Does USB-C signify the end of ThunderBolt? If so, where does that leave nMP?

[MacVidCards]

But far more north than USB overhead.

Ah, your learning.

USB has no less than 12 different adapters. When Thunderbolt has half of that, then let us talk.

Even if Manufactures could/would port over their existing products from Thunderbolt to USB-C, do you think it will be cheaper?.....No.

Ahh, having a hard time accepting the obvious I see.

No worries.

You do realize that "USB-C" is actually a different thing then "USB", yes?

TB2 devices already in the pipe may still pop out for awhile.

But if this new shiny thing can be made to fit Apple's purposes, TB is done.

It was already paddling upstream in a leaky canoe. Apple just served notice that the may be taking the paddle.

 

[mcnallym]

Thunderbolt = Pro Crowd who want the ability to connect the higher performance systems to there devices, i.e. things like this http://www.netstor.com.tw/_03/03_02.php?MTA4

USB-C = Consumer level where whilst perhaps not as fast or as expandable as Thunderbolt will be good enough for what consumer people want.

Lets face it with the Macbook specs then are you really going to be attaching Multibay Enclosures like the Netstor to it, packing it out with HDD Controller, additional Thunderbolt Audio devices etc to a CoreM device, so why pay the extra cost to put Thunderbolt in there?

The two will quite happily co-exist

 

[theluggage]

Thunderbolt = Pro Crowd who want the ability to connect the higher performance systems to there devices
USB-C = Consumer level

Fine, but if you establish that TB is for Pros only (and we're talking particular types of "Pro" here, not just anybody who buys a Mac with "Pro" in the name) then you need to start asking which future Mac models are going to support it. The only computer of any brand that really doesn't make sense without Thunderbolt is the nMP. What is the demand for, say, laptops that can attach to that monster RAID/PCIe enclosure you linked to?

I think USB-C is the end of Thunderbolt as a universal display and laptop docking connection. Thunderbolt basically becomes an external PCIe bus, on systems that suffer from the lack of PCIe. Third party displays are going to start sporting one-wire USB-C display, hub and charge connectors, and it sounds like USB-C is going to start offering DisplayPort 1.3 as soon as GPUs support it (while TB will have to wait for TB3, new connectors and, presumably, Intel integrated graphics with DP1.3 support).

That's the big problem: TB3 will restore the 'clear blue water' between TB and USB... but it isn't here yet.

Plus, the USB-C connector (as opposed to the USB 3.1 protocol) is potentially capable of carrying PCIe in the same way it carries displayport - so, if there's demand, future USB-C devices could compete with TB more directly.

I guess it would be theoretically possible for Thunderbolt to adopt USB-C as its new connector? Unlikely that Intel and the USB people would share their toys and play nicely, though.

 

[linuxcooldude]

You do realize that "USB-C" is actually a different thing then "USB", yes?

USB-C still uses the USB protocol, so its still USB, with yet another connector.

But if this new shiny thing can be made to fit Apple's purposes, TB is done.

They apparently already have on a consumer device thats targeting, mom, pop-grandma and not the power user/professional.

It was already paddling upstream in a leaky canoe. Apple just served notice that the may be taking the paddle.

Yet another knee jerk reaction just because they're using a new USB port with no Thunderbolt port. Because it makes no sense putting Thunderbolt on such an underpowered processor.

 

[slughead]

One is an external, independently controlled bus for PCI-E, one is USB.

USB-C still uses the USB protocol, so its still USB, with yet another connector.

USB-C can run 2 lanes of unencapsulated PCIe--just not in Apple's new laptop.. With PCIe that's the speed of TB2, with DP1.3 (which it can also do) that's TB3, with USB 3.1.

Seems at least similar. It's not survival of the fittest, it's survival of the "good enough." People who were previously hard-line about everyone being absolutely ready to go from PCIe to TB/2 are now stating there's no way they'll go from TB to USB.. Why the heck not? TB was a downgrade from PCIe and was astronomically more expensive for the same solutions (which were already on the market), USBC is only a mild downgrade by comparison, and will already ship with a lot of PCs.

Also: Any thoughts on security? In the PC world, they're still worried about ram dumps through TB (Apple patched up this problem, not sure other OS/hardware implementations have)... Will USBC have the same adoption hurdles?

 

[drewsof07]

USB-C can run 2 lanes of unencapsulated PCIe--just not in Apple's new laptop.. With PCIe that's the speed of TB2, with DP1.3 (which it can also do) that's TB3, with USB 3.1.

Seems at least similar.

I also thought TB3 would be capable of delivering power.

 

[deconstruct60]

It's funny, back in the day when people were asking about why there was no PCIe TB cards, you were among the first to point out that the standard required DP and how it was a kludge with discrete GPUs.

Again not really factual. I outlined how Thunderbolt naturally aligns with an embedded GPU but I did neither a. blur the distinction between a GPU and a GPU, "standard PCI-e socket" card (they are not the same) nor b. positioned that there were no solutions with discrete GPUs ( an exceeding dubious position since as I mentioned Apple shipped dGPU + TB solutions in the first batch of system the put on the market )

Yes, Thunderbolt controller chips have DP pins as one of their inputs. As long as that DP input signal is delivered to the controller's pins, integrated ( inside the CPU package and/or leverages the main RAM ) or discrete ( an independent GPU package with independent VRAM ) is largely immaterial. To the pins on the board soldered to is the critical issue. Intel from the beginning outlined that this was a logic board targeted solution.

The Rube Goldberg designs which push the DP signal off the logic boards, internal an external cable and than back down to the logic board level to get to the TB controller's input pins.... yeah that is goofy. Never was going to pass muster at an Apple design review. It is giant kludge.

Yes, Intel's integrated GPUs were even simpler (less components ) to implement in a TB system, but it is never the case they were the only elegant solution enabled as some protested.

There's no doubt that the inclusion of DP (and the requirement for an integrated GPU or a cable kludge) really limited TB adoption on conventional desktops/workstations.

The only requirement for a cable kludge is so that GPU card vendors can hock more more GPU cards. Those are the folks ( and their zealot customers ) are the ones to cling to the cable kludge solution. It is an odd set of external constraints that drive that as being a "good" , "well designed" solutions.

Put the DP signals on the logic board relatively near the set of PCI-e lanes pulled from the chipset ( PCI-e v2 x4 ) and Thunderbolt is highly straightforward to implement. If draw two inputs from the same board it is easy.

You get a giant disconnect when want to wrestle control and close proximity of the output connectors the TB controller is coupled with away from the controller. Either a. pump the DP output through some other connector first before pass through TB or b. block or impeed the other other input.

Conventional desktops from were impeded by stubbornness not Thunderbolt. The vast majority of desktop CPU packages that Intel has shipped over the last 3 years have had GPUs and DP out capabilities. All that was needed was to instead of shipping that to ancient VGA , or slightly less ancient DVI port the designers simply routed it to a TB controller. They refused. Apple didn't.

The whooping and goofy hand waving was largely consumed by wailing about how to get AMD and Nvidia after market cards looped into the system. Those two had no interest either in letting go of control of the DP signal in a standardized way or in weaving TB onto their products.

The doubt should be there. Since Intel has relaxed the constraints, has there been an exponential explosion of TB deployments? Nope. There is exceedingly little empirical evidence that doing so sooner would have had any material impact on adoption rates.

From what I understand, USB-C doesn't require display data like TB (it's completely optional), liberating it from this constraint.

There is no USB-C; it is a term that Apple seems to be promoted that likely will cause as much confusion long term as clarity (closely followed by "Thunderbolt bus"). There is Type-C connector. Yes, the alternate modes are all optional. It's "required even if you don't think you need it" is the USB 2.0 channel. Hence some weak footing behind merging "USB" and "C" into a noun.

One of the big additions that Type-C brings is a standardized mechanism for other protocols to share non-concurrent access to the high speed data lanes of the connector. Thunderbolt didn't want to create "yet another" connector. Then needed one which that standards body was willing to share physical properties with. USB turned them down (and actually were and still are too slow to exclusively cover long term goals) and DisplayPort folks said yes.

On systems with a minimized set of ports sharing is going to happen. It isn't a huge constraint to jump over.

 

[deconstruct60]

BTW, here's a good article on USB Type-C and DisplayPort Alternate Mode.

http://www.anandtech.com/show/8558/displayport-alternate-mode-for-usb-typec-announced

One thing worth noting is that USB-C alternate mode can use the 4 differential pairs dedicated for USB Superspeed for other protocols besides DisplayPort...

The huge short term problem in displacing Thunderbolt (and DP standard mini-DP connector ) is that there are 4 active pairs in the connection. USB Type C only has two active pairs. You use either the "A" side or "B" side, but not really both.

This is part why the Apple USB 3.0 and Video out ( HDMI / VGA ) sink to 1080p output limitations when put the Type-C connector in to a mode where DP can only get out on one single differential pair. ( and Apple's 4K max resolution is likely limited to just 30Hz).

The Display Port "alternate mode" is something Thunderbolt already had from the beginning. It is merely the legacy DP compatibility mode where DP signal is simply just passed through to the connector with a DP cable plugged in.

The function of the connection will be negotiated by the device when connected. So if this comes to fruition, TB will be redundant.

Not really because don't cover the same bandwidth and still isn't going to cover longer than "just to docking station" chains very well.

If USB-C can do the things TB does, for less money AND is flexible, ie you can implement which parts you want for different devices, who would ever want TB?

There is zero reason why a TB "light" can't join as yet another alternate mode.

Two factors from the anandtech article. First, is that you need a multimode mux and/or a-b switch as a backing to make the Type C connector multimodal. USB doesn't do it. The connector in and of itself doesn't do it. There will be a need to put a "capable enough" switch behind the connector to make all this happen.

As pointed out TB already has the DP legacy pass through mode built in. How hard would it be to add a USB 3.0 SuperSpeed pass through mode to the infrastructure of the controller? Not very hard a all. This is all bascially A-B-C crossbar switch stuff. ( not even like a Ethernet switch which as to multiplex the connector. It is either turn on or turn of like a keyboard/video/mouse switch only as higher throughput speeds and inputs from logic board traces not external cables. )

Second. there is a higher speed for fewer pairs trend.

http://www.anandtech.com/show/8972/...a-standard-display-stream-compression-arrives

eDP 1.4 is not so much so can attach 8K display to laptops as much as run a 2K-4K display with fewer wires. What does Type-C have? Fewer wires. A Type-C pragmatically only has 12. mini-DisplayPort has 20. With just one single different pair USB 3.0 is down to just a highly theoretical 2.5Gbps. Even 3.1 is back down to just the 5Gbps. A 40Gps TB with 20 pins is probably down to 10Gbps. As long as TB can do more with less ( and with an external transciever increasing the signal to noise ratio on the copper line that is likely to stay true for a long while) it still has a place.

And for whoever thinks TB is PCIE without overhead...try putting one of those new SM951 blades in a TB2 enclosure. It runs out of bandwidth well south of the claimed 20.

Sigh.... goof grief more FUD drivel. Thunderbolt bandwidth is not PCI-e bandwidth. PCI-e device traffic to a single destination on a TB network are gated as much by the x4 PCI-e v2.0 conectitivy at the single destination as by the actual TB bandwidth. TB bandwidth is higher because need a higher "back haul" bandwidth if want to manage latency and flow control issues ( http://en.wikipedia.org/wiki/Fat_tree ).

If put multiple x4 PCI-e drives on a single x4 PCIe connection.... duh it will get throttled. TB v3 will enable PCIe v3 connectivity. Some controllers probably at x4 PCI-e v3 which will handle multiple v2 drives. USB 3.1 is basically x2 PCI-e v2. Two x4 v2 drives will easily swamp it.

So TB3 can be counted on for falling short of the promised 40.

PCI-e v3 x1 is 8Gb/s. So yeah the top end controllers will be x4 PCI-e v3 ( around 32Gb/s ). It still will be based on fat-tree network design principles. When you string up a network of 2-4 TB v3 devices and 2-4 USB 3.1 devices you'll probably be able to tell why fat-tree principles have benefits too.

Even if USB-C is never quite as "good" as TB at some things, the ability to power devices will drown out other concerns at Apple.

TB over a Type-C connector will have the same power capabilities and the greater bandwidth and greater multiplexing ability over the same highly limited set of lanes available.

Conceptually, TB can actually get three protocols out on a Type-C connector. DP v1.2 , PCI-e v2 , and USB 3.x. USB can't. In and of itself the Type-C connector doesn't multiplex. Thunderbolt does. Right tool for right job is whether it will get placed or not.

Apple was trying to give TB an artificial boost by placing it on EVERY machine they sold. They have now served notice that it WON'T be on every new machine. And there goes the one piece of continuing momentum. When TB3 comes out with new connectors and everyone needs a fleet of adapters to connect old and new things is when the real fun will begin.

More FUD. Two doses here. First, all but the Mac systems except the Mac Pro actually clearly needed Thunderbolt. It was not "artificial" for Apple to roll it out for the vast majority of its lineup. In terms of Macs sold and deployed and even smaller "artificial" justification as the classic Mac Pro was such a small and rapidly shrinking subset.

The new Mac Pro constraints is far more so driven by the real need for the Mac Pro to be a Mac as opposed to different for different sake ( or to justify some 3rd party business interest).

The second does is that Intel ( and USB-IF) haven't really released the full details on TB v3 yet. Just like there was single connector variants of TB controllers, there may be a dual connector path for Thunderbolt going forward. One simpler, cheaper connector ( Type C alt Thunderbolt ) and another "maximum bandwidth" mini-DP.

The DisplayPort folks didn't announce they were scrubbing all other DP connectors in any of the "Type C alternate mode" announcements. If they don't scrap mini-DP connectors then there is little good reason for Thunderbolt to walk away either. Having cabling with Type-A on one side and Type-B on the other side didn't kill USB. It probably wouldn't kill Thunderbolt either.

This whole "single cable to rule them all" is even more on heady duty hallucinogen drugs than the "single connector" notion. Only the zealots and FUD clowns are pitching that nonsense.

Even if Thunderbolt somehow goes 'rogue' Type-C and concurrently uses both 'A" and "B" sides, you'd still probably need new cables because can't use the transceivers at the high bandwidth rates required for the dramatically higher speeds. ( TB v1 to v2 was basically the same raw cable speed. The aggregation, mux, and underlying protocol transport was adjusted. No new transceivers needed. ). You can't run 10Gps Ethernet on old cat-5 (or older) cable either. Over time, faster means new cables eventually. Eventually, Wailing about new cables is wailing about wanting to be stuck in the past.

 

[VirtualRain]

The huge short term problem in displacing Thunderbolt....

TLDR;

So is Apple going to remain committed to TB or not?

 

[deconstruct60]

TLDR

If don't read, probably won't comprehend much.

So is Apple going to remain committed to TB or not?

Probably Yes. Nothing in Apple's recent roll out blocks that the 1-2 revisions down the road the MacBook having a TB mode. Apple added TB v2 to the MBA. Does that scream abandonment? No.

For the future MacBook is more a matter of space availability ( additional TB controller next to Type-C connector ), provisioning PCIe v3 lanes, and a version of a TB approved connector that will fit the device. Space should come with time ( logic process will get smaller allowing more stuff in less space) and TB v3 ( smaller z height ) options are coming. It just isn't going to come in 2015. Big deal. Coming in 2015 really doesn't say jack squat about commitment to Thunderbolt long term or not. Even more so in the light that word is already out that a smaller z-height option is coming with Thunderbolt v3 ( ~3mm http://www.extremetech.com/computin...-and-100w-power-delivery-for-single-cable-pcs ). Provisioning more PCI-e v3 lanes will come with next iteration of Core M processor in 2016. There are no long term problems here with weaving in TB.


Type-C " Receptacle opening: ~8.4 mm x ~2.6 mm "
http://www.anandtech.com/show/8377/usb-typec-connector-specifications-finalized

2.6mm and 3.0mm are about the the same size. Both are on track with Apple's "thinner" design constraints.

Type-C supports alternate modes of DP . TB v3 supports superset of those modes ( USB 3.1 , TBT , DP 1.2 (if pure passthrough 1.3 isn't a stretch since simple, very short crossbar), HDMI 2.0 via conversion ). New connectors for both, really not clear advantage either way.

Type-C supports 100W. TB v3 will support 100W. Nothing particularly differentiating here either.


It is highly likely that TB will either leverage Type-C or minimally there will be a very close Type-C like clone for TB. So "thinner and newer" doesn't block TB at all.


Apple probably will eventually go with USB 3.1. Is there USB 3.1 on Mac systems now? Nope. If Apple doesn't jump on board 2015 is that doom and gloom for USB 3.1? Nope. ( Remember back in early Thunderbolt days when genius predictions of USB 3.0 is never coming to Macs was being thrown around? Same fundamentally flawed inferenced BS now about TB going away. ) When Intel weaved USB 3.0 into the core chipset, Apple followed. Same is likely going to be true for vast majority of Mac systems over next two years. Stuck on USB 3.0 because the default chipsets are stuck on USB 3.0. Nothing "Apple hates USB 3.1" about it at all. TB isn't on the MacBook for largely same core reasons USB 3.1 isn't on MacBook; provisioning requires future chipsets.



Apple is going to hand wave the USB 3.1 gap by doing a bunch of marketing two-stepping about USB 3.1 gen 1. That is just USB 3.0, but it sells the illusion that Apple is delivering 3.1 They aren't. What they are probably saying though is Apple is only going to deliver USB 3.1 with Type-C connectors. ( Technically they don't have to but they probably will). They are also likely doing a shot across the bow that Type-A is dead connector standard walking. Just like Apple didn't do a mix of USB 2.0 and USB 3.0 sockets when transition came. Uniformity and only "as many as necessary" is likely design constraints.

Thunderbolt is probably not going into the Intel core I/O chipset. Its placement requirements next to the connector and adoption level don't really justify it. That won't stop Apple from using it. Type-C actually drives the need for a switch on the backend of the connector. Thunderbolt has had and will continue to have a switch as a core component. It fits the need and adds more value ( higher multiplexing ). Where is the huge justification for dropping it??? Race to the bottom system pricing? Race to the bottom cable prices? Systems with fewer ports that also do alot less?


For Mac Pro specifically (since this is a Mac Pro thread)? I suspect Apple will not be an early adopter of USB 3.1. The underlying chipset isn't going to pick it up for multiple years and the Mac Pro is short on PCI-e lanes already ( going from a x1 to x2 USB controllers is spending/diluting lanes they don't have. ). When 3.1 does come to the Mac Pro it extremely likely will come in Type-C connectors though. Folks highly dependent upon Type-A only license key dongles should be activity talking to the software vendors hooked on those


I somewhat doubt that Mac Pro's will be first/early adopter of TB v3 either. Again a PCI lane consumption problem. x8 PCI-e v3 can cover 3 * x4 PCI-e v2. It won't cover 3 * x4 PCI-e v3. The Xeon E5 aren't going to move to PCI-e v4 for another iteration or two.

The Mac Pro has higher needs right now. Better OpenCL 2.x and SPIR support (i.e., newer GPUs). Better SSD. Better AVX 2 support.

With three TB controllers the higher 40Gps isn't all that urgent when sitting on overall 60Gps right now. 40 on fewer pairs make far more a difference when suck with just 1-2 connectors versus 6. 4 USB 3.0 ports will largely work just fine with USB 3.x devices.

And no, the Mac Pro not being the "first" for every new technology doesn't say anything new about Apple's strategic decision making.

 

[MacVidCards]

If don't read, probably won't comprehend much.



Probably Yes. Nothing in Apple's recent roll out blocks that the 1-2 revisions down the road the MacBook having a TB mode. Apple added TB v2 to the MBA. Does that scream abandonment? No.

For the future MacBook is more a matter of space availability ( additional TB controller next to Type-C connector ), provisioning PCIe v3 lanes, and a version of a TB approved connector that will fit the device. Space should come with time ( logic process will get smaller allowing more stuff in less space) and TB v3 ( smaller z height ) options are coming. It just isn't going to come in 2015. Big deal. Coming in 2015 really doesn't say jack squat about commitment to Thunderbolt long term or not. Even more so in the light that word is already out that a smaller z-height option is coming with Thunderbolt v3 ( ~3mm http://www.extremetech.com/computin...-and-100w-power-delivery-for-single-cable-pcs ). Provisioning more PCI-e v3 lanes will come with next iteration of Core M processor in 2016. There are no long term problems here with weaving in TB.


Type-C " Receptacle opening: ~8.4 mm x ~2.6 mm "
http://www.anandtech.com/show/8377/usb-typec-connector-specifications-finalized

2.6mm and 3.0mm are about the the same size. Both are on track with Apple's "thinner" design constraints.

Type-C supports alternate modes of DP . TB v3 supports superset of those modes ( USB 3.1 , TBT , DP 1.2 (if pure passthrough 1.3 isn't a stretch since simple, very short crossbar), HDMI 2.0 via conversion ). New connectors for both, really not clear advantage either way.

Type-C supports 100W. TB v3 will support 100W. Nothing particularly differentiating here either.


It is highly likely that TB will either leverage Type-C or minimally there will be a very close Type-C like clone for TB. So "thinner and newer" doesn't block TB at all.


Apple probably will eventually go with USB 3.1. Is there USB 3.1 on Mac systems now? Nope. If Apple doesn't jump on board 2015 is that doom and gloom for USB 3.1? Nope. ( Remember back in early Thunderbolt days when genius predictions of USB 3.0 is never coming to Macs was being thrown around? Same fundamentally flawed inferenced BS now about TB going away. ) When Intel weaved USB 3.0 into the core chipset, Apple followed. Same is likely going to be true for vast majority of Mac systems over next two years. Stuck on USB 3.0 because the default chipsets are stuck on USB 3.0. Nothing "Apple hates USB 3.1" about it at all. TB isn't on the MacBook for largely same core reasons USB 3.1 isn't on MacBook; provisioning requires future chipsets.



Apple is going to hand wave the USB 3.1 gap by doing a bunch of marketing two-stepping about USB 3.1 gen 1. That is just USB 3.0, but it sells the illusion that Apple is delivering 3.1 They aren't. What they are probably saying though is Apple is only going to deliver USB 3.1 with Type-C connectors. ( Technically they don't have to but they probably will). They are also likely doing a shot across the bow that Type-A is dead connector standard walking. Just like Apple didn't do a mix of USB 2.0 and USB 3.0 sockets when transition came. Uniformity and only "as many as necessary" is likely design constraints.

Thunderbolt is probably not going into the Intel core I/O chipset. Its placement requirements next to the connector and adoption level don't really justify it. That won't stop Apple from using it. Type-C actually drives the need for a switch on the backend of the connector. Thunderbolt has had and will continue to have a switch as a core component. It fits the need and adds more value ( higher multiplexing ). Where is the huge justification for dropping it??? Race to the bottom system pricing? Race to the bottom cable prices? Systems with fewer ports that also do alot less?


For Mac Pro specifically (since this is a Mac Pro thread)? I suspect Apple will not be an early adopter of USB 3.1. The underlying chipset isn't going to pick it up for multiple years and the Mac Pro is short on PCI-e lanes already ( going from a x1 to x2 USB controllers is spending/diluting lanes they don't have. ). When 3.1 does come to the Mac Pro it extremely likely will come in Type-C connectors though. Folks highly dependent upon Type-A only license key dongles should be activity talking to the software vendors hooked on those


I somewhat doubt that Mac Pro's will be first/early adopter of TB v3 either. Again a PCI lane consumption problem. x8 PCI-e v3 can cover 3 * x4 PCI-e v2. It won't cover 3 * x4 PCI-e v3. The Xeon E5 aren't going to move to PCI-e v4 for another iteration or two.

The Mac Pro has higher needs right now. Better OpenCL 2.x and SPIR support (i.e., newer GPUs). Better SSD. Better AVX 2 support.

With three TB controllers the higher 40Gps isn't all that urgent when sitting on overall 60Gps right now. 40 on fewer pairs make far more a difference when suck with just 1-2 connectors versus 6. 4 USB 3.0 ports will largely work just fine with USB 3.x devices.

And no, the Mac Pro not being the "first" for every new technology doesn't say anything new about Apple's strategic decision making.

Have to agree on tldr

Too many words.

If you actually have a point, try making it in fewer words. Basic debate class stuff from high school.

Before I dozed off I read something about putting multiple PCIE 2.0 x4 lane drives over TB to saturate a TB2 connection and THAT is when I quit reading.

See, the difference between us is I actually test stuff and you wheel out enormous walls of text that you copy & paste from tech articles trying to impress people with the "brilliant but socially unable to communicate college professor" vibe.

I can take a single (1) PCIE SSD and run it in a cMP and achieve 1500 MB/s read and write.

If I put it in a TB2 enclosure and connect to nMP It gets strangled down to 1,200-1,300.

Barefeats got same numbers. So there is TB2 throughput. Not theoretical horse twaddle read from a tech blog, just what it actually does.

 

[DonMega]

Have to agree on tldr

Too many words.

If you actually have a point, try making it in fewer words. Basic debate class stuff from high school.

Before I dozed off I read something about putting multiple PCIE 2.0 x4 lane drives over TB to saturate a TB2 connection and THAT is when I quit reading.

See, the difference between us is I actually test stuff and you wheel out enormous walls of text that you copy & paste from tech articles trying to impress people with the "brilliant but socially unable to communicate college professor" vibe.

I can take a single (1) PCIE SSD and run it in a cMP and achieve 1500 MB/s read and write.

If I put it in a TB2 enclosure and connect to nMP It gets strangled down to 1,200-1,300.

Barefeats got same numbers. So there is TB2 throughput. Not theoretical horse twaddle read from a tech blog, just what it actually does.

No respect for anyone who has to rip on someone to make themselves feel good or needed or valuable. Pathetic.

 

[MacVidCards]

No respect for anyone who has to rip on someone to make themselves feel good or needed or valuable. Pathetic.

I see you didn't read the Great Wall of China text either. Plenty of ripping on others in there trying to make himself look like a genius.

 

[linuxcooldude]

Have to agree on tldr

Too many words.

If you actually have a point, try making it in fewer words. Basic debate class stuff from high school.

Before I dozed off I read something about putting multiple PCIE 2.0 x4 lane drives over TB to saturate a TB2 connection and THAT is when I quit reading.

See, the difference between us is I actually test stuff and you wheel out enormous walls of text that you copy & paste from tech articles trying to impress people with the "brilliant but socially unable to communicate college professor" vibe.

I can take a single (1) PCIE SSD and run it in a cMP and achieve 1500 MB/s read and write.

If I put it in a TB2 enclosure and connect to nMP It gets strangled down to 1,200-1,300.

Barefeats got same numbers. So there is TB2 throughput. Not theoretical horse twaddle read from a tech blog, just what it actually does.

Ok, so you can't keep up, fair enough. ( Strangely, no tech articles seem to pop up when I randomly copy paste his posts on Google )

1,200-1,300MB/s is quite respectable on an external PCIe type bus like Thunderbolt.

It would be quite interesting to see the results using it on the USB 3.1 bus.

 

[rdav]

nMP(7)2015: Replace USB-3 with USB-C /?

It's seems likely that the nMP(7)2015 will retain the current six TB2 ports. (Since TB3 will probably not be widely available until some time in 2016). And with enhanced GPUS, they would be used for existing monitors etc.

But could replace the USB-3 ports with USB-C (x4). Not for Power, but for enhanced peripheral support & data speeds. Hopefully, USB-C to USB-3 hubs will be available soon, to make this transition easier.

 

[deconstruct60]

....
But could replace the USB-3 ports with USB-C (x4). Not for Power, but for enhanced peripheral support & data speeds.

USB Type-C connectors don't necessarily bring data speed increases. The actual 3.1 data increases are not only provisioned through Type-C. The MacBook is pragmatically (from a data perspective ) straight USB 3.0; same as previous Mac laptops.

http://arstechnica.com/apple/2015/03/explaining-the-usb-3-1-gen-1-port-in-the-retina-macbook/

[ The corner issues which drives the USB 3.1 Gen 1 labelling is for the need to enhanced cabling/adapters to do the correct handshake ID and autoconfiguration that Type-C opens the door for. The subset of the USB 3.1 update that pertains to Type-C is not the whole of 3.1. Neither the other way around. The whole of the changes arriving with 3.1 are not the SuperSpeed+ data rate improvements (another subset ). ]

Hopefully, USB-C to USB-3 hubs will be available soon, to make this transition easier.

USB Type-C Connectors in and of themselves aren't really something that "hubs". May have hubs with one "host side" Type-C connector that has Type-A connectors for the "downstream" side fan out ports, but whether that is distributing USB 3.0 or not depends upon what the Type-C connector is hooked to internally. There is no requirement that there is USB 3.0 there. (probably is in most implementations, but not required. )

The "to USB-3" (really probably just more Type A ports ) runs counter to the claim that this is an enhanced (expanded) peripheral support move.

Eventually Apple will put USB 3.1 data rate capability inside some future generation Mac Pro. When that happens it will extremely likely be through Type-C connectors. ( Type-A is 15+ years old. Apple walks away from legacy stuff after a while) However, it is doubtful that is going to happen in 2015 ( or 2016 ) if there is any systems upgrade. For ultra slim MacBooks there is upside in USB 3.0 through a Type-C connector. For a Mac Pro there are no upsides.

TB v3 will likely show up on a future Mac Pro before USB 3.1 does. TB v3 could bring a Type-C clone connector with it if that connector fully unlocks TB v3's max data rates.

 

[Steve USB]

The huge short term problem in displacing Thunderbolt (and DP standard mini-DP connector ) is that there are 4 active pairs in the connection. USB Type C only has two active pairs. You use either the "A" side or "B" side, but not really both.

Type C has 4 high speed pairs, not 2, it can support two USB3.1, one USB3.1 and 2 lanes of DP, or 4 lanes of DP.

Not really because don't cover the same bandwidth and still isn't going to cover longer than "just to docking station" chains very well.

Type C was designed for 20Gb, USB 3.1 Gen 2 is currently 10Gb. DP 4.something Gb. Plenty of bandwidth.

Type-C pragmatically only has 12. mini-DisplayPort has 20. With just one single different pair USB 3.0 is down to just a highly theoretical 2.5Gbps. Even 3.1 is back down to just the 5Gbps. A 40Gps TB with 20 pins is probably down to 10Gbps. As long as TB can do more with less ( and with an external transciever increasing the signal to noise ratio on the copper line that is likely to stay true for a long while) it still has a place.

Type C has 24 pins, 4 grounds, 4 power (and RF ground) 4 high speed pairs (8 pins), 4 pins for USB2, 2 for control, and 2 that become the Aux for DP.

USB3.1 Gen 1 is 5Gb (4Gt), Gen 2 is 10Gb (10Gt).

TB over a Type-C connector will have the same power capabilities and the greater bandwidth and greater multiplexing ability over the same highly limited set of lanes available.

Same power capabilities?!? Type C can supply 20V at 5A, 100W. No need for an additional power connection.

Type C is a single connection dock.

Conceptually, TB can actually get three protocols out on a Type-C connector. DP v1.2 , PCI-e v2 , and USB 3.x. USB can't. In and of itself the Type-C connector doesn't multiplex. Thunderbolt does. Right tool for right job is whether it will get placed or not.

Then why did Apple join USB to help develop Type C, that is so close in capabilities to Thunderbolt? Why did Apple support the decision to allow Alternate Modes, and drive the spec for Display Port as the first Alternate Mode?

The second does is that Intel ( and USB-IF) haven't really released the full details on TB v3 yet.

And if they don't....

This whole "single cable to rule them all" is even more on heady duty hallucinogen drugs than the "single connector" notion. Only the zealots and FUD clowns are pitching that nonsense.

Count me as one of those zealots

You can't run 10Gps Ethernet on old cat-5 (or older) cable either.

Actually you can, just for a shorter distance, in a bundle (crosstalk between cables). Out of the bundle further.

One of the speed limits with Type C is, of the 20dB loss budget, 14 dB is in the devices (7dB per device), with only 6dB allocated for the cable. Move the chips closer to the connector, and you can run the cable faster.

 

[linuxcooldude]

Type C has 4 high speed pairs, not 2, it can support two USB3.1, one USB3.1 and 2 lanes of DP, or 4 lanes of DP.

Seems he is saying while it has 4 pairs, only two active pairs at one time.

 

[Steve USB]

Seems he is saying while it has 4 pairs, only two active pairs at one time.

?!? The only time only 2 pairs are active is when only a single USB 3.1 port is in the device (computer or peripheral).

Currently DP & USB are the only approved modes, so: Two USB 3.1, one USB3.1 and 2 lanes of DP, or 4 lanes of DP, use all 4 lanes.

 

[MH01]

To be honest, Apple has a history of introducing standards and yet not implementing them across thier range. The problem is that USB-C needs dongles, until it becomes mainstream , if it becomes mainstream, right now it's kind of an expansive early adopter journey.

 

[AdonisSMU]

So basically Thunderbolt is dead. That was fast. I was hoping it would last a bit longer and people would use it.

 

[MacVidCards]

So basically Thunderbolt is dead. That was fast. I was hoping it would last a bit longer and people would use it.

Hence me starting the thread. This is how I see it.

 

[AdonisSMU]

Hence me starting the thread. This is how I see it.

I would've quite liked a new display for at home as I have a 13" rMBP and sometimes it's nice to have a larger display. With the new 12" rMB I wonder why Apple just released a new 13" machine given that the writing is on the wall with TB and Lightening? Thunderbolt doesn't seem like it's been out very long for them to already be discontinued but it would seem like it hasn't really taken off outside of MacWorld.

----------

To be honest, Apple has a history of introducing standards and yet not implementing them across thier range. The problem is that USB-C needs dongles, until it becomes mainstream , if it becomes mainstream, right now it's kind of an expansive early adopter journey.

It does feel like USB-C generation too early....just like the rMB. Apple does tend to release get some data let some time pass then to a MAJOR upgrade on the second version of said product and that's the version that lasts for awhile and then back to more incremental upgrades.

 

[ha1o2surfer]

Wow, totally agree.

Again.

Got to be a record.

If USB-C can do the things TB does, for less money AND is flexible, ie you can implement which parts you want for different devices, who would ever want TB?

And for whoever thinks TB is PCIE without overhead...try putting one of those new SM951 blades in a TB2 enclosure. It runs out of bandwidth well south of the claimed 20. So TB3 can be counted on for falling short of the promised 40.

Even if USB-C is never quite as "good" as TB at some things, the ability to power devices will drown out other concerns at Apple.

They are happy to chuck the baby with the bathwater to gain thinness in their machines, been proven many times. That combined with the love of "fewer cables" led to the amusing combo of the G4 Cube....running a 17" ADC CRT through it's own power supply so the display would have just 1 cable. Anyone with some knowledge of electronics and the delicate Cube's brick & VRM would cringe at the combo.

Apple was trying to give TB an artificial boost by placing it on EVERY machine they sold. They have now served notice that it WON'T be on every new machine. And there goes the one piece of continuing momentum. When TB3 comes out with new connectors and everyone needs a fleet of adapters to connect old and new things is when the real fun will begin.

Thunderbolt has a lot of overhead you are right and most people don't know this.. Like your example. I use a GTX 970 over a TB1 connection and I can't get anywhere near 1GB/s for memory copies via CUDA. MAYBE 750MB/s if I'm lucky..... TB isn't as flexible as one may think; and what I mean by that are the two giants behind this tech....

The minute I saw Apples little Mac Pro Thunderbolt diagram about what ports are on which bus and how many monitors you can have on each really seemed to rub me the wrong way. I had to carefully layout which monitor was on which port and what devices I had on which bus or some monitors wouldn't work... idk who is to say USB 3.1 will have similar problems.. I just don't know. I like the olden days

 

[flowrider]

And another article:

http://appleinsider.com/articles/15...-c-for-its-all-new-12-inch-macbook-and-beyond

Lou