Apple Still Considering USB 3.0 in Addition to Thunderbolt?

[sennekuyl]

Connecting 120+ devices.

As most usb devices only have the one connection I'm puzzled where people a daisy chaining USB to 5+ devices on one bus. I've seen Apple keyboards monitors as hubs and 1 (!) printer, but every other device is single port. You could connect several hubs in series, but after a few hubs surely the performance is too low and it is too cumbersome.

 

[benthewraith]

I have to wonder if we will see support for more things (like USB 3.0) now that Cook is CEO. In other words, maybe they won't take such a hard line against things like blu-ray, etc.

Here's to hoping the next MBP I purchase has an HDMI port.

 

[John.B]

As most usb devices only have the one connection I'm puzzled where people a daisy chaining USB to 5+ devices on one bus. I've seen Apple keyboards monitors as hubs and 1 (!) printer, but every other device is single port. You could connect several hubs in series, but after a few hubs surely the performance is too low and it is too cumbersome.

Exactly.

When I read people claiming USB 3.0 supports daisy chaining 127 devices, that tells me they are clueless. It would take a fleet of hubs before getting anywhere near that sort of theoretical limit, and you're certainly not going to get theoretical bandwidth with even one device, let alone dozens.

 

[repoman27]

Here's to hoping the next MBP I purchase has an HDMI port.

And here's to hoping you'll realize that you can just buy a Mini DisplayPort to HDMI with Audio adapter cable for $6 instead of depriving everyone else with common sense 15 mins of battery life through the pointless inclusion of a second digital display interface port on a laptop that already has a perfectly good one.

 

[repoman27]

When I read people claiming USB 3.0 supports daisy chaining 127 devices, that tells me they are clueless. It would take a fleet of hubs before getting anywhere near that sort of theoretical limit, and you're certainly not going to get theoretical bandwidth with even one device, let alone dozens.

USB doesn't support true daisy chaining so much as a tiered star arrangement. You can have up to 5 tiers of hubs connected to a single controller, each hub theoretically capable of having as many as 15 downstream ports. Currently, the most downstream ports you can get in a standalone USB 3.0 hub is four, so it would take 32 hubs to connect the maximum number of addressable devices. And you'd only be left with 95 ports for actual devices because each hub counts as a device. And it would cost $1000. And, as you said, it wouldn't perform particularly well.

 

[repoman27]

People aren't really thinking about the sorts of things they can do with TB before writing it off like mindless idiots. TB means you can have a laptop docking station that you might actually use unlike the old ones which were specific to a given model of laptop. TB docks will be useful for any Mac.

Actually, TB "docks" such as the new ATD will be useful for any computer with a Thunderbolt port. That is, once Intel can increase yields to the point where they have sufficient stocks of Thunderbolt controller chips to service other customers, and at such time as they are at liberty to sell to PC manufacturers at large.

 

[Waragainstsleep]

As most usb devices only have the one connection I'm puzzled where people a daisy chaining USB to 5+ devices on one bus. I've seen Apple keyboards monitors as hubs and 1 (!) printer, but every other device is single port. You could connect several hubs in series, but after a few hubs surely the performance is too low and it is too cumbersome.

Has anyone here ever even wanted to connect 127 devices to a USB port, let alone actually tried it?

 

[Sol]

Has anyone here ever even wanted to connect 127 devices to a USB port, let alone actually tried it?

I think someone on this forum read that USB can do that and presented it as evidence of its superiority over Thunderbolt that is limited to six or seven. As I explained before, one Thunderbolt bus can theoretically host those 127 USB devices with the proper adaptor, ie. a Thunderbolt to USB 3 adaptor.

 

[Buildbright]

Given that Thunderbolt can carry USB 2, Firewire, audio, video and ethernet simultaneously on the Apple Thunderbolt Display, it may be able to do USB 3 too, given the right adaptor. I am only speculating so if anyone knows if this is possible please set the record straight.

Personally I have no use for it. External FireWire drives are fast and reliable enough with the option to be bus-powered without taking a second port (like silly USB splitter cables) and before anyone mentions SSDs, they perform better installed internally via SATA, not in external cases. If I ever get a RAID it would be far better to pay extra for the Thunderbolt version that is faster, instead of a slightly cheaper USB 3 version that compromises performance. Likewise, for audio and video peripherals I would stick to the proven FireWire or the Thunderbolt versions. USB 2 audio gear has been a step back, technology-wise (USB audio seems limited to 96kHz, 24-bit while FireWire goes up to 192kHZ-24-bit and can be used without a host computer) but like most USB peripherals, it is cheaper, thus its prevalence.

Then you can pay for these crazy adapters to convert FireWire to thunderbolt for $150. USB 3.0 will be the standard like it or not. Thunderbolt devices will be more scarce and expensive then even FireWire because they require expensive intel chips for cables and terminals.

 

[theluggage]

However, the gist of the original article seemed to be that Apple was considering inclusion of USB 3.0 prior to the introduction of Ivy Bridge

I wonder if they're just looking at official OS X USB3 support in advance of Ivy Bridge?

That would make it easy to add USB3 PCIe cards to the MP, and ExpressCard/TB-based USB3 controllers to other models.

Although adding a discrete USB3 controller to the Pro is perfectly plausible it still seems a bit odd to make the effort to include it when (as lots of people have pointed out) they'll get the hardware anyway when Ivy Bridge comes out, whether they like it or not. Supplying the OS support so that any MP buyer who cared about USB3 could slap in a $30 PCIe card...

If so, however, not using a USB 3 controller in the new Thunderbolt display seems like a missed opportunity (has it actually shipped yet?)

 

[bigwig]

If the article quoted above, claiming that USB 3.0 will be built into Intel's chipsets but that Thunderbolt will not, is correct, Thunderbolt is dead on arrival. Computer manufacturers are notorious cheapskates, and if they can choose a cheaper chipset that omits TB they inevitably will. TB penetration in the PC space will thus be essentially zero, and since that drives the manufacturing decisions of peripheral makers the number of available TB devices will also be essentially zero.

 

[sennekuyl]

If the article quoted above, claiming that USB 3.0 will be built into Intel's chipsets but that Thunderbolt will not, is correct, Thunderbolt is dead on arrival. Computer manufacturers are notorious cheapskates, and if they can choose a cheaper chipset that omits TB they inevitably will. TB penetration in the PC space will thus be essentially zero, and since that drives the manufacturing decisions of peripheral makers the number of available TB devices will also be essentially zero.

How much greater does TBolt cost to implement than/with USB3?

 

[AidenShaw]

flip the question

Has anyone here ever even wanted to connect 127 devices to a USB port, let alone actually tried it?

Instead, ask "has anyone here ever wanted to connect more than 6 devices to a USB port, or done it".

Note that I started by saying 8 TBolt devices, not 127.

------

How much greater does TBolt cost to implement than/with USB3?

Intel hasn't discussed much about the TBolt controllers, nor published them in its price lists, so we have no idea about the simple material cost for a pair of controllers.

I say "pair", because you need a TBolt controller both in the computer and in the device.

There are very, very few TBolt devices on the market, so we can't even make comparisons of "this USB 3.0 disk costs $X, and the same disk with USB 3.0 + TBolt costs $Y".

About the only firm datapoint is:

• $49.00 - Apple 2m TBolt cable
• $3.99 - 2m USB 3.0 cable (Link Depot cable at Newegg)

 

[sennekuyl]

Biggest difference though? the tbolt cable is an active cable, whereas the usb is passive. With eventual* scaling of cables a tbolt cable will probably ~$15. Apples probably wont go much under 30 though.

* I waited years after USB & Firewire were announced to see them in midrange to cheap computers --- Firewire only just 2 years ago. There were devices sure, but it took ages for them to become ubiquitous. TBolt has been 8 months out the door. F' give it a chance! (Do some words have evil magic here?)

 

[repoman27]

I say "pair", because you need a TBolt controller both in the computer and in the device.

But if you own a Thunderbolt equipped computer, you've already paid for the host controller that's in it. Just like you've paid for the onboard USB controllers. It really comes down to the additional cost to add Thunderbolt controllers and ports to an accessory device, and the premium the cables will demand over time.

About the only firm datapoint is:

• $49.00 - Apple 2m TBolt cable
• $3.99 - 2m USB 3.0 cable (Link Depot cable at Newegg)

Well, to be fair, let's compare Apple's to Apple's... Apple makes and sells quite a few different cables and they range in price from $19 to $99, so the Thunderbolt cable is actually pretty mid-range for them. The Apple Copper Fibre Channel Cable which doesn't even provide half the bandwidth of a Thunderbolt cable despite fairly similar construction goes for $89. It is very reasonable to expect Thunderbolt cables to retail in the $10-15 range once third parties are able to bring them to market. More expensive than USB 3.0, for sure, but with a much smaller target audience and a very different set of capabilities.

If so, however, not using a USB 3 controller in the new Thunderbolt display seems like a missed opportunity (has it actually shipped yet?)

The Apple Store still says "Ships: 2-3 weeks," so I'm guessing it's not out quite yet. Also, some back of the envelope calculations show that the ATD with just the connections it does provide could utilize 84% of the 10 Gbps available on one Thunderbolt channel. The video signal alone requires at least 5.8 Gbps, so there wouldn't be much room to breathe if you added USB 3.0 to that pipe as well.

Then you can pay for these crazy adapters to convert FireWire to thunderbolt for $150. USB 3.0 will be the standard like it or not. Thunderbolt devices will be more scarce and expensive then even FireWire because they require expensive intel chips for cables and terminals.

Do you really not understand that pretty much every PC shipping these days natively supports USB, PCIe, SATA and DisplayPort, and that these things are not in any way interchangeable?

DisplayPort and HDMI are currently duking it out for dominance in the digital display interface space, but neither is close to being replaced by USB 3.0, nor are PCIe or SATA going the way of the dodo anytime soon. Thunderbolt combines PCIe and DisplayPort on a single external port. It is meant to supplant DisplayPort, HDMI, DVI, ExpressCard and PCI Express External. It will exist alongside USB 3.0, just as the protocols it is based on currently do inside your computer.

Your latter argument is also absurd. Most PCs contain far more expensive chips from Intel, and that hasn't stopped them from taking off. And the chips in Thunderbolt cables aren't even made by Intel.

If the article quoted above, claiming that USB 3.0 will be built into Intel's chipsets but that Thunderbolt will not, is correct, Thunderbolt is dead on arrival. Computer manufacturers are notorious cheapskates, and if they can choose a cheaper chipset that omits TB they inevitably will. TB penetration in the PC space will thus be essentially zero, and since that drives the manufacturing decisions of peripheral makers the number of available TB devices will also be essentially zero.

I know it's hard to get used to the fact that times have changed, but Apple isn't the little guy anymore. Macs currently account for more than 10% of the US PC market, which means that more than 10% of the PCs bought in America today have Thunderbolt ports (many of them ALSO have FireWire 800 ports), because Apple isn't a cheapskate. What drives the decisions of the accessory makers is whether or not there is a market and whether or not they can make a profit competing in it. For general purpose mass storage devices, USB 3.0 will probably dominate the market. For solutions which can truly benefit from Thunderbolt's unique capabilities, there will be very little competition from USB 3.0.

 

[AidenShaw]

I know it's hard to get used to the fact that times have changed, but Apple isn't the little guy anymore. Macs currently account for more than 10% of the US PC market, which means that more than 10% of the PCs bought in America today have Thunderbolt ports (many of them ALSO have FireWire 800 ports), because Apple isn't a cheapskate.

However, only a tiny fraction of a fraction of a percent of PCs *in use* are Apples with TBolt today.

What drives the decisions of the accessory makers is whether or not there is a market and whether or not they can make a profit competing in it.

For general purpose mass storage devices, USB 3.0 will probably dominate the market.

For solutions which can truly benefit from Thunderbolt's unique capabilities, there will be very little competition from USB 3.0.

I agree with your second sentence, bit it almost implies that the major accessory market will not be interested in TBolt.

The third sentence is consistent with TBolt as a high-priced niche interconnect.
------

I don't disagree with anything you say, but with 10% penetration into an established market it will take years to get critical mass, and if the current universal standard is "good enough" for a large part of the market - TBolt may have a hard time getting the volumes that will bring its price down and generate volumes.

 

[SilianRail]

However, only a tiny fraction of a fraction of a percent of PCs *in use* are Apples with TBolt today.




I agree with your second sentence, bit it almost implies that the major accessory market will not be interested in TBolt.

The third sentence is consistent with TBolt as a high-priced niche interconnect.
------

I don't disagree with anything you say, but with 10% penetration into an established market it will take years to get critical mass, and if the current universal standard is "good enough" for a large part of the market - TBolt may have a hard time getting the volumes that will bring its price down and generate volumes.

Especially since there will be a widely supported and low cost 32 Gbps PCI Express interconnect in 2013.

 

[repoman27]

However, only a tiny fraction of a fraction of a percent of PCs *in use* are Apples with TBolt today.

Right you are, but better than 10% of all PCs shipping with a new technology only 8 months after introduction is a fairly astonishing feat. If you're a device manufacturer who's looking ahead 5 years, you'll see more than 80 million PCs with Thunderbolt host controllers even if nobody but Apple decides to pick it up.

I agree with your second sentence, bit it almost implies that the major accessory market will not be interested in TBolt.

The third sentence is consistent with TBolt as a high-priced niche interconnect.

Thunderbolt, like FireWire, will always be more expensive to implement than USB and never achieve the same volume of devices. If USB is perfectly suitable for a given device, you wouldn't really consider a Thunderbolt version as it wouldn't be terribly marketable. There are, however, still quite a few PCIe expansion cards on the market for which USB 3.0 isn't viable. If you've already developed one of these, you could, with relative ease, repackage it in an external enclosure along with a Thunderbolt controller and expand your market to all the folks who don't have a spare PCIe slot but do have Thunderbolt. With the increasing popularity of notebooks, sub-notebooks, all-in-ones and small-form-factor PCs, that market has the potential to become quite large.

FireWire devices, and now Thunderbolt, will always have a customer base as long as music and movie producers can afford them. I'm pretty sure that's why RIAA and MPAA are fighting so hard to stop illegal downloading and file sharing. It's obvious, really.

 

[toke lahti]

I have always wondered what it would be like living in the movie Blade Runner.

BladeRunner2 is coming and probably in 3D, so maybe you can try in a theather near you...

...oh and two dedicated 2 DisplayPort channels.

Which neather of them are able to even comply DPv1.2 from year 2009...

You could combine these two:
Image
http://www.sonnettech.com/product/usb3expresscard34.html
Image
http://www.sonnettech.com/product/echoexpresscard34thunderbolt.html

And with this amazingly cluttered option, you can achieve the same thing that would cost $3 in a mac (usb3 port), but you have to pay something like 50 times more.
How simple!

 

[repoman27]

Especially since there will be a widely supported and low cost 32 Gbps PCI Express interconnect in 2013.

Thunderbolt is designed to scale along with PCIe and DisplayPort, so it's likely that a Thunderbolt controller with connections for four lanes of PCIe 3.0 and two DisplayPort 1.2 links will already be shipping before PCIe External Cabling even gets out the gate. And considering how quickly PCIe External has taken off at the 2.5 and 5 Gbps signaling rates, I'm sure it will be ubiquitous as soon as the 8 Gbps version is finalized. It will be an instant hit, even though there will be a more capable, proven technology with over 50 million host controllers already shipped in the market.

Aside from not existing except on paper, not being daisy-chainable, and not supporting a digital display interface protocol, what are the reasons why one would prefer PCIe External?

I'm not even sure why I'm arguing with you. In your mind Thunderbolt is DOA and somehow PCI External Cabling is not a candidate for being relegated to obscurity.

 

[repoman27]

Which neather of them are able to even comply DPv1.2 from year 2009...

Actually, Thunderbolt provides two 10 Gbps channels, either of which are capable of transporting a DisplayPort 1.1a main link which has a maximum bandwidth of 8.64 Gbps. DisplayPort 1.2 requires 17.28 Gbps for the main link. It would have been a pretty big ask to expect Thunderbolt to provide two 20 Gbps channels upon its initial release in order to handle DP 1.2. I reckon you have a bunch of consumer electronics cables lying around your cave that can handle 40 Gbps?

 

[toke lahti]

Actually, Thunderbolt provides two 10 Gbps channels, either of which are capable of transporting a DisplayPort 1.1a main link which has a maximum bandwidth of 8.64 Gbps. DisplayPort 1.2 requires 17.28 Gbps for the main link. It would have been a pretty big ask to expect Thunderbolt to provide two 20 Gbps channels upon its initial release in order to handle DP 1.2. I reckon you have a bunch of consumer electronics cables lying around your cave that can handle 40 Gbps?

When people buy computers, which are about the most pollutive commodity, they should think about future expandability, not just what they need now.
(Like someone said, skate where the puck will be...)
You probably know that Airs have only half of the TB bandwidth?
At the same time we are seeing first 4k 3D displays:
http://www.studiodaily.com/main/tec...Your-Eyes-Approaches-4K-Resolution_13406.html

It would have been pretty wise for Apple to do like Sony and put TB in usb socket. That way they wouldn't have to cripple the DP.
But maybe Apple wanted to cripple macs to shorten their life to sell more often the same thing to same customer.

 

[repoman27]

When people buy computers, which are about the most pollutive commodity, they should think about future expandability, not just what they need now.
(Like someone said, skate where the puck will be...)
You probably know that Airs have only half of the TB bandwidth?
At the same time we are seeing first 4k 3D displays:
http://www.studiodaily.com/main/tec...Your-Eyes-Approaches-4K-Resolution_13406.html

Apple obviously should have made sure that all Macs could drive the full resolution of $11,000 televisions that don't exist yet. How wasteful of them not to. They also should have just skipped ahead and included Ivy Bridge so that we won't have more capable Macs on the market next year, tempting people to upgrade yet again. I guess you missed my earlier point about a cable that can handle 40 Gbps not being in the realm of consumer products at this point in the technological advancement of the human race.

The Thunderbolt port on the MacBook Air provides the same amount of bandwidth as all Thunderbolt ports, 10 Gbps x 2 channels. The Eagle Ridge TB controller used only has one DisplayPort connection, which is fine since there is no dGPU to hook to a second one, and only supports a single TB port, which is also fine since there isn't really room on the Air for more than one. See, Apple used a less expensive chip with a smaller die and provided the end user with the maximum capabilities of the platform at the same time, maybe they aren't so wasteful after all...

DisplayPort 1.1a supports displays with resolutions of 2560x1600. Thunderbolt ports can drive up to 2 of these. How many displays with resolutions north of 2560x1600 are currently on the market? I realize that there is an announced home theater standard which would run at 4096×2160p24, but how many of these currently exist or will exist in three years? Is there any content yet released in this format? 99% of what you'll view on one of these screens in the near future will be scaled anyway, so what difference does it make if the output from your MacBook Air is also scaled? It would also suck to use one of these as a PC display at full resolution because the refresh rate is only 24 Hz.

It would have been pretty wise for Apple to do like Sony and put TB in usb socket. That way they wouldn't have to cripple the DP.
But maybe Apple wanted to cripple macs to shorten their life to sell more often the same thing to same customer.

Yeah Sony is clearly designing products with the maximum possible lifespan, hoping that their customers buy once and won't come back anytime soon. Unlike Apple, whose products just happen to retain their resale value better than any other PC brand...

It would be pretty wise for you to quit trolling.

 

[AidenShaw]

If you're a device manufacturer who's looking ahead 5 years, you'll see more than 80 million PCs with Thunderbolt host controllers even if nobody but Apple decides to pick it up.

True, but unfortunately our stock markets discourage most companies from looking more than 3 months ahead.

On the other hand, if half the PCs sold soon add TBolt, you'll have half a billion potential customers in 5 years. (very rough estimate)

The problem for TBolt, however, is whether any significant fraction of those 80M/500M systems have any use for TBolt. If 99% of the market is happy with a USB 3.0 external drive or two, then your potential market is .8M/5M systems.

I'm in the 1% who would add a PCIe->TBolt card to my system tomorrow, and get a RAID controller with at least 8 bays. (I have 10 bays of eSATA drives on my home PC right now, and a 5-bay eSATA array on my home server.)

Realistically, though, it's likely that the price will be outrageous due to the low volume of TBolt customers. (Not that there won't be enough systems with TBolt, but that few people will have any need to pay the TBolt tax on peripherals.)

 

[bigwig]

On the other hand, if half the PCs sold soon add TBolt, you'll have half a billion potential customers in 5 years. (very rough estimate)

Which manufacturer is going to pay extra, either for Intel's TB support or for a third-party chip? Given that most look for any way to save mere pennies from their costs, I'd say none. That's why Intel's purported move to omit TB from its standard chipset is so full of failure. The only reason you do that is if you want to kill TB, just like lack of standard FireWire support from Intel doomed FW to irrelevance. If Intel really wanted to stand behind TB they'd make it standard and require anybody who wanted to put an Intel Inside sticker on their box implement it.