Intel Thunderbolt Support Under Linux
|
|
|
Comments (8)
What IS it?
They say it's a combination of displayport and pci express, which makes it sound like a video architecture, but then they say the initial drivers are peer to peer, making it sound like some sort of networking/data transfer architecture.
Anybody have a more clear understanding of exactly what it is?
Quoting:What IS it?
Couple good demos
http://www.youtube.com/watch?v=1ox_inwLSl0
http://www.youtube.com/watch?v=lCzHcxUXmqw
Other demos
I'm still under the impression that this is yet another "special Apple display cable" (which apparently also is intended to replace USB2.0/3.0, allowing Apple to remove still more "inelegant" connectors from its devices). Is anyone else actually implementing it right now?
So from reading the WP page, it's basically a combination of a video port and a data port..interesting idea, but...WHY?
Ok, so it is high bandwidth. The Video side only seems useful for on-board graphics. The couple times I've messed with Display port plugs (some of our Lenovo's at the hospital have them), I found them to be kind of a pain to work with.
So for me it comes down to:
1. Why do we need a combined video/data interface when the video side is likely only useful for integrated graphics systems (and why do we need a combined video/data interface to begin with, it just confuses things and means more hubs and cabling).
2. Why use a display system (displayport) that seems to have very little uptake in the market (Newer Lenovo machines seem to have them, and I THINK I may have seen a laptop of another brand once with one)?
Why?
From Charlie Demerjian over at SemiAccurate (http://semiaccurate.com/2011/06/16/intel-declares-open-war-on-mobo-makers/):
More on Thunderbolt from Charlie in these posts (always opinionated but fun to read if you're hardware-literate)...
http://semiaccurate.com/2011/02/24/intels-thunderbolt-in-a-nutshell/ --what it is
http://semiaccurate.com/2011/02/24/intel-takes-the-light-from-light-peak/ -- how it came about
http://semiaccurate.com/2011/05/09/iphone-5-has-thunderbolt-locks-users-in/ -- why Apple likes it
http://semiaccurate.com/2011/04/15/usb-design-allows-for-25-gbps/ -- USB alternative
http://semiaccurate.com/2011/06/23/pci-wants-to-squash-thunderbolt/ -- even PCI might become an alternative
Bottom line, to my reading: something to avoid.
From Charlie Demerjian over at SemiAccurate (http://semiaccurate.com/2011/06/16/intel-declares-open-war-on-mobo-makers/):
Quoting: ...they are desperately trying to kill USB3 because it competes with their own (not) Light (any more) Peak bus, aka Thunderbolt. Thunderbolt is obsolete, technically dubious, vastly more expensive than USB3, and offers no advantage to the user, but lots of disadvantages. The biggest problem is that Thunderbolt is proprietary, so Intel can use it to force peripheral vendors to pay them huge fees while locking out competition.
More on Thunderbolt from Charlie in these posts (always opinionated but fun to read if you're hardware-literate)...
http://semiaccurate.com/2011/02/24/intels-thunderbolt-in-a-nutshell/ --what it is
http://semiaccurate.com/2011/02/24/intel-takes-the-light-from-light-peak/ -- how it came about
http://semiaccurate.com/2011/05/09/iphone-5-has-thunderbolt-locks-users-in/ -- why Apple likes it
http://semiaccurate.com/2011/04/15/usb-design-allows-for-25-gbps/ -- USB alternative
http://semiaccurate.com/2011/06/23/pci-wants-to-squash-thunderbolt/ -- even PCI might become an alternative
Bottom line, to my reading: something to avoid.
I have a question:
I understand that USB3 will put a fair bit of load on the CPU -- does Thunderbolt manage to avoid this? (if it does, then their might be an excuse/real use for it).
I understand that USB3 will put a fair bit of load on the CPU -- does Thunderbolt manage to avoid this? (if it does, then their might be an excuse/real use for it).
@bernardswiss: The mechanism for that is DMA, Direct Memory Access, wherein the CPU is suspended for a few cycles while extra io hardware accesses a buffer in main memory, rather than an ISR, interrupt Service Routine, wherein the CPU has to execute the memory access and io access itself, costing a bunch of cycles (including saving and restoring what it was doing when the interrupt came through). From a quick look around, both USB and PCIe (an underpinning of Thunderbolt) are spec'd such that transfers can be sped up by using DMA if the specific chips support that, so, no, it's not a differentiator.
Post a comment
A free account is needed to post. Write your comment here first: your text is kept in this browser while you sign up or log in, and is waiting for you when you come back.
Your message:Markdown: **bold**, *italic*, [text](https://link), - lists, > quotes, $x^2$ for math. Formatting help
Sign up and post (it is free) or Log in and post
