Slashdot Mirror
Thunderbolt vs. SuperSpeed USB
Lucas123 writes "When it comes to performance, power and size, external I/O interconnect Thunderbolt handily beats SuperSpeed USB, but in the one critical category — ubiquity — it has an almost impossible uphill battle. Thunderbolt has a maximum 10Gbps signaling rate to SuperSpeed USB's 6Gbps and it offers more than twice the power to devices. To date, however, Apple is the only systems manufacturer to adopt Thunderbolt, and it has done so as an additional device connectivity port, keeping SuperSpeed USB on its computers. No other systems manufacturer has committed to Thunderbolt. In contrast, SuperSpeed USB has been installed on 10 billion pieces of hardware, with numbers continuing to grow."
SuperSpeed USB has been installed on 10 billion pieces of hardware
No it hasn't. USB may have been installed on 10 billion pieces of hardware, but SuperSpeed USB is nowhere near as ubiquitous yet. SuperSpeed USB may be able to compatibly downgrade to full-speed USB communication, but that doesn't mean that anything you plug a SuperSpeed device into is magically SuperSpeed.
Anyway, I like the idea of Thunderbolt, especially the thought that it could become the holy grail of single cable interconnects. But just because I like a thing and it's technically better doesn't mean the world will adopt it. Unfortunately, I've learned that politics and money will drive the decision, not technology.
John
When the Compact Disc first came out not many had any chance to play one. It was expensive, part of extravagant home theatre systems, and only the rich could afford it. Years later it was adopted by the masses once it was able to be cheaply reproduced. The same goes for this piece of technology. While truly innovative and new technology almost never starts out as being ubiquitous; it does move us forward. This is my point: it is better, faster, and eventually it will be cheaper too. That and I heard Apple had exclusivity on the hardware totally until 2012 with regards to it. http://www.appleinsider.com/articles/11/02/24/intel_details_thunderbolt_as_exclusive_to_apple_until_2012.html
Help me, help you. - Jerry McGuire
Also, Apple has not have SuperSpeed USB on any of it's computers.
Huh? When did Apple start offering USB 3.0? AFAIK they're still shipping USB 2.0 only.
Dear Slashdot,
How about taking a little more pride in what you choose to show to the 8 bajillion people who hit your front-page every day?
Hmm, this sounds just like the Firewire vs. USB competition with Apple pushing Firewire. We saw how that turned out. We all know that being better doesn't mean anything in this industry.
If you actually read TFA, it does't say that SuperSpeed USB has been installed in 10 million devices. It says that USB has been installed in 10 million devices.
10 gps vs 4.8 gps isnt enough to make me want to add an extra port.
As you can find usb 3.0 pci-e cards and having Thunerbolt on the main system board is slowing down roll out.
Acer and Asus have signed up for Thunderbolt and are expected to deliver PCs with Thunderbolt next year. Except more motherboards to have Thunderbolt as well, and once that occurs, Dell and other has-beens will do the same.
How Will Thurberbolt work with pci-e video cards?
As you need a way to link the display port bus to the MB based TB data bus and I don't see any plans any where saying how they plan to pull that off.
I'm looking to increase my backup times to an external USB3 drive. My PC only supports USB2, but I'm looking to drop in a card to support USB3. Last I checked, all the on-board USB controllers on new motherboards are still somewhat dicey (crap). Does anyone here have a good recommendation as to what USB3 controller based card I should get (2 to 4 port is ok)? The goal is for direct drop-in (no fancy drivers) for Server 2008 and Win7, and the least hardware bug prone. Cost is not an issue.
Life is not for the lazy.
It is different, hence why Intel makes both (Thunderbolt is Intel's not Apple's). Thunderbolt is basically just an external PCIe bus. While that has a benefit of great speed and low latency, it has drawbacks. The client device has to be more complex (and thus expensive) since it needs a PCIe controller on it. Also a device can hose your system, being PCIe it has DMA and can write or read any memory.
USB is much simpler. Slave devices need little logic to handle it. Also it is handled through the CPU which, while slower, is safer meaning an errant device can't as easily trash your system.
It is if transferring an hour of 1080p video.
Thunderbolt is interesting because of the potential maximum cable length. The current cupper cables are limited to 3 meters, but once optical cables are available, "10s of meters" will be possible.
Since you can run both display, keyboard and mouse over one cable natively, this means that you can put your computer with its noisy fans into the basement, use a single thunderbolt cable, and just have an extremely thin client at your workstation.
And apple seems to be the only place to get a Thunderbolt cable right now.
Did you buy your UID from somebody else? Because otherwise, you would have to be hopelessly optimistic. Or unable to notice a pattern that has held true over a number of years...
Sleep your way to a whiter smile...date a dentist!
This would be more like Intel releasing a new x86 processor that is *almost* as good as a hypothetical ARM chip that beats the x86 offering on price, performance, and electricity usage.
Does this parable refer to the Atom CPU?
Thunderbolt was just released a few months ago. USB3 has been out for almost 2 years and it is finally starting to get a little traction in the marketplace. Something else to keep in mind is that Thunderbolt is Intel tech, not Apple. Intel is pushing Thunderbolt so Thunderbolt will be on 95% of mobos in a couple years. Since Thunderbolt is Intel tech, Microsoft will support it as well. Don't discount how much damage was done to FW by shitty MS firewire drivers that barely worked. Intel, Microsoft, and Apple will all be pushing Thunderbolt to succeed.
One last thing, look for video card manufacturers to be pushing TB as well to get rid of DL DVI,DVI, and VGA cables.
Thunderbolt will succeed.
I'm still having a bit of trouble understanding the exact use case you're thinking of. Where are you getting the 1080p video and where are you putting it such that you need to transfer it far faster than real-time (0.05 Gbps or less)?
10 GB/s vs 4.8 GBs? Those are theoretical. What really matters is practical, real life speeds, or failing that, benchmarks.
Thunderbolt is more like an option that wants to displace DVD and BluRay.
Collectively DVD and BD have this massive legacy library. A newer player has legacy support for the older content.
An entirely new option has to completely ditch all of that and may have inferior selection. There may be things that are simply not available on the "hot shiny new" option despite the fact that it is "hot, shiny, and new".
A Pirate and a Puritan look the same on a balance sheet.
That's like responding to someone saying they don't want a sports car with "You would if you were in a race!"
As much as a small segment of the population may want the faster transfer speeds of Thunderbolt, the majority will probably opt for the tried and true USB form factor and it's backward compatibility. Currently the few USB 3.0 devices I've seen have all worked with USB 2.0 as well, which is another big factor.
It's interesting to be sure, and I will probably opt for a thunderbolt compatible motherboard on future builds in the interests of future-proofing, but I'm not going to throw my dollars behind supporting Thunderbolt devices until it becomes more ubiquitous and a LOT more 3rd party hardware vendors start supporting it.
> It is if transferring an hour of 1080p video.
Even USB2 will do for that. Not a big deal. There are plenty of better options available assuming I tire of the relative slowness of USB2.
I move files around by the terabyte. In that use case, the slowness of other options becomes annoying. However, there's only so much I am willing to spend on a better solution. Also, if I blow a big wad of cash for it I will have very high expectations and be very easily disappointed.
My non-cheap machines all have USB3. My cheap machines all have eSATA.
A Pirate and a Puritan look the same on a balance sheet.
This is exactly like when Firewire came out and USB was it's inferior little cousin. Apple adopted it and it was relegated to the realm of more expensive devices. I have the feeling this will happen again. SSUSB will prosper while Thunderbolt will be used almost exclusively by the people fond of a particular red fruit.. Here's the interesting part. Will the iPad/2 support Thunderbolt backward compatibility? Considering there isn't and "backward" for Thunderbolt to go, I doubt it. The industry is going to go to the easiest connection to support and if SSUSB/USB3 is fully backward compatible to USB1.0 then this is probably the last you hear of Thunderbolt if it doesn't come from an iFanboi telling you how inferior SSUSB is.
Now with Thunderbolt do you really want to have unplug your display to mount and unmount an ext hdd?
Why not chain the HDD off the display instead?
Life is not for the lazy.
One thing that is really holding Thunderbolt back (in my opinion) is the holding back of a Thunderbolt spec. The claim at NAB was that the SDK would come out this year, but apparently they only meant it would come out to a select group of companies making high end equipment directed at graphics specialists. I have lots of ideas for what I'd do with PCIe over displayport, but there's no way for me as an individual to get the SDK, and there's no contact person at Intel to email to even ASK how I'd get it. If it's anyone's fault the technology doesn't get adopted, it's Intel's.
You can watch "The Secret Life of Walter Mitty" on them?
Thunderbolt and SuperSpeed USB aren't even competing for the same market and don't have the same purpose. You will probably never have Thunderbolt mouse, keyboard, headset, printer, etc. There's no reason to use such expensive cabling and ridiculous bandwidth for devices like that.
Docking stations, breakout boxes, external PCIe cards, displays and high speed RAID arrays are what Thunderbolt is for. A box with say... 4 ethernet adapters is good for Thunderbolt. Your webcam though? Uhhhhh, no.
I imagine much of this will be decided by how difficult it is to program for. One of the things that ended up keeping USB back from some of the higher end uses (like external video cards) was how huge of a pain it was to program for... at my previous company I always heard grumbling from the USB programmers.... if Thunderbolt is layed out better for these types of devices then, even if it *gasp* costs a whole $1 per port then it might have a place.
Oh god I agree with APK.
I'm looking at my laptop, I see no less than 4 different ports for bidirectional data. USB, Firewire, Thunderbolt, ExpressCard/32... Ethernet if that counts would make it 5.
There's space on motherboards everywhere for thunderbolt. Especially considering that the mini DP is free to license and use as a port, and it plays nice with DisplayPort devices... I don't see how this is a bad situation.
Non impediti ratione cogitationus.
Since either are faster than ssd or hdd speeds, it does not really matter, your bottleneck is your drive and sata connecters. I dont see paying extra for something that will not work with any devices I have, or will have to pay an extra premium for later.
the display needs to be thunderbolt for that to work but a Display port display can be at the end of chain.
I stand corrected. I was under the impression that the new 27" Apple Displays had two Thunderbolt ports for daisy-chaining to another display. Guess not. Bummer.
Life is not for the lazy.
Perhaps you can have the TB controller on the video card, you then have all you need, the video, and the PCIe. and Bingo.
instead of signals transmitted using electricity. That essentially remove surges between your equipment from the equation. I have no longer any count of how many times I have experienced small shocks from connecting an external hard drive with its own power cord to my laptop or desktop. That may not stop, but I will at least know this no longer impact my machine. On the other side - I am not going to run for Thunderbolt until I know the failure rate of fiber cables due to bending etc. vs current USB cables... Especially if the price for new cables is substantially more expensive.
true, but there is 1 crucial factor. When you plug a SuperSpeed device into an existing port, it still works.>
Lets be clear here - the only super speed devices we'll see really are storage devices.
In that case the devices do not really "work". Yes you'll be able to get to files on them in a pinch, but they will not be usable for the reason you got the external super-speed storage for in the fast place - either large image or video editing. The thing will be dog slow and basically unusable.
It's the same way a modern car will "work" when it goes into emergency lockdown limp mode and proceeds at no more than 10MPH. Yes it is "working" but what good is that really?
I have an external USB 3.0 case but there's really no point in ever connecting it via slow USB.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
Since Thunderbolt can be used to drive displays, I personally want it just so I can daisy chain two large displays off from one Macbook Pro. If it also means that I can plug peripherals into the monotor or onto the end of the chain, well that's just an added bonus. This is the one new feature on that has actually made me really want to upgrade sooner rather than later.
Apple often has a knack for promoting superior-but-niche technologies, and probably this is no different. But can we please let at least the USB connector design take its place alongside RJ-45/11 as things we don't screw around with? Personally I still miss the durable and reliable DB connectors, but that's a lost cause. Let video satisfy our need for connection anarchy *looks at the four incompatible video connectors on this PCs*.
"....that while Thunderbolt may offer twice the bandwidth of SuperSpeed USB, most people simply won't need it..."
Then MB drives were enough.
Then GB drives were enough.
Now TB drives are enough.
1 GB RAM was enough.
Then 2 GB RAM was enough.
,
.
.
See were I'm going?
Vote monkeys into Congress. They are cheaper and more trustworthy.
Of course not, but if you wanted to make something that appealed to the largest market possible, you probably would. Unfortunately, when a 3rd party manufacturer has to decide which connectivity option they want to use with their device, I'll bet you any amount of money they're going to pick the one with the largest market share regardless of capabilities.
Niche developers making niche products will do nothing for widespread adoption of the standard, just look at Firewire.
Just about every post is rated (-2, wrong). This makes me smile.
The thing with USB3 is that nobody needs it. USB2 is plenty fast for just about anything you want to connect over USB (external hard drives, usb sticks and keyboard/mouse). For everything else it plain sucks, you need a hub and the latency is horrible and the toll on the CPU for high data transfers is horrible so you're not guaranteed a certain speed. And it's not future-proof with a single 6Gb line where 10Gb has been in data centers for years now as an interconnect for fast data.
Thunderbird is what USB3 should've been - an extension of the PCIe bus directly on the CPU. Sure it's not widely implemented yet but it's pretty darn cool what the possibilities are even in the datacenter (for compute clusters)
Custom electronics and digital signage for your business: www.evcircuits.com
Another anonymous post clued me in: Somebody working in video production might be trying to get lossless or otherwise intraframe-only video out of camcorders and into editing equipment. At the highest professional level, this isn't 2K (1080p); this is 4K (2160p).
1080p video is far from 0.05 Gbps. 1080p/60 as transmitted over HDMI1.0 to 1.2 is 3.96 Gbps. HDMI 1.3 with deep color goes up to 10.2 Gbps (hmmm ... the same as Thunderbolt ... not surprising considering that is one of the interconnects that Thunderbolt is supposed to be able to do).
If you mean a compressed data stream encoding a 1080p video content, as transmitted over the air or via satellite or cable, that is much lower; somewhere under 20 Mbps. However, an hour of 720p or 1080i HD content, even compressed to hell using H.264, is somewhere around 1 GB. I know I sure spend a lot of time twiddling my thumbs slinging those files over either gigabit ethernet or USB 2.0 as I have to do presently. The difference between 4.8 Gbps and 10.2 Gbps for transferring an hour of 1080p is something like 2 minutes vs 1 minute. Sounds worthwhile to me, if I have, say 100 or 1000 hours of 1080p to transfer.
Here we are in 2011, the digital camera revolution is over and everyone (and their mother!) owns a digital camera, most of which will do 720p, the rest will do 640x480, even the crapy vivitar walmart ones for $20. It's difficult to find a cell phone or MP3 player that doesn't record video these days. I was part of a very technology/photography oriented group of friends growing up, and now many of us are starting families. Not once has any of them said "gosh, I am spending too much time transferring all this raw footage!". The great thing about pictures is that no post-processing is needed and can be shared instantly on facebook. Video is time consuming and only 2-3 people ever watch your home videos. "Consumer HD video" is the "video phone" technology of our generation. Everybody has access to it, but nobody actually wants to use it. You have a few amateur die-hards who actually put their products through their paces, but the vast majority of people will probably record 6-8 hours of memorable video throughout their child's life.
Sure, your new display technology can stream 2048p video at 4x realtime speed, but the use for that feature to the average consumer is vanishingly small. Daisy-chaining expensive prehiprials to a single port is a neat feature, but by generation 3 or 4 (think usb, firewire) you end up with 4-6 ports on the back, and another 2 on the front. The fact that the cable is limited to 3m doesn't help.
This would be neat if they released the fiber version, and I could keep my PC in the closet in the basement, far away from my mouse, keyboard, and two silent monitors. But instead we're tethered within 8 feet of the big PC box yet again.
moox. for a new generation.
It happened in networking too, with faster and faster ethernet being good enough, and cheap.
Wow, sent an e-mail as suggested when clicking on "use classic" banner, and got a fast response that addressed my msg
1080p video is far from 0.05 Gbps.
I was guessing based on the maximum bit rate of Blu-ray Disc video.
1080p/60
Feature films on BD are 1080p/24. As I understand it, feature-length motion pictures aren't produced in high motion because theaters aren't set up to show high motion. Pretty much the only thing that needs high motion is sports and perhaps overcranked effect shots that will be slowed down before playback.
3.96 Gbps
You have a point about uncompressed video, as might be used in video production. But how big would the RAID 10 array for saturating a 10 Gbps link need to be?
However, an hour of 720p or 1080i HD content, even compressed to hell using H.264, is somewhere around 1 GB. I know I sure spend a lot of time twiddling my thumbs slinging those files over either gigabit ethernet or USB 2.0 as I have to do presently.
Wouldn't a two-hour movie compressed to 2 GB (16 Gbit; 2.2 Mbps) take under twenty seconds to copy over a 1 Gbps link? Even assuming a slightly more generous 6 GB and 6.6 Mbps, it's still one minute to copy something that will take two hours to watch.
Sounds worthwhile to me, if I have, say 100 or 1000 hours of 1080p to transfer.
Apart from professional video production, where would one get these hundreds of hours of 1080p video?
There are lots of hopelessly optimistic people on /. with lower UIDs. :)
Note, I'm not one of them.
Why not chain the HDD off the display instead?
As long as the device you want to remove is at the end of the chain, you're golden. However, as the chain gets longer, the chances of the device you want to remove being the device at the end of the chain diminish.
I don't care if it's 90,000 hectares. That lake was not my doing.
Except the camcorder isn't likely to store the data on anything that can actually saturate either usb3 or thunderbolt so the use case is meaningless. So basically thunderbolt would be a useless interface for connecting to his SAN.
I have a hard time believing that people on Slashdot are actually calling it "SuperSpeed" USB, instead of just 3.0.
Seems to me, most Slashdotters should be aware that today's "SuperSpeed" is tomorrow's snail crawl. Open mouth, remove foot, please.
Sure I spend of lot of time twiddling my thumbs waiting for my files to transfer too, but the bottleneck isn't my 1Gb/s network. It's my slow sata drive! I would venture a guess that if you were able to fully utilize that 10Gb/s connection thunderbolt gives you, you'll be entirely limited by what's on either end. Only high end SAS arrays will be able to write/read that fast.
In other words the only way to utilize all that bandwidth would be to aggregate 100s of connections, ala an ethernet switch. Which is an entirely DIFFERENT market than what they're targeting for thunderbolt.
Aside from the spec sheet factoids that we all already knew, I see nothing else that's factually correct in that summary. Let's go through the rest of it:
To date, however, Apple is the only systems manufacturer to adopt Thunderbolt
Incorrect. Sony shipped the Vaio Z21 earlier this year with Thunderbolt, though it was branded under the old codename of Light Peak.
and it has done so as an additional device connectivity port
Incorrect. It's replacing mini-DisplayPort with Thunderbolt on the models that are being updated since they share the same connector. Doing so means no additional ports. Instead, since Thunderbolt can support different protocols, people who have DisplayPort devices can continue to use them without issue, and they'll have Thunderbolt there for later when more peripherals arrive.
keeping SuperSpeed USB on its computers.
Apple doesn't have SuperSpeed USB on any of its computers. They never have. The MacBook Pro, MacBook Air, iMac, and Mac mini have all been updated with Thunderbolt, yet they're all still using USB 2.0. The Mac Pro is their only computer which hasn't been updated since Thunderbolt debuted, and it certainly doesn't have USB 3.0 either. In fact, Apple is the only major holdout on USB 3.0. It even says so at the end of the introduction on Wikipedia's USB 3.0 page.
No other systems manufacturer has committed to Thunderbolt.
Besides Sony, which I've already mentioned, both Acer and Asus have committed to having Thunderbolt devices out in the future.
In contrast, SuperSpeed USB has been installed on 10 billion pieces of hardware, with numbers continuing to grow.
Incorrect, and entirely inane. Of course the numbers continue to grow. Was there a concern that they'd be shrinking? And 10 billion is patently false. The best numbers I could find in a quick search were that only 14M USB 3.0 devices were sold in 2010 (so you're only off by three orders of magnitude), and that projections by analysts peg sales at 1.7B per year by 2014.
Intel's codename for it was Light Peak, but just checking Wikipedia's page for Thunderbolt will confirm that Intel owns the registrations on the trademark for Thunderbolt, and they've switched to calling it Thunderbolt in all of their marketing material as of several months ago.
Ahh yes, so if you define "working" in a very specific way
A specific way for a specific need - high speed disk use.
Lots of people don't need high speed disks, and they would be fine with a standard USB drive. I use them myself for some things where speed is not needed.
People who actually by USB 3.0 storage devices are very likely to buy them exactly because of the speed. So then when they operate at a greatly reduced speed, they are not usable for the reason you bought them.
Think of it this way - if you buy a USB 3.0 drive and no matter what you do, it works at USB 2.0 speeds - would you keep or return it? Even though it's "working" according to YOUR definition.
The use of "fanboi" has jumped the shark when people accuse you of being a "fanboi" just because you want to use a technology for what it was meant to do!
"There is more worth loving than we have strength to love." - Brian Jay Stanley
Benchmarks of TB with both single and multiple streams showed it doing 10Gb/sec effective, both ways at the same time. That 10gb/sec *doesn't* include overhead. (20Gb/sec combined)
So, assuming maximum frame size, you will see an aggregate payload bandwidth of 10Gb/sec per direction.
Also, for other posts, TB isn't PCIe, it tunnels the PCIe protocol over it. PCIe slots are both a layer1 and layer2, but TB has its own layer 1 protocol, but implements the PCIe layer 2 protocol.
One of the reasons for Lightpeak was that USB is about at its max speed. It may have a few revisions left, but from what I've read, USB won't be able to get much further without breaking backwards compatibility. Once you lose backwards compatibility, you lose its greatest advantage. Or even if it does keep it, USB will have to trade some benefit for backwards compatibility, as nothing is "free".
Once you level the playing field by saying neither USB4 nor TB work with USB1-3, then which one do you choose? I'd say it's a choice weighted combination of versatility, cost, and performance. As it sits right now, TB is the underdog because of it's lack of install base and not much can make full use of it.. yet.. but if it is true that USB won't scale much further without breaking its greatest asset, then which will take over?
Also, lightpeak is meant to replace PCIe physical interfaces for Intel. We may start to see motherboards with no slots, but a cluster of fiber ports, then you just mount your cards(video/network/raid/etc) in the chassis. If intel pushes for this style, we may eventually see lightpeak in every computer anyway, then it's not a question of either/or. I'm sure we'll see both ports as fab processes will push the cost down to almost free in no time.
We'll have the BluRay vs HD-DVD for USB vs TB soon enough, except it'll be affordable and fairly convenient to own both, unlike two expensive optical disc formats.
Again, I'm not sure if it's true, so only time will tell.