Intel Drops Thunderbolt 3 Royalty, Adds CPU Integration and Works Closely With Microsoft (windowscentral.com) 107
An anonymous reader quotes a report from Windows Central: Over the last few days, Thunderbolt 3 has been a hot topic amongst Windows users especially with its notable absence with the new Surface Pro and Surface Laptop. Part of the problem is adoption, integration, cost, and consumer confusion according to Microsoft. Intel is aware of the current roadblocks to Thunderbolt 3 implementation, which adds 40Gbps data transfers along with charging and display support for USB Type-C. Today, the company announced numerous changes to its roadmap to speed up its adoption, including: Dropping royalty fees for the Thunderbolt protocol specification starting next year; Integrating Thunderbolt 3 into future Intel CPUs. The good news here is that Intel is dropping many of the roadblocks with today's announcement. By subtracting the licensing costs for Thunderbolt 3 and integrating into the CPU, Intel can finally push mass adoption. Getting back to Microsoft, Intel noted that the two companies are already working closely together with the latest Creators Update bringing more OS support for the protocol. Roanne Sones, general manager, Strategy, and Ecosystem for Windows and Devices at Microsoft added that such cooperation would continue with even more OS-level integration coming down the road.
Re:Huh, someone was paying attention to Firewire (Score:5, Insightful)
I would hope so. I'd love it if the USB-C/Thunderbolt port became the new standard port that's built into everything and is used for everything. For any device you have, you will only need to plug in a single cable to a port following a single standard (excepting when you need an additional power cable). I'd even like to see it used on servers. I could see a scenario where every server in a rack is plugged into a single Thunderbolt switch/hub that provides networking, DAS/NAS/SAN, KVM, lights-out management, everything.
Re: (Score:2, Informative)
Your wish will never happen because Intel chips will still be the only CPU's allowed to use it. Intel's problem with Thunderbolt has always been that it won't license it on FRAND terms to all comers (including AMD), as a result even if it's cheaper it will be integrated into nothing. USB-3 license costs are pennies and there is no requirement on who you are to use it.
Thunderbolt has massive restrictions on WHO is allowed to use it, even if they drop the price to the same as USB they will never allow it to b
Re:Huh, someone was paying attention to Firewire (Score:5, Informative)
Thunderbolt has massive restrictions on WHO is allowed to use it..
I wish the TFS would've linked to Intel's actual press release [intel.com]. The article that the summary linked to left out one of the biggest pieces: "...next year Intel plans to make the Thunderbolt protocol specification available to the industry under a nonexclusive, royalty-free license..." which then goes on to explicitly state they want third-party chipset makers producing Thunderbolt chipsets. This announcement is NOT about them letting motherboard makers slap the port and an Intel chipset on their board royalty-free. This is about letting other chipset makers produce their own TB chips royalty-free. They are basically releasing it industry-wide, and it will no longer be limited to Intel chipsets.
So yes, expect to see AMD-compatible motherboards sporting TB3 support. I'd be more doubtful that AMD will be able to license it for integration into their CPUs (though they might; certainly wouldn't be remotely the first Intel technology they licensed to put directly in their silicon), but the new licensing does not preclude third-party chipsets being put onto AMD-supporting motherboards. And I completely expect the likes of Asus, Gigabyte, ASRock, etc. to do this. Probably on their high-end/gamer boards only at first, but still.
Re: (Score:1)
And I completely expect the likes of Asus, Gigabyte, ASRock, etc. to do this. Probably on their high-end/gamer boards only at first, but still.
Do high end gamer boards have sockets for AMD? I am pretty sure I had a motherboard with an AMD 8088 chip on it. It wasn't a gamer board. I think my PC Junior had an Intel chip.
Re: Huh, someone was paying attention to Firewire (Score:1)
Re: (Score:1)
The best way to get something universally adapted is to make that something universally available. Sure monopolizing a thing may be profitable in the short run, but for the long haul only openness matters.
Re: (Score:1)
Yeah, just look at that marvel of pursuing open standards, the Apple Inc, with that lovely $800 billion market capitalization and $300 billion-ish pile of cache. Opennes works, clearly.
Re: (Score:3)
Do you mean Apple deciding to ditch their proprietary ADB in favour of USB?
Apple adopting using UNIX as the core of their OS and having it certified as conforming to the Single UNIX Specification?
Apple using an implementation of the OpenStep specification as their GUI app development framework?
Apple replacing ADC with DVI?
Apple contributing their mini DisplayPort plug to the DisplayPort consortium, royalty free?
Apple replacing their proprietary MagSave power co
Re: (Score:2)
I would guess he means the old iPhone and current lightning connectors.
Re: (Score:3)
Apple Inc, with that lovely $800 billion market capitalization and $300 billion-ish pile of cache.
Good Lord, Apple! Clean your cache!
Re: Huh, someone was paying attention to Firewire (Score:1)
Re: (Score:2)
Probably what will happen is rackmount servers will have two or even three TB3 ports, each with 100w, that way you have redundancy when the port or cable fails (on either end) on both power and network. Failure of one or more cables would probably allow the server to run in reduced power mode indefinitely.
I don't think we're far out from phones having 256gb drive and 8gb ram as normal, at which point you just plug it in to a KVM kiosk and use that as both your cell phone and desktop. Even a year ago
cell phone and desktop they need to dump roms (Score:2)
cell phone and desktop they need to dump roms then / make it so that your phone carrier can't per load and lock in crap + slow down updates.
Works closely with... (Score:3)
Re:Works closely with... (Score:5, Interesting)
The NSA?
Re: (Score:1)
Re: (Score:3)
Probably. Thunderbolt is great for spies and law enforcement. DMA access to the whole address space, just rip those encryption keys and passwords right of of RAM.
Best to avoid anything like that: Thunderbolt, PC Card, PCMCIA, Firewire, hot plug PCI(e).
Re: (Score:2)
As if Wintel hasn't been a term since at least the mid 90s.
To wait, or not to wait... (Score:3)
So the question for someone looking to buy a revised MacBook Pro this year would be, buy it now for a battle-tested Thunderbolt 3 connection, or wait for the chip integration for performance gains even though it will be a fist gen thing next year...
Re: (Score:3)
I wouldn't wait on integration. Even after it gets integrated into the CPU how long will it take Apple to switch to that CPU model?
Then how long will it take for them to change motherboards to ones that support on-chip?
THEN exactly how long will it take until the peripherals perform well enough for it to make a difference?
Sure it's probably better to have the on-chip version, but the rest of the ducks are going to take their time lining up.
Re: (Score:3)
That was the problem with FireWire. FireWire was supposed to be able to do nearly everything we're doing with Thunderbolt (just not as fast). The problem was the people making the peripherals rarely made them right. Theoretically I should be able to plug a FireWire cable from my computer, to a hard drive, to an external CD drive, to another computer. Both computers should have access to both peripherals and they should be able to network with one another over the cable.
I was never able to get two comput
Re: (Score:2)
The drive manufacturers deliberately didn't implement those parts of the protocol, because the computer operating systems didn't support sharing drives in that manner. Two different computers with write access to the same drive, the same filesystem at the same time... Your data will be corrupted pretty damn fast.
OS manufacturers were not interested in solving a relatively tricky problem, when anyone who wants that kind of thing can just use a NAS. Okay, limited to gigabit ethernet speeds, but not worth the
Re: (Score:2)
I will argue that had they actually solved it we would be using FireWire revision H or so by now and they company that actually solved how to make it work cheaply would have made so much money licensing we would have a new power player in the data market (unless of course it was an existing power player).
Re: (Score:2)
The NAS idea isn't so bad, considering that FireWire supported networking anyways and it doesn't have to be I.P. networking to be networking (remember protocols before the whole world jumped on the TCP I/P bandwagon? some weren't so bad IPX excelled in LAN setups without having to configure anything). Also I had a couple of NAS drives from the era that were natively Ethernet and they had dismal read/write speeds that didn't come anywhere close to saturating a 10 Mbps connection, much less the 100 Mbps they
Re:To wait, or not to wait... (Score:5, Informative)
If you are looking for a MacBook Pro with a discrete video card, then I would not wait because of this [1]. Apple has done a lot of work with Intel on integration of the Thunderbolt chips to allow for the mux'ing of the discrete and integrated video streams. My guess is that Apple will continue to use the parts that use the external chips to preserve that work, at least on those computers that have discrete and integrated video parts[2].
[1] At this point you would have to be nuts to buy any Apple product in the next three weeks. Wait until after Tuesday of WWDC (major stuff is announced Monday, then minor bumps come out on Tuesday), then evaluate what you are going to buy.
[2] Technically some iMacs have both, but I would exclude them from this list as it is missing this mux'ing system since the screen is only ever driven by the discrete part. There are some decode functions that are used in the integrated part, but that never goes anywhere but back across to main memory.
Re: (Score:1)
Thanks, with the question I actually meant to compare buying after the imminent update vs next year, not right away. :-)
Good point about the video aspect, I do need a model with a discrete card (and am really hoping for a boost in that regard in the update).
If you are willing to go non-Mac (Score:2)
Have a look at a Gigabyte Aero 15. It is about half a pound heavier than a MBP but still pretty light n' thin, and with that you get a GTX 1060. This generation of nVidia mobile cards almost exactly match their desktop counterparts specs wise, so that's a lot of dGPU power.
Re: (Score:1)
Although I appreciate the thought (I would dearly love the GTX or any nVidia chipset), I need a Mac as most of my work is IOS related and while I'd be OK with a hackintosh as a desktop, I can't go that way for a laptop...
I'm hoping they switch chipsets for the update having come to there senses that a lot of devs wan nViidea chips, we'll see.
Also can't really do heavier as I travel a lot, already debated downsizing to a 13" or Air, and getting a desktop (but then it probably would be a Hackintosh deal so I
Re: (Score:2)
Re: (Score:2)
I have never used a touchpad like the one included in MacBooks. I don't know if it's the hardware or the driver, so perhaps a Hackintosh with the same hardware would work just as well, but I have a feeling that isn't the case.
Re: (Score:2)
Re: (Score:3)
"Buy now or wait for something better" is a dilemma faced by anyone who has ever bought any computer.
Re: (Score:1)
Yes, but... this adds one more potential brick on the side of waiting if it would help improve system performance for those using Thunderbolt 3 for heavy I/O (like an external storage device striping together a few SSD's). That's the main thing I was wondering about missing out on, beyond all the usual points about storage and processor and GPU increasing in speed.
It sounds like it's not significant enough to wait on if you were thinking about buying one of the updated MacBook Pro models due to come out so
Re: (Score:2)
This won't ripple through the Apple hardware product line for a year or more. Just buy This Year's Model [youtube.com] and deal with it. Your PC-using friends will be polite enough not to say much about it to you.
Re: (Score:2)
Out of interest, what do you want Thunderbolt for? The only really useful thing I can see is external PCIe for a GPU or some other high bandwidth card... But such a machine would be stuck with a mobile CPU and mobile thermal limits, so even then only useful for some fairly niche applications.
For everything else USB 3 has plenty of bandwidth.
Re: (Score:2)
So the question for someone looking to buy a revised MacBook Pro this year would be, buy it now for a battle-tested Thunderbolt 3 connection, or wait for the chip integration for performance gains even though it will be a fist gen thing next year...
This is always the question, right? Whether to wait for new tech, or buy now, has been a consideration since at least 1986, when I first started buying them, (I decided to wait because, c’mon, expandable to 4 MB of RAM? Hell, yeah!). But, over time, I decided that, whenever I need a new machine to just go out and get it. Otherwise, you’ll be waiting forever, because there’s always something better coming.
eGPU! (Score:2)
Finally!, now hopefully a simple box consisting of a PCIE -> TB3 interface and a cheapo PSU will cost less than $299.
Re: (Score:2)
and be capped at pci-e X4 3.0. And you better hope that it's not tied to the DMI bus.
AMD has scared intel. (Score:1)
So now instead of a premium product thunderbolt is free and integrated.... as long as you are using an intel cpu. Of course AMD won't be able to add it to their CPUs. So the pending Ryzen mobiles will either have to do without or pay for the space, power, and expense of a discrete chip.
Re:AMD has scared intel. (Score:5, Interesting)
With the protocol now being royalty-free, what's stopping AMD from adding it to their CPUs?
Re: (Score:3)
Being "royalty free" can mean different things, so I checked the source [intel.com], which is quite explicit about this point:
In addition to Intel's Thunderbolt silicon, next year Intel plans to make the Thunderbolt protocol specification available to the industry under a nonexclusive, royalty-free license. Releasing the Thunderbolt protocol specification in this manner is expected to greatly increase Thunderbolt adoption by encouraging third-party chip makers to build Thunderbolt-compatible chips.
So yeah, seems there shouldn't be any legal reasons preventing AMD to implement it.
Re: (Score:2)
Re: (Score:3)
Re: (Score:1)
They have invented the damned thing, what "being ahead on integration" are you taking about, son?
Interesting is to know the practical downsides of connecting TB via PCIe lanes, as oppossed to "In CPU".
Heck, the best part about TB for me is the symmetric port.
Re: (Score:2)
Hopefully common sense. It's a gaping security hole.
Re: (Score:2)
Re: (Score:2)
AMD is a big company, and even if there were up-front costs, that wouldn't be an obstacle. Intel explicitly said this move is intended to get TB3 in third-party chips:
In addition to Intel’s Thunderbolt silicon, next year Intel plans to make the Thunderbolt protocol specification available to the industry under a nonexclusive, royalty-free license. Releasing the Thunderbolt protocol specification in this manner is expected to greatly increase Thunderbolt adoption by encouraging third-party chip makers to build Thunderbolt-compatible chips. We expect industry chip development to accelerate a wide range of new devices and user experiences.
https://newsroom.intel.com/editorials/envision-world-thunderbolt-3-everywhere/
"OS-level integration" (Score:1)
Re: "OS-level integration" (Score:1)
Re: (Score:1)
Intel does care about Linux. Unlike AMD or Nvidia, they directly contribute to MESA and make a free software graphics driver that actually works without major issues. And about Thunderbolt, they actually submitted a large patchset to lkml a few days ago.
Re: (Score:2)
Only the non pro version. They still publish a proprietary driver. And their vulkan driver is being promised to be open sourced but no sign of it, while volunteers are working on radv.
Re: (Score:2)
Re: (Score:2)
I miss the word "lamer".
It must be due for a comeback.
Re: (Score:1)
The Linux VM for Windows is strange. It seems like it's just a VM, nothing else. You could never actually use it to say, use the bash shell with the windows file system in a unix-style manner.
Re: (Score:2)
The website it is sourced from simply hugely biased the story to M$, who are providing nothing, basically just another company accessing the now royalty free hardware design. M$ will of course shit on about not providing thunderbolt to older versions of windows, as far as they are concerned either pay for the probe or fuck off. Intel will of course make the driver directly accessible but in the end by far the majority of thunderbolt connections will be android, then apple and then windows anal probe 10 in t
Re: (Score:2)
Windows 10 Progressing As Planned... (Score:3, Insightful)
Just another step towards locking out any OS but Windows 10, including locking out older versions of Windows. Either that, or badly cripple the few OS's that do still run on it, likely requiring approval by MS in order to get their boot loader signed for UEFI Secure Boot.
And right while people are finding out that even Enterprise edition Windows 10 [theinquirer.net] refuses to stop talking to external servers (including ad servers), even though companies pay through the nose for this specific version of Windows so that they can prevent just that sort of thing.
Make no mistake about it. Microsoft wants to own your computer, and intends on leaving you no choice in the matter, except perhaps not to own a computer at all. We're going to be seeing a lot worse coming down the pike very shortly at this rate.
Thunderbolt requires "active" cables - Fail (Score:5, Insightful)
A connector standard that requires expensive active electronics as an integral part of any cable is sure to fail with regards to mass market penetration.
Re: (Score:3)
I don't see how you can get the data rates and power delivery capabilities of TB3 without an active cable. Unless it's a very very short cable. Seems like a price worth paying for one-cable-to-rule-them-all type situations.
My biggest beef is that they didn't go with a mag-connector for USB-C. Seems like an oversight to me.
Re: (Score:3)
The problem with Thunderbolt is that they wanted to make the ports cheap and burden the price X times on the buyer of the cables, later. Won't fly.
Re: (Score:1)
As you can see from the almost zero prices of (even longer) TOSlink cables, such cables are very cheap -
TOSlink is a crap plastic based fiber, that exists for cheaply/lazily removing ground loops, not for high data rates. There is a reason 1+ Gbps fibers are glass, and not displaced by dirt cheap plastic fibers.
there almost no limits to the data rates possible on optical fibers.
When there are high loses and dispersion, you will quickly hit a data rate limit...
Re: (Score:2)
The issue with optical cables is that they don't like being bent too much. People will expect to be able to wrap and bundle them like they can with USB cables.
Also, the cost of optical cables that can carry tens of gigabits per second is much higher than TOSlink which is under megabits, i.e. several orders of magnitude lower.
Re: (Score:2)
My biggest beef is that they didn't go with a mag-connector for USB-C. Seems like an oversight to me.
Probably because having the cable fall out when you look at it the wrong way is several orders of magnitude more annoying when it stops actual functionality, as opposed to simply reverting to battery power.
Re: (Score:2, Informative)
No it doesn't. Active cables are required for if you want to run at 40Gb/s on a cable longer than 1.5 feet. If can accept lower speeds, however, or short distances passive cables are standard compliant.
Re: (Score:1)
I bet this protocol is so fucking horrid it makes USB (another Intel 'standard') look fun by comparison.
Re:Thunderbolt requires "active" cables - Fail (Score:5, Interesting)
Re: (Score:2)
They should just use USB C cables. They have the bandwidth and the USB C standard supports using the ports to carry other signals in a safe, compatible way. It could also allow peripherals to be multi-mode, USB and Thunderboilt.
Re: (Score:3)
"They should just use USB C cables"
All Thunderbolt 3 cables are guaranteed to work as USB-C cables.
Standards wars (Score:2)
I've seen plenty of wars over which standard comes to dominate. Beta-VHS, MS-keyboard vs USB (initially), SCSI vs. RS-232 (as a philosophical choice of parallel vs serial). But I don't understand this one.
Why competition between Thunderbolt 3 and USB-C? Aren't they nominally identical in all respects? Or are they exactly identical (?), in which case I am exposing my ignorance.
Re:Standards wars (Score:5, Informative)
USB-C is the connector type, not the protocol. Thunderbolt 3 is the protocol. All current Thunderbolt 3 implementations use USB-C as the connector type. Not all USB-C connectors attach to devices that support the Thunderbolt 3 protocol. It's essentially 1 cable/connector that supports many different protocols (USB 3.1, Ethernet, Thunderbolt 3, DisplayPort, etc).
Re: (Score:2)
Wow. Thanks to both of you for clearing this up!
Still DMA out the Side? (Score:5, Interesting)
Is this still architected so that anybody who can stick a cable in the side of your computer can suck down its memory contents?
Re: (Score:3)
There is an MMU which is supposed to manage the DMA and could in theory prevent these kinds of attack, but at the moment it isn't used like that and even if it was the implementation would be vulnerable to attack. This is Intel, the guys being the Management Engine fiasco that they were warned about for years, so security isn't very high up their list.
Working Closely with Microsoft? (Score:1)
Screw Microsoft in the ass. Why are they not working closely with Redhat and the Linux Kernel Developers?
Re: (Score:2)
I would rather Microsoft's only contact with Red Hat be through the Linux Kernel Developers who happen to work for Red Hat.
how many buses and will severs be stuck with video (Score:2)
how many buses and will severs be stuck with lowend video on die to drive this?