats a$$ about how fast a legacy notebook drive dongle works on TB. Yeah the port is faster but that isn't why the port exists by itself, rather it is a forward looking way to handle external communications for future products.
Thus when is see these discussions about USB vs TB the only thing I can think of is man do these people not get it. TB is not a replacement for USB, so any discussion that revolves around TB replacing USB is misplaced at best.
Unlike Thunderbolt, USB 3 will be on every netbook, notebook, desktop, and server in the coming months. Thunderbolt will only be available on higher end products.
USB 3 will become the standard and Thunderbolt will become the next FireWire...
I keep seeing this sentiment but I don?t get it. It sounds like people are drawing comparison based on the speed of the port, but that isn?t a fair comparison. FW was a market failure compared to USB for factors that do not affect TB.
For starters it?s supported by Intel. Second, it not encumbered with the same excessive licensing fees FW was. Third, the mDP port now has more of a reason to catch on as it?s a svelte, free option that can be used for fast data and display.
I see no reason why USB3.0 and TB will not be widely popular.
All technologies are transitional, Thunderbutt included. It's a shame Intel can't move faster to adopt USB 3.0.
Based on the timing and their backing of TB I think they simply choose not to include USB3.0 until they could also include TB/LightPeak.
Wherever did you get the idea that USB doesn't support DMA? Having worked on many USB chips and device drivers, I can assure you that even the cheapest USB chips found in PCs support DMA.
That is not to say that there isn't a *tremendous* overhead in the USB stack. But this is a software issue, not a result of the lack of DMA.
It will be interesting to see what the software overhead is for Thunderbolt. Not saying it will be any different from PCIe -- I'm saying I don't know and we'll have to see before making major proclamations about its performance.
USB does pretty good with one device attached. When start attaching a half dozen, the software stack really starts to chew up time. Hopefully, this won't be true for Thunderbolt.
You're right, I was wrong the USB controller card got DMA, but the USB device doesn't have access directly to the ram to push or read data directly like the FW. It's like TCP/IP, the software driver control each block making a lot of CPU overhead.
The main appeal of TB it's all hardware base, their is any software part of the communication process and doesn't add any layer on top of PCI. But it's still wise of you to wait and see in the real world with real product was is mean in term of real performance.
I'm assuming TB will allow bus-powered drives and charging for future iOS devices when sync'ed, though not sure I've seen this for sure. Seems necessary for it to be successfully in a broad way.
Yes, 10 Watts. More than USB but less than Firewire.
Standard USB hub ports can provide from the typical 500 mA [2.5 W] of current, only 100 mA from non-hub ports. USB 3.0 and USB On-The-Go supply 1800 mA [9.0 W] (for dedicated battery charging, 1500 mA [7.5 W] Full bandwidth or 900 mA [4.5 W] High Bandwidth), while FireWire can in theory supply up to 60 watts of power, although 10 to 20 watts is more typical.
I think, but I?m not sure, Thunderbolt?s 10W is constant regardless of the connection type. Is this enough to connect a external 3.5? HDD without a supplemental power source? I sure hope so.
It's ridiculous to even talk about this. As I've said from the beginning, Apple will support USB 3.0 when Intel supports it and we now know that will happen with IvyBridge.
Macs don't have USB 3.0 because Apple would have to add an additional chip to support it. Once Intel natively supports USB 3.0, Apple would have to go through a similar level of effort to avoid supporting it. There's no reason for Apple to do that.
Nothing to do with "acting monopolistic".
Well, Sony, Dell, toshiba and many others have added the chip with their new sandy bridge 2 computers to allow USB 3.
I stand by my thoughts that apple can and ought to support USB 3 now. Will those iMacs coming in via boats from china right now have USB 3? Probably not.
So if they aren't monopolistic based on hardware ... Jobs penchant for hating blu-ray, hating adobe's flash. Or as evidenced by the recent closed wall approach to iPad, iPhone and now the mac app store. Where they are demanding $ from any company selling services/content on iOS devices... And blocking those who won't pay the vig fee...
Apple loves to keep their overall costs down. It gives them higher margins. Then they love to close their os and charge others 30% to do business there.
I love apple products, but kind miss the old apple which was open more and... Well even if the products cost more... Offer more. You know... Offer USB 3 and blu-ray when others are
Sorry but I ain't the only person noticing the recent years less than nice and cuddly approach of apple.
Based on the timing and their backing of TB I think they simply choose not to include USB3.0 until they could also include TB/LightPeak.
Let's not forget SATA III.
I expect the delays are all part of an agreement between Intel and Apple.
Let's not forget SATA III.
I expect the delays are all part of an agreement between Intel and Apple.
Good point. Are these MBPs the first Macs to use SATA III?
When costs come down enough, it will be possible to use optical buses for another 10x boost in speed, or up to 100Gb/s. And optical Thunderbolt should allow cable runs of up to 100 meters. I doubt any copper-based technology could get anywhere near that performance level. Any future USB 4.0 (if there is one) will need to go optical to compete.
Thunderbolt is already twice as fast as USB 3.0. TB has a theoretical 10Gb/s throughput, while USB 3.0 maxes out at 5.0 Gb/s. But that's just the beginning for Thunderbolt.
Isn’t TB 10Gbps in each direction while USB3.0 is 5Gbps total (theoretical speeds for both)?