AkG
Well-known member
- Joined
- Oct 24, 2007
- Messages
- 5,270
Closer Look at USB 3.1
The easiest way to start to describe what has changed with USB 3.1 standard is to start with what has been carried over from previous generations. First and foremost Type A and Type B connectors are still around and a USB 3.1 Type A port is identical to a USB 3.0 Type A port. The same holds true for Type B ports. That is to say both are physically the same as their USB 3.0 predecessor.
This in turn means that USB 3.1 is based upon a 4 data lane configuration - just as USB 3.0 was. More importantly, USB 3.1 Type A and Type B ports are fully backwards compatible with USB 2.0 and USB 3.0 devices - they just will not work at USB 3.1 speeds.
This backwards compatibility was done on purpose. USB 3.1 does indeed represent a new direction and approach for the USB standard but USB-IF wanted consumers of existing devices to not worry about compatibility. Unlike Apple who threw their existing user-base under the bus numerous times, if your device works with USB 3.0 Type A or Type B ports it will fully connect and work via USB 3.1 Type A or B. More importantly consumers should notice almost no differences between connecting them via USB 3.0 and USB 3.1 controllers and ports.
For this reason, USB Type A and Type B ports will still be a part of the computing landscape and in all likelihood Type A's will still be the de-facto standard ports found on motherboards for the foreseeable future. We will see some of the new 'Type C' ports on motherboards but Type A will be the most common - just as when USB 3.0 was released and motherboards came with 2.0 and 3.0 ports, expect both A and C type USB 3.1 ports to co-exist.
This backwards compatibility and physical layout is nearly the grand total of what has been carried over to the next generation 'SuperSpeed Plus' USB standard. In fact, if it was not for backwards compatibility USB 3.1 in all likelihood would not have even exhibited this amount in common with its predecessors. We will get to the new Type C connector later but even excluding this new connector type USB 3.1 is an entirely new breed of USB built on a completely new foundation.
In some ways USB 3.1 is actually a return the original USB founders’ goal of replacing as many different and competing standards as possible. In the 1990s this meant simply being 'plug and play' via one all-encompassing USB driver set. Now the landscape is very different and in order to replace everything from HDMI to ThunderBolt and even power ports requires a new way of doing things.
With this in mind, the USB-IF started by changing the very encoding scheme USB uses. In the past USB generations, 8-bit data chunks would be encoded into 10-bit symbols and then passed over the USB interface, then at the other end of the connection this 10-bit encoding would then be decoded into the original 8-bits. The extra 2-bits of data was the sum total of the Error-Correcting Code (ECC) and this amounted to a twenty percent overhead packet loss, thus reducing speeds even further.
With USB 3.1, the USB-IF has moved to a new and highly sophisticated encoding scheme they have dubbed Gen X. The Gen X scheme does things differently and is best compared to how Ethernet transmits data. Much like your wireless Ethernet connection, USB 3.1 packets are much, much larger. Instead of USB 3.0's 10-bit packet that has only 8-bits of data, USB 3.1 sends data packets that contain 128 bits of data. Also like Ethernet, USB 3.1 uses a 'header' that contains the ECC for each packet as well as the instructions for what is inside the packet. This 4 bits of data also has an Error Correction Code built into itself and can be reassembled as long as at least 3 bits are intact.
Obviously this 4+1 ECC is much more advanced than the original 2-bit ECC used in USB 3.0, but also allows USB 3.1 to boast an theoretical overhead of only 3%. This increase in packet size, in-conjunction with better ECC, is precisely how the USB-IF was able to push theoretical maximum speed from 5Gbit/s to 10GBits/s, even though USB 3.1 uses the same 4 data lanes that was first introduced in USB 3.0 specification.
Bulk Only Transport (BoT) Protocol has also been updated and improved. The Bulk Transport protocol is a specific mode meant solely for transporting large amounts of data over USB. Nearly every motherboard gives their take on BoT implementation a different name, but ASUS uses the apt description of 'Turbo Mode'. When enabled, different software drivers are used for USB file transfer. These drivers allow a USB connection to consume as much bandwidth as it can, with little regards for other devices attached, and use greatly increased packet sized. For best results BoT should be used on a clear USB channel with no other devices attached to it.
In previous generations BoT did improve performance somewhat but the end result was extremely variable. In order to improve upon USB 3.0 BoT performance, USB 3.1 not only adds in SCSI command support - to reduce delays between command phases - but also adds in a caching element in which the controller uses a portion of its onboard cache for BoT I/O's. Unfortunately, Command Queuing is still absent and the I/O requests are processed in the order they are received, just as with USB 3.0. As such it is best to only transfer one file at a time using Turbo Mode, otherwise overall performance will suffer.
Interestingly, ASUS' next generation Turbo Mode also supports standard SCSI commands over USB and not just for USB attached SCSI devices (UASP). This is an important feature as next generation Solid State controllers are starting to include SCSI command capabilities, and as such ASUS motherboards may in fact provide improved performance over competitors' models in the future.
A doubling in the performance department is certainly impressive, but sheer speed is only one of the improvements the USB-IF is counting on to eliminate the competition. Up until USB 3.1, a USB port and USB cable could really only be used to transmit USB encoded data. For example, if a consumer wanted to add an external monitor to their system they either had to use a built-in controller and port, or they would have had to purchase USB based external display adapter and controller and use it between the monitor and the computer. USB 3.1 eliminates the need for specialized ports and external 'adapters' - be they displaybased, Ethernet, or other. Instead, monitor outputs, Ethernet cables, and nearly every other connector found on the typical desktop, laptop, and hand held computer can be used via the USB 3.1 port.
USB 3.1 is able to boast such impressive abilities due to a new addition to the actual USB standard. Since USB 3.1 already uses a header for their data packets it was relatively simple to encode in an additional code to tell the 'other end' of the connection that a given packet was not encoded via the USB standard but instead was encoded via some other standard. For example if the header states a given packet is encoded using the DisplayPort standard, the client side of the connection will treat it as an audio/visual package - just as if it was sent via a DisplayPort connector and cable. This new mode is aptly called 'alternate mode' and it can be used on any - or all - of the four data lanes at any given time.
If we use the same display output analogy as above, a compatible monitor both receive audio and video via a single USB 3.1 cable while it is also being used as a USB 3.1/3.0/2.0 hub with a keyboard, mouse, printer, etc also connected to this one cable. Alternately if you use a HDMI to USB adapter cable monitors with 'just' an HDMI port can still use a single cable to connect to the computer - as long as the monitor supports the Mobile High-Definition Link standard.
At this time the DisplayPort and Mobile High-Definition Link (MHL) Consortium have already agreed to their perspective standard being used via USB. Meanwhile Ethernet and even PCIe governing bodies are in talks with the USB-IF. For laptop and tablet users, once the "Media Agnostic USB specification and protocol" is finalized, future portable devices may look a great deal sleeker with drastically fewer port types.
Being able to provide audio and visual data via USB is in and of itself very, very interesting, but on its own would have proven to be of limited use for laptops, tablets and other portable devices. To this end, the USB-IF also increased the USB Power Delivery standard.
In previous generations, USB Power Delivery Protocol was limited to a maximum of 5 amps at 5 volts, or 25 watts total. With USB 3.1 this has been increased to a maximum 5amps at 20 volts - or a whopping 100 watts. In theory this means one USB 3.1 port could be used as a power-in port on UltraBook while another is used to power external devices such as monitors, external storage arrays, or even printers.
There has been some confusion regards this new Power Delivery standard and it is not directly tied to the new Type-C port, rather it is tied to the controllers connected to the port and the cables themselves. What this means is that while we could in theory see Type A ports sporting 100 watt capabilities this is unlikely due to their backwards compatibility; using a standard Type A cable would result in a fire hazard with such a massive increase in power flow. Instead 100 watt connections will most likely be reserved for Type-C ports, and Type-C cables. Even then -thanks the auto negotiation chips in the client controller and host controller, not every Type-C cable will be 'allowed' to handle 100 watts of power.
Closer Look at USB 3.1
The easiest way to start to describe what has changed with USB 3.1 standard is to start with what has been carried over from previous generations. First and foremost Type A and Type B connectors are still around and a USB 3.1 Type A port is identical to a USB 3.0 Type A port. The same holds true for Type B ports. That is to say both are physically the same as their USB 3.0 predecessor.
This in turn means that USB 3.1 is based upon a 4 data lane configuration - just as USB 3.0 was. More importantly, USB 3.1 Type A and Type B ports are fully backwards compatible with USB 2.0 and USB 3.0 devices - they just will not work at USB 3.1 speeds.
This backwards compatibility was done on purpose. USB 3.1 does indeed represent a new direction and approach for the USB standard but USB-IF wanted consumers of existing devices to not worry about compatibility. Unlike Apple who threw their existing user-base under the bus numerous times, if your device works with USB 3.0 Type A or Type B ports it will fully connect and work via USB 3.1 Type A or B. More importantly consumers should notice almost no differences between connecting them via USB 3.0 and USB 3.1 controllers and ports.
For this reason, USB Type A and Type B ports will still be a part of the computing landscape and in all likelihood Type A's will still be the de-facto standard ports found on motherboards for the foreseeable future. We will see some of the new 'Type C' ports on motherboards but Type A will be the most common - just as when USB 3.0 was released and motherboards came with 2.0 and 3.0 ports, expect both A and C type USB 3.1 ports to co-exist.

This backwards compatibility and physical layout is nearly the grand total of what has been carried over to the next generation 'SuperSpeed Plus' USB standard. In fact, if it was not for backwards compatibility USB 3.1 in all likelihood would not have even exhibited this amount in common with its predecessors. We will get to the new Type C connector later but even excluding this new connector type USB 3.1 is an entirely new breed of USB built on a completely new foundation.

In some ways USB 3.1 is actually a return the original USB founders’ goal of replacing as many different and competing standards as possible. In the 1990s this meant simply being 'plug and play' via one all-encompassing USB driver set. Now the landscape is very different and in order to replace everything from HDMI to ThunderBolt and even power ports requires a new way of doing things.
With this in mind, the USB-IF started by changing the very encoding scheme USB uses. In the past USB generations, 8-bit data chunks would be encoded into 10-bit symbols and then passed over the USB interface, then at the other end of the connection this 10-bit encoding would then be decoded into the original 8-bits. The extra 2-bits of data was the sum total of the Error-Correcting Code (ECC) and this amounted to a twenty percent overhead packet loss, thus reducing speeds even further.
With USB 3.1, the USB-IF has moved to a new and highly sophisticated encoding scheme they have dubbed Gen X. The Gen X scheme does things differently and is best compared to how Ethernet transmits data. Much like your wireless Ethernet connection, USB 3.1 packets are much, much larger. Instead of USB 3.0's 10-bit packet that has only 8-bits of data, USB 3.1 sends data packets that contain 128 bits of data. Also like Ethernet, USB 3.1 uses a 'header' that contains the ECC for each packet as well as the instructions for what is inside the packet. This 4 bits of data also has an Error Correction Code built into itself and can be reassembled as long as at least 3 bits are intact.
Obviously this 4+1 ECC is much more advanced than the original 2-bit ECC used in USB 3.0, but also allows USB 3.1 to boast an theoretical overhead of only 3%. This increase in packet size, in-conjunction with better ECC, is precisely how the USB-IF was able to push theoretical maximum speed from 5Gbit/s to 10GBits/s, even though USB 3.1 uses the same 4 data lanes that was first introduced in USB 3.0 specification.

Bulk Only Transport (BoT) Protocol has also been updated and improved. The Bulk Transport protocol is a specific mode meant solely for transporting large amounts of data over USB. Nearly every motherboard gives their take on BoT implementation a different name, but ASUS uses the apt description of 'Turbo Mode'. When enabled, different software drivers are used for USB file transfer. These drivers allow a USB connection to consume as much bandwidth as it can, with little regards for other devices attached, and use greatly increased packet sized. For best results BoT should be used on a clear USB channel with no other devices attached to it.
In previous generations BoT did improve performance somewhat but the end result was extremely variable. In order to improve upon USB 3.0 BoT performance, USB 3.1 not only adds in SCSI command support - to reduce delays between command phases - but also adds in a caching element in which the controller uses a portion of its onboard cache for BoT I/O's. Unfortunately, Command Queuing is still absent and the I/O requests are processed in the order they are received, just as with USB 3.0. As such it is best to only transfer one file at a time using Turbo Mode, otherwise overall performance will suffer.
Interestingly, ASUS' next generation Turbo Mode also supports standard SCSI commands over USB and not just for USB attached SCSI devices (UASP). This is an important feature as next generation Solid State controllers are starting to include SCSI command capabilities, and as such ASUS motherboards may in fact provide improved performance over competitors' models in the future.

A doubling in the performance department is certainly impressive, but sheer speed is only one of the improvements the USB-IF is counting on to eliminate the competition. Up until USB 3.1, a USB port and USB cable could really only be used to transmit USB encoded data. For example, if a consumer wanted to add an external monitor to their system they either had to use a built-in controller and port, or they would have had to purchase USB based external display adapter and controller and use it between the monitor and the computer. USB 3.1 eliminates the need for specialized ports and external 'adapters' - be they displaybased, Ethernet, or other. Instead, monitor outputs, Ethernet cables, and nearly every other connector found on the typical desktop, laptop, and hand held computer can be used via the USB 3.1 port.
USB 3.1 is able to boast such impressive abilities due to a new addition to the actual USB standard. Since USB 3.1 already uses a header for their data packets it was relatively simple to encode in an additional code to tell the 'other end' of the connection that a given packet was not encoded via the USB standard but instead was encoded via some other standard. For example if the header states a given packet is encoded using the DisplayPort standard, the client side of the connection will treat it as an audio/visual package - just as if it was sent via a DisplayPort connector and cable. This new mode is aptly called 'alternate mode' and it can be used on any - or all - of the four data lanes at any given time.

If we use the same display output analogy as above, a compatible monitor both receive audio and video via a single USB 3.1 cable while it is also being used as a USB 3.1/3.0/2.0 hub with a keyboard, mouse, printer, etc also connected to this one cable. Alternately if you use a HDMI to USB adapter cable monitors with 'just' an HDMI port can still use a single cable to connect to the computer - as long as the monitor supports the Mobile High-Definition Link standard.
At this time the DisplayPort and Mobile High-Definition Link (MHL) Consortium have already agreed to their perspective standard being used via USB. Meanwhile Ethernet and even PCIe governing bodies are in talks with the USB-IF. For laptop and tablet users, once the "Media Agnostic USB specification and protocol" is finalized, future portable devices may look a great deal sleeker with drastically fewer port types.

Being able to provide audio and visual data via USB is in and of itself very, very interesting, but on its own would have proven to be of limited use for laptops, tablets and other portable devices. To this end, the USB-IF also increased the USB Power Delivery standard.
In previous generations, USB Power Delivery Protocol was limited to a maximum of 5 amps at 5 volts, or 25 watts total. With USB 3.1 this has been increased to a maximum 5amps at 20 volts - or a whopping 100 watts. In theory this means one USB 3.1 port could be used as a power-in port on UltraBook while another is used to power external devices such as monitors, external storage arrays, or even printers.
There has been some confusion regards this new Power Delivery standard and it is not directly tied to the new Type-C port, rather it is tied to the controllers connected to the port and the cables themselves. What this means is that while we could in theory see Type A ports sporting 100 watt capabilities this is unlikely due to their backwards compatibility; using a standard Type A cable would result in a fire hazard with such a massive increase in power flow. Instead 100 watt connections will most likely be reserved for Type-C ports, and Type-C cables. Even then -thanks the auto negotiation chips in the client controller and host controller, not every Type-C cable will be 'allowed' to handle 100 watts of power.
Last edited by a moderator: