the names are terrible but I think the biggest issue are the "standards" that are not really standard where you can have 3 different types of USB C all with different "standards" so you end up getting dumb names for the same type of connection that has different standards. I.E. Standards are not really standards if you are catering to different types.
so its now PCI in smaller form. how long till we see 99999999999999999+ pin CPU sockets as a replacement for USB? edit: i edited the cpu pins. happy now?
This should have been THE means to have a beefy GPU that you can hide behind the monitor in a separate case so your pc can be small. 1 of these OCuLink cables would be enough for a 2080 at least.
@@paulmichaelfreedman8334Their would be a latency concern though and their's already sfx cases that can fit large graphics cards in them anyways using unique case layouts.
I saw a kid NOT MUCH older than that in a grocery store one day. He was sitting in a shopping cart with a phone in each hand playing two different games. I was so flabbergasted that I stood there with my mouth a gap. The Mom turned around and saw me and started laughing. I asked, his he really playing those games? She responded with, Yes, he is, and he's doing better than I could do with just one of them. I was like... no kidding... I can't even figure out what he's doing. So... tech savvy, maybe not... but def a more skilled end user than I will ever be!
Yea but it's a proprietary version of it so you can only buy their 2x MSRP GPUs for it. I have the x13(the first device with that connector) and it sucks balls having such an awesome connector with no real way to utilize it other than spending more than the cost of the device
@@m.devellis If they come out with an EGPU Enclosure with the XG Mobile interface, I would actually sell my desktop without the GPU and just use the X13 for everything
OCuLink would be great in those laptops where the manufacturer cheaps out on USB4 w/ PCIe tunneling or Thunderbolt. All that's needed, is rerouting 4 PCIe lanes and soldering a connector and maybe a redriver, but no controller chip, which makes it cheap. It's a sheer and utter shame that it hasn't seen a lot of adoption, since it's currently the best solution for connecting an eGPU.
There is another big drawback to OCuLink you should have mentioned: OCuLink is more fragile, it's not rated for as many plug/unplug as a lot of other external connectors like USB, HDMI or Ethernet
Ratings are overrated... micro-USB is rated for 10.000 insertions, while in reality if you're lucky it survives 1000. Meanwhile mini-USB is only rated for 5000 insertions and i never have seen one fail, despite that i use that at work connecting it multiple of times every day (for years) and moving the device around a lot with the cable attached. (PICkit 3 Programmer) (We also use the PICkit 4 with micro-USB, but that thing had contact problems after just 2 weeks of use and we now glue on that cable so that it doesn't move.)
TOSLink already existed and would do the same thing better. What you may not realize if you are younger is that the PC industry has always been and always will be a grift to steal your money.
Good idea, it'd be easier than swapping the gpu module, and there's already external mobile gpus docks and desktop gpu enclosures on the market for it. Mobile handhelds have already been using it and those manufacturers also made their own egpu docks, just look at onexplayer and gpd.
Agreed! I've been using the Asus ROG Flow series gaming 2-in-1 tablets which use a proprietary implementation of OCuLink for their XG Mobile eGPUs and they are amazing. But Framework's modularity is a game changer too. I'd LOVE to see all this in one product line, especially if its eGPUs weren't at all proprietary and vendor-locked!
@@AquaeAtrae yeah those look great, my biggest issue with those are that it's a closed garden, so i can't just chuck my existing gpu in a box and hook up, or use a different computer (unless it's a asus with XG mobile stuff) on it.
Another point for thunderbolt is that the USB-C port they use is physically smaller than an Oculink port, taking up less space if space is a concern. Like on the side of a laptop where an external graphics card would be most likely be used. The kind of laptop that would most likely put one on is also one they would cram a decent graphics card inside anyway.
Another point for thunderbolt is that cheap ass manufacturers can scam the masses with a low speed USB-C port. USB-C is a total scam. They did that crap on purpose and advertised 'USB-C' as 'High Speed' when it is only a standard for a physical connector. It is like saying your wall outlet can deliver 5 million volts just because it is an electrical connector. Try it. I dare you. Why does Thunderbolt use a USB-C connector? So they can scam people into thinking they have Thunderbolt when they don't. The old 1394 would have worked for the same speeds and been distinctively a high speed only connector. That's what Thunderbolt would have used if the industry wasn't a complete bait and switch scam job.
OCuLink comes in x4 and x8 variants. Max it can do is 64Gbps (16GB/s) @ x8. In x4 operation it's 32. Like you just worked out... They did mention it's two variants but yeah. They definetly didn't make that very clear. The way it's written makes it seem the 4 lane is what does 64Gbps. (they even put that equation up and its wrong lol). So don't blame you for being confused. That also means the 4 lane version (it's most common variant in the consumer space) is only on par with Thunderbolt at the moment. You should point this out to them.
@@nated4wgy No, that's very much what was said. And shown! At 1:10 ! Literally giving 4 x 8 maths on the picture! The 8-lane connection is mentioned entirely separately. In fact, at 0:58 they even show a timeline with OCuLink 2 at 2017, showing 4 lanes, and 64Gbps. Something's definitely off here. Edit: okay, I went and looked. _Apparently_ (though I honestly can't find much in terms of easy to parse information) the "64" in this case should be gigaTRANSFERS per second, GT/s. That is to say, 32Gbps, both ways at once. Like an Ethernet cable. That would also make your math make sense, since 16GB/s would be multiplied by 8 for bits, resulting in 128Gbps. But since it's actually half the throughput, running duplex - it's multiplied by 4 instead, giving 64GT/s. But... it's 64GT/s in its normal x4 configuration, not x8 like you said so... Either way the video's very wonk with the numbers. Also why is there like two websites that even have the numbers?!
1:55 Oculink is just a pci-e x4 slot in a different form factor, just like M.2 or U.2. Thunderbolt however, requires a thunderbolt controller to enable compatibility with daisy chaining, plug and play, charging, USB 2, USB 3 and DisplayPort. Oculink is just a direct connection from a plug to a cpu. It's not so much "using cheaper components" and more "get rid of >50$ controller". Oculink is faster than Thunderbolt, because Oculink can transport pcie 4.0 x4 (64gbit), while thunderbolt devices can only use up to pcie 3.0 x4 (32gbit).
Yes, but the controller actually costs $10, and maybe another $10 for all the signals on the PCB. Also Thunderbolt 5 is already out and even Intel's pathetic Thunderbolt controller that can only power one of the ports at full speed can do 64 Gb/s. And when the ASMedia controller comes out it will likely have PCIe 5.0 x4 or even x8 to allow 64 or even 80/120 Gb/s per port.
Recognized the connector in the thumbnail from my Valve Index and Samsung Odyssey+. So cool to see this format getting some recognition! Though I'm sure in a few years USB4 will make it completely obsolete.
Ive been wanting LTT to do a deep dive on occulink. We gotta get this port more news! Its on a few handhelds and mini pcs, hoping it picks up some more steam.
Nice to see some OcuLink coverage from LMG, it would be nice to see some more eGPU coverage since OL is making big strides in that regard with the m.2 to OL adapters + OL eGPU boards/enclosures that outperform TB3/4 and USB4.
Why this never became THE means of having a small pc case with a beefy GPU like a 2080Ti, in its separate case behind the monitor (or out of sight), is beyond me. Now if you want a beefy GPU your case MUST be huge unless its a custom build and very expensive.
the connector has terrible longevity, think some of the handhelds/mini laptops makers have had to put in a lot of work to make it more durable. a connector like that couldnt really be on a consumer product if it could degrade so quickly
@@TheAmazingCowpig Or AMD/NVidia develop GPU boards specially made for water cooling so they can be half length with two boards stacked. Such a cooling solution is possible in a relatively small form factor. But an external GPU really is best if they're always big.
I can't wait for the first TB5 controllers to come out, just so that they can't reach their advertised bandwidth because the controller chip wasn't connected with enough PCIe lanes or with a lower PCIe version.
Your precious Thunderbolt is really just a scam to make the average consumer think their USB-C port is fast. On every PC I have with a USB-C port, the USB-C is slower than some of the USB-A ports. That is reality for most consumers yet they are led to believe that port can do 50Gbps just because it is USB-C. Frankly, Thunderbolt is an overpriced license and really only worth the money for enterprise users. I wish they'd write a new 1394 standard because that would be royalty free. a 1394 cable can definitely hit Thunderbolt speeds with a modern controller. At the end of the day, Thunderbolt is no more of a transfer speed guarantee than a random USB-C port. But did Thunderbolt go with something distinctive? no. They chose USB-C because they got bought off so manufacturers could imply their USB-C ports are faster than they really are. It is sort of like 80 Plus ratings on PSUs where the 80 Plus really says nothing about quality or real world performance. I'm not sure why you'd need it on your gaming PC unless you wanna hook up an external GPU to a desktop, in which case, you might want to seriously re-evaluate your life choices. Your 10k sub YT channel doesn't justify Thunderbolt for transferring your video files. If you think otherwise, find a good therapist! The port doesn't magically make your flash drive faster. They will still be slow AF. Even if you have an NVME reader that nominally meets TB5 speed, it will never run anywhere near that fast. The potential transfer speeds far outstrip the capability of average NAND flash to expel data. Thunderbolt is mostly a fad that companies make a ton of money from, selling un-needed capability that really doesn't meet the claims to people who don't know any better. If you want fast transfer, open up PC 1, take the NVMe drive out and put it in PC 2. A $20 NVME to PCIe adapter will be faster than Thunderbolt in its wettest dream. Anything else is just jerking off. Seriously, if you want the fastest, Thunderbolt is a waste of time. It only exists because people are lazy and clueless. You don't even need it to hook a dGPU up to a laptop and direct connection to PCIe is always gonna be faster. Unless you have all thumbs, the time taken to swap a NVMe drive still doesn't make TB faster. And if you are all thumbs, this really isn't the hobby for you. You can also set up a server and that could probably transfer files between devices faster than Thunderbolt, while spending a fraction of the money. The thing is, you'll run into caching issues long before transfer speeds come into play and you might be able to max out that Thunderbolt interface...for about 5 seconds, and then the transfer rate will drop to prehistoric levels. By the time it is all said and done, you might as well have sent it over a 10Gbps interface, most of which can maintain that speed consistently these days. Thunderbolt is just gonna be hurky-jerky in its data delivery. You literally only need google 'external GPU over Thunderbolt' to see it isn't that impressive, a literal direct visual performance test. PCIe it is not. By the time it is actually affordable, it will have been greatly exceeded by a new standard you haven't even heard of yet. But if the industry really cared about your transfer speeds, USB would have been optical from 2.0 onwards. Or ditch USB altogether. A TOSLINK cable would be faster for digital data transfer. It was faster 20 years ago than USB and would be faster today if they had kept developing it for binary data transfer.
@@Lurch-Bot Oh look who can't detect sarcasm and thus wasted a bunch of time writing paragraph after paragraph of salty rants. Get some help dude, this is unhealthy.
IMOH, best use case of OCuLink is link between light-weight business laptop computer or mini PC and external powered GPU (like RTX4080) enclosure. Gaming laptop is too expensive, too heavy, too bulky and short on battery time. Thunderbolt 3 or USB4 port are too expensive for a light-weight budget business laptop. OCuLink can be add to such laptop with minimal additional cost.
You do not want a port that enables anybody with the right tool to pwn you, on a laptop, especially not a business laptop. You need a port with somekind of protocol, and setup and drivers, and that gets you to USB or Thunderbolt.
@@Carewolf You don't understand what you're talking about. Thunderbolt is literally PCIe with more steps and it doesn't need "setup and drivers". Thunderspy and Option ROM attacks are proof that Thunderbolt isn't safe. Thunderbolt 4 allows any device to connect by default. Secure Boot and IOMMU are bulletproof protections from these kinds of attacks, and both also work on OCuLink and are enabled by default. Also it's much easier to pwn someone with USB, all it takes is a cheap microcontroller that acts like a keyboard. Thunderspy needs $300 worth of bulky equipment and in some cases disassembling the laptop.
@@null-nl5su The attacks on Thunderbolt at all attacks that circumvent the built-in protections. You are advocating for a connection with NO built-in protection.
@@Carewolf You didn't understand what I wrote AT ALL. Option ROM attacks don't circumvent anything and don't work when Secure Boot is enabled. Thunderspy does circumvent Thunderbolt security levels but it requires dissasembly and an SPI programmer with default settings and Thunderbolt 4 also has "no built-in protection", it allows any device to connect by default. Thunderspy doesn't work when IOMMU is enabled and it is by default.
PCIE actually has a hot plugging feature. It's just not a thing in consumer devices. So Oculink could definitely do hot plugging if a manufacturer is willing to pony up for the feature.
I've read some tech reviews for current gen GPUs where significant reduction in card throughout and performance has been observed when running over OCuLink. Seems the tech needs a bit more debugging to get some GPUs to performance parity on some boards.
Question - HDMI 2.1 vs DP 1.4 - Let's say I get the Asus PG32UCDM (yum), then I get a 5090 when they become available. What method of connection would be better? Wouldn't the higher bandwidth of HDMI 2.1 be preferable? or is there some downside I'm missing?
Nice video. I knew what it was. I always though that OcuLink could of been "Fixed". The Fiber optic connecetion with the fiber optic cables was also a under-used option. I loved being able to have a portable CD-2-MD recording setup with the Fiber Optic functionality between the CD and MD. I even did a few CD to MD transfers for friends and family.
Very, very different sort of fiber. The old audio TosLink used a plastic fiber with basic LED and phototransistors transducers, one way. Very very cheap, very low bandwidth - the main reason to use optical was as a way to escape the curse of ground loops and differentials, the hated electrical problem that is usually the source of that humming noise in the speakers you can't get rid of. Optical components for high-speed data communication are a lot more expensive.
@@vylbird8014 An example of the cost difference is that a 5m standard DP cable is like $20 while the same length cable but with optical fibre instead of copper is $130. Source: I found out that 5m DP cables cannot carry 3440x1440@100Hz without issues while my optical one can do so with ease.
@@emu071981 Yeah. It's the same story with Thunderbolt: The early versions were intended to run over short-range optical cables. Intel initially called the standard Light Peak. Once the engineers got to work on it though, it was eventually found that designing copper connections capable of handling the bandwidth was actually a lot easier than trying to make cheap enough optical interfaces - and there was no demand for five-meter-long cables in the intended use cases. anyway. So the optical element was removed and it became a copper-only technology.
You didn't mention the most important element. The latency is much lower. its influence the frame rate to a big extent And it is very possible that the moment a cable become a smart cable, it gain latency. So it possible that even thunderbolt 5 can't really perform better than oculink
around 1:50 ... sure a passive cable is cheaper, than a TBT host and device chipset, which gears the PCIe Gen3x4 and two DP ports into two 20Gb/s lanes (40Gb/s on TBT5?)
A part of me wishes this was canon, because I've really wanted something to explore the pre-city age, but it's probably for the best to keep this as an alternate universe.
I actually knew about this cable, when I owned a steam deck the only way to use an external graphics card was via one of these cables, ofc I never went through with it though, it was way better for me to just sell the steam deck and get a laptop with far better specs while maintaining portability.
Thunderbolt has broader adoption because it's the ONE standard out there that actually means a damn thing anymore, now that the USB group has completely given up even pretending to care about any sort of standards compliance. If you see thunderbolt you know what you're getting, it's that simple. If you see USB it's a crapshoot.
Occulink doesn’t support power delivery, does it ? And no hotplugging either which limits its usefulness for certain devices. But still it is a good alternative to have. Where’s the pcie 5 version ?
@@zgriggs it doesn’t hot plug I tested it. That’s where thunder bolt has it beat. It also doesn’t tunnel usb or DisplayPort. But unlike thunderbolt, there’s no need for a thunderbolt or usb4 host controller. So occulink has merit especially for eGPUs. But for other types of media like storage not so much because what if you need to plug and plug the storage while the computer is still running. No hot plug support is no bueno.
This connector should be on the next generation steam deck, to allow true high performance when docked, would allow a steam deck 2 to fully replace desktop PCs for at keast some people
I think the most important thing for me, as a consumer, is ease of use. I can never go back to a connection that I have to orient correctly to plug in. The USB-C connection is such a great evolution and platform to grow from. However, ease of use isn’t as important if it is a set it and forget it connection such as a monitor. I will say that plugging in HDMI and DisplayPort is much more pleasant than DVI or VGA. So if we can make a mini OCuLink that doesn’t need proper orientation then I am all in replacing all the proprietary bottlenecks of Thunderbolt.
You do realize that, for the average consumer, whatever USB-C port they have is slower than most of their USB-A ports... The average consumer buys thinking USB-C=Thunderbolt. The whole thing is a scam. A true next-gen high speed data transfer standard would use its own connector. And it would be optical. They are literally giving us today what we should have had 15 years ago.
In Enterprise, OCuLink is nowhere near the popularity of Slim-SAS (SFF-8654) for PCIe Gen4. Also, MCIO (SFF-TA-1016) is the dominant PCIe Gen5 connector. I don't see much future for OCuLink...
having a high performance GPU in a self contained unit that has the cooling and power required for the GPU and just needs one of these OCuLink cables to connect to the PC would be a huge advantage... Imagine being able to connect that thing not only to a PC tower but also to a laptop and finally have reliable high performance GPU performance and no temperature issues or melting connectors or other expensive issues along those lines
3:26 not even a problem, I really don't know why everyone is allergic to cables. Also I don't think video would be a issue considering the data can be sent in a different format and decoded on the monitor side.
The new Ayaneo Flip models come with an Oculink connector on the top. Seems like more devices could benefit from a similar approach (and would be a lot cheaper than the ROG propeitary connector in the Ally).
There is now a Thunderbolt 5 dock from Huber, aka HyperDrive Next Thunderbolt 5, specs are nice, 5GE Ethernet. @Techlinked, may be have Linus or the team take a crack at that one vs the TB5 standard. Nick Savage video on TB4 Apple cables vs the cheap cables, non trivial differences
I've found displaylink and hdmi over usbc either has resolution problems or latency/frame drops. I've experienced this displaylink problem on both windows and mac
GPD, a company that has been making PC handhelds since before the Switch was even a thing, uses Oculing ports on their modern handhelds. They also sell an EGPU with an oculink port.
@@rightwingsafetysquad9872 Thunderbolt 4 may not be a bottleneck, but you need an intel CPU to use it. AMD can use USB4, but USB 4 is equivalent to Thunderbolt 3 in terms of bandwidth. You can buy various external GPU shrouds with Oculink too. The point is less about the GPD's EGPU and more about the fact that their handhelds have the port out of the box, which is rare in consumer devices.
@XzaroX Thunderbolt 4 is just Thunderbolt 3, but some previous optional features are now mandatory. USB4 and Thunderbolt 4 have the same bandwidth. And the only OCulink GPU expansion I've seen looks super sketchy.
Daddy Intel have disappointed me again by not having Thunderbolt 5 integrated into their latest Meteor Lake 'Core Ultra' chips. If the MSI Claw came out with Thunderbolt 5, it would be a definitive feature since a single cable connection from eGPUs to Handheld Gaming Console PCs make the most compelling argument. OCulink has gained renewed interest because of its adoption on handhelds from likes of GPD. Propriety solutions exist like on the ROG Ally or the Surface Connect. I'm very excited for Intel to enable Thunderbolt Share, which is a peer to peer ethernet networking. I can imagine two handheld gaming console PCs connected by a single USB-C cable for local multiplayer gaming.
Yep, the only pre-assembled cable I could find on digikey is £61 - I'll just keep using flat ribbons for short range and DP cables for longer range PCIe connections.
The issue is that, unlike pure image signals and HDMI/Display Port, external PCIe is itself a bit of a niche on the consumer market. Like, I don't even really need Thunderbolt.
Well not sure, that's like saying PCI-E is a bit niche. What's niche about having an extensibility port? Just because it's not used by an average user whose work doesn't involve using special hardware with their PC, for example, it doesn't become niche . Instead I would say that people are just too used to their laptops not being extensible that the idea of extensibility is a bit alien.
@@Alias_Anybody not necessarily due to lack of demand, so far eGPU options been subpar (high penalty, compatibility issues because of another protocol on top) and costed a lot with Thunderbolt. It's not entirely fair to use such argument when the market supply wasn't there. People just got used to work that way and may be reluctant to changing their way.
It's nice that in the video you make the argument for calling the elements their latin names. Yes, we should do that. English names are actual garbage. Latin, however, makes sense.
Minis forum i love you. I remember the day you came out, most people say you such sketchy ass brand.. but now.. you are making innovations left and right (most killing the overpriced product)
I don't think I have the emotional bandwidth for yet another connector standard. I'm so close to purging Micro USB from my life. I can't handle another one.
Only place I've seen OcuLink is on VR headsets, though in those cases it also supplies power to the headset so fair to say it's not real OcuLink, just the connector.
With the image at :59 seconds if TB 3 and TB4 have the same pcie and 32 gbps speed and both cant transfer video....whats the differences? Seems the same with an updated name.
The PCI group should lend their naming guy to the USB forum guys.
what, you don't like the name USB 3.1 2X2 gen 2 XL ?
@@oussama7132nah it needs more keywords like Pro or Ultra
Did you forget that they change the naming scheme for USB 3?
the names are terrible but I think the biggest issue are the "standards" that are not really standard where you can have 3 different types of USB C all with different "standards" so you end up getting dumb names for the same type of connection that has different standards. I.E. Standards are not really standards if you are catering to different types.
so its now PCI in smaller form.
how long till we see 99999999999999999+ pin CPU sockets as a replacement for USB?
edit: i edited the cpu pins. happy now?
Fun Fact some VR Headsets like the Valve Index use the OCuLink connector for their main cable harness.
This should have been THE means to have a beefy GPU that you can hide behind the monitor in a separate case so your pc can be small. 1 of these OCuLink cables would be enough for a 2080 at least.
@@paulmichaelfreedman8334Their would be a latency concern though and their's already sfx cases that can fit large graphics cards in them anyways using unique case layouts.
And those are only rated for like 50 insertions. Hardly the "best connector" or even "one you've never heard of"
@@boboe2244I've plugged and unplugged my Index hundreds of time, connector is fine
Holy sh*t, only 50 insertions? That has to be a typo, right? Lololol that’s wildly bad.
me: *sees thumbnail*
also me: "DisplayPort!"
Disappointed
Me too lmao
i honestly thought it was some sort of composite video connecter (like used on the ps1 or the n64)
Ya saw it was released 12hr ago. So brain automatically said nope, not displayport, propably oculink or mcio
Yeah, my brain short circuited a bit when he said "oculink can't do video" when it looks EXACTLY like a Displayport cable.
"For a less tech savvy user"
Yes, that baby is definitely a less tech savvy user.
He looked so offended.
I saw a kid NOT MUCH older than that in a grocery store one day. He was sitting in a shopping cart with a phone in each hand playing two different games. I was so flabbergasted that I stood there with my mouth a gap. The Mom turned around and saw me and started laughing. I asked, his he really playing those games? She responded with, Yes, he is, and he's doing better than I could do with just one of them. I was like... no kidding... I can't even figure out what he's doing. So... tech savvy, maybe not... but def a more skilled end user than I will ever be!
That explains why the ASUS XGMobile has 2 connectors and requires you to properly power off the system before disconnecting it, its using an OCuLink
Yea but it's a proprietary version of it so you can only buy their 2x MSRP GPUs for it. I have the x13(the first device with that connector) and it sucks balls having such an awesome connector with no real way to utilize it other than spending more than the cost of the device
@@m.devellis Blame Nvidia for that one. They're the ones controlling what devices their products are allowed in.
@@m.devellis If they come out with an EGPU Enclosure with the XG Mobile interface, I would actually sell my desktop without the GPU and just use the X13 for everything
@@agentcrm I keep hearing that but then it why does the Radeon XG mobile cost so much too?
@@agentcrmso nvidia forced the use of a proprietary connector? What do they get out of it? There are no royalties for OcuLink
OCuLink would be great in those laptops where the manufacturer cheaps out on USB4 w/ PCIe tunneling or Thunderbolt. All that's needed, is rerouting 4 PCIe lanes and soldering a connector and maybe a redriver, but no controller chip, which makes it cheap. It's a sheer and utter shame that it hasn't seen a lot of adoption, since it's currently the best solution for connecting an eGPU.
I'm buying the first laptop that does this with a decent enough CPU.
@@m.devellisThere is a mod for the lenovo ideapad that has the oculink connector in the m.2 slot and a spring loaded cover like the rj45 on laptops.
looks AMD has already dropped the ball since razer just announced the 2024 edition of their laptop will have thunderbolt 5.
You probably do not want an external pwn me port on a laptop. Though just that was actually not uncommon back in the firewire days.
@@m.devellisLenovo is supposed to releasing something “soon” with OCuLINK. GPD already has a handheld size laptop with a 7840M in it… (Win Max 2 2023)
Btw, VR headsets, like the Valve Index, HP Reverb or Vive Pro, use this connector.
My Reverb G2 uses USB-C and DisplayPort on the PC side. I'd never thought to consider it was an OCuLink on the HMD side! Today I realised...
The josh hutcherson with the whistle reference made me giggle
Bruh, the timing. I read your comment, did the whistle myself and not even a second later that part of the video played.
There is another big drawback to OCuLink you should have mentioned: OCuLink is more fragile, it's not rated for as many plug/unplug as a lot of other external connectors like USB, HDMI or Ethernet
Ratings are overrated... micro-USB is rated for 10.000 insertions, while in reality if you're lucky it survives 1000. Meanwhile mini-USB is only rated for 5000 insertions and i never have seen one fail, despite that i use that at work connecting it multiple of times every day (for years) and moving the device around a lot with the cable attached. (PICkit 3 Programmer) (We also use the PICkit 4 with micro-USB, but that thing had contact problems after just 2 weeks of use and we now glue on that cable so that it doesn't move.)
@@bjiirnmini USB was the worst connector I have ever used, easily worse than microUSB.
I dunno micro usb is pretty bad
TOSLink already existed and would do the same thing better. What you may not realize if you are younger is that the PC industry has always been and always will be a grift to steal your money.
@@FrodoMaster1996 Nah Mini USB is little larger but its lot more durable than Micro USB
Rally up the boys and take over this connection
"Jack of all jacks." ... well played.
Came to the comments to appreciate it. 👍🏻
3:55 he got us😂
If the Framework 16 comes with a 8x oculink2 expansion bay module at some point, it would easily replace my desktop.
I think the ports are only 4 lanes.
Good idea, it'd be easier than swapping the gpu module, and there's already external mobile gpus docks and desktop gpu enclosures on the market for it. Mobile handhelds have already been using it and those manufacturers also made their own egpu docks, just look at onexplayer and gpd.
Agreed! I've been using the Asus ROG Flow series gaming 2-in-1 tablets which use a proprietary implementation of OCuLink for their XG Mobile eGPUs and they are amazing. But Framework's modularity is a game changer too. I'd LOVE to see all this in one product line, especially if its eGPUs weren't at all proprietary and vendor-locked!
@@jolness1 there is a 8 lane standard as well.
@@AquaeAtrae yeah those look great, my biggest issue with those are that it's a closed garden, so i can't just chuck my existing gpu in a box and hook up, or use a different computer (unless it's a asus with XG mobile stuff) on it.
Another point for thunderbolt is that the USB-C port they use is physically smaller than an Oculink port, taking up less space if space is a concern. Like on the side of a laptop where an external graphics card would be most likely be used. The kind of laptop that would most likely put one on is also one they would cram a decent graphics card inside anyway.
Another point for thunderbolt is that cheap ass manufacturers can scam the masses with a low speed USB-C port. USB-C is a total scam. They did that crap on purpose and advertised 'USB-C' as 'High Speed' when it is only a standard for a physical connector. It is like saying your wall outlet can deliver 5 million volts just because it is an electrical connector. Try it. I dare you.
Why does Thunderbolt use a USB-C connector? So they can scam people into thinking they have Thunderbolt when they don't. The old 1394 would have worked for the same speeds and been distinctively a high speed only connector. That's what Thunderbolt would have used if the industry wasn't a complete bait and switch scam job.
PCIe does have hotplugging functionality. If a particular mainboard doesn't support it, it's because the manufacturer was a cheapskate.
I love how 4 PCIe lanes * 8 Gbps = 64Gbps 😭🥶
OCuLink comes in x4 and x8 variants. Max it can do is 64Gbps (16GB/s) @ x8. In x4 operation it's 32. Like you just worked out...
They did mention it's two variants but yeah. They definetly didn't make that very clear. The way it's written makes it seem the 4 lane is what does 64Gbps. (they even put that equation up and its wrong lol). So don't blame you for being confused. That also means the 4 lane version (it's most common variant in the consumer space) is only on par with Thunderbolt at the moment.
You should point this out to them.
@@nated4wgy No, that's very much what was said. And shown! At 1:10 ! Literally giving 4 x 8 maths on the picture! The 8-lane connection is mentioned entirely separately.
In fact, at 0:58 they even show a timeline with OCuLink 2 at 2017, showing 4 lanes, and 64Gbps. Something's definitely off here.
Edit: okay, I went and looked. _Apparently_ (though I honestly can't find much in terms of easy to parse information) the "64" in this case should be gigaTRANSFERS per second, GT/s. That is to say, 32Gbps, both ways at once. Like an Ethernet cable. That would also make your math make sense, since 16GB/s would be multiplied by 8 for bits, resulting in 128Gbps. But since it's actually half the throughput, running duplex - it's multiplied by 4 instead, giving 64GT/s. But... it's 64GT/s in its normal x4 configuration, not x8 like you said so...
Either way the video's very wonk with the numbers. Also why is there like two websites that even have the numbers?!
They made a mistake, Gen4 lane is 16Gb/s, 4 of them make 64Gb/s
isnt it 32 in one direction and 32 in the other so 32 * 2 = 64?
@@nated4wgy That was my point 😔
1:55 Oculink is just a pci-e x4 slot in a different form factor, just like M.2 or U.2.
Thunderbolt however, requires a thunderbolt controller to enable compatibility with daisy chaining, plug and play, charging, USB 2, USB 3 and DisplayPort. Oculink is just a direct connection from a plug to a cpu. It's not so much "using cheaper components" and more "get rid of >50$ controller".
Oculink is faster than Thunderbolt, because Oculink can transport pcie 4.0 x4 (64gbit), while thunderbolt devices can only use up to pcie 3.0 x4 (32gbit).
Yes, but the controller actually costs $10, and maybe another $10 for all the signals on the PCB. Also Thunderbolt 5 is already out and even Intel's pathetic Thunderbolt controller that can only power one of the ports at full speed can do 64 Gb/s. And when the ASMedia controller comes out it will likely have PCIe 5.0 x4 or even x8 to allow 64 or even 80/120 Gb/s per port.
Recognized the connector in the thumbnail from my Valve Index and Samsung Odyssey+. So cool to see this format getting some recognition! Though I'm sure in a few years USB4 will make it completely obsolete.
Ive been wanting LTT to do a deep dive on occulink. We gotta get this port more news! Its on a few handhelds and mini pcs, hoping it picks up some more steam.
DisplayPort cables are also suitable for PCIe (same impedance requirements). I even seen SATA cables used for PCIe lanes.
Good job editor for all the footage 😄
I have one of these connecting a 6800xt to my strix g16 with an i9, draws desktop level 450w when gaming and is extremely plug and play
I could watch this funny man all day you have no idea
LOL that Daddy Intel stock footage
Nice to see some OcuLink coverage from LMG, it would be nice to see some more eGPU coverage since OL is making big strides in that regard with the m.2 to OL adapters + OL eGPU boards/enclosures that outperform TB3/4 and USB4.
Why this never became THE means of having a small pc case with a beefy GPU like a 2080Ti, in its separate case behind the monitor (or out of sight), is beyond me. Now if you want a beefy GPU your case MUST be huge unless its a custom build and very expensive.
the connector has terrible longevity, think some of the handhelds/mini laptops makers have had to put in a lot of work to make it more durable. a connector like that couldnt really be on a consumer product if it could degrade so quickly
@@konzo5942 So they redesign the connector. Cheap? No. Doable and viable? Absolutely.
A SFF build in something like a Fractal Ridge case isn't big, but it does come with some ITX tax.
@@TheAmazingCowpig Or AMD/NVidia develop GPU boards specially made for water cooling so they can be half length with two boards stacked. Such a cooling solution is possible in a relatively small form factor. But an external GPU really is best if they're always big.
mid atx case ftw
I thought it was the PlayStation 2 video output connector.
It looked like DisplayPort to me lol
@@greatwavefan397 It does look like that, but it looked too big, even in the picture. Plus, the title kinda rules it out.
Well that's just insulting.
I can't wait for the first TB5 controllers to come out, just so that they can't reach their advertised bandwidth because the controller chip wasn't connected with enough PCIe lanes or with a lower PCIe version.
Your precious Thunderbolt is really just a scam to make the average consumer think their USB-C port is fast. On every PC I have with a USB-C port, the USB-C is slower than some of the USB-A ports. That is reality for most consumers yet they are led to believe that port can do 50Gbps just because it is USB-C. Frankly, Thunderbolt is an overpriced license and really only worth the money for enterprise users. I wish they'd write a new 1394 standard because that would be royalty free. a 1394 cable can definitely hit Thunderbolt speeds with a modern controller. At the end of the day, Thunderbolt is no more of a transfer speed guarantee than a random USB-C port. But did Thunderbolt go with something distinctive? no. They chose USB-C because they got bought off so manufacturers could imply their USB-C ports are faster than they really are. It is sort of like 80 Plus ratings on PSUs where the 80 Plus really says nothing about quality or real world performance.
I'm not sure why you'd need it on your gaming PC unless you wanna hook up an external GPU to a desktop, in which case, you might want to seriously re-evaluate your life choices. Your 10k sub YT channel doesn't justify Thunderbolt for transferring your video files. If you think otherwise, find a good therapist!
The port doesn't magically make your flash drive faster. They will still be slow AF. Even if you have an NVME reader that nominally meets TB5 speed, it will never run anywhere near that fast. The potential transfer speeds far outstrip the capability of average NAND flash to expel data. Thunderbolt is mostly a fad that companies make a ton of money from, selling un-needed capability that really doesn't meet the claims to people who don't know any better.
If you want fast transfer, open up PC 1, take the NVMe drive out and put it in PC 2. A $20 NVME to PCIe adapter will be faster than Thunderbolt in its wettest dream. Anything else is just jerking off. Seriously, if you want the fastest, Thunderbolt is a waste of time. It only exists because people are lazy and clueless. You don't even need it to hook a dGPU up to a laptop and direct connection to PCIe is always gonna be faster. Unless you have all thumbs, the time taken to swap a NVMe drive still doesn't make TB faster. And if you are all thumbs, this really isn't the hobby for you.
You can also set up a server and that could probably transfer files between devices faster than Thunderbolt, while spending a fraction of the money.
The thing is, you'll run into caching issues long before transfer speeds come into play and you might be able to max out that Thunderbolt interface...for about 5 seconds, and then the transfer rate will drop to prehistoric levels. By the time it is all said and done, you might as well have sent it over a 10Gbps interface, most of which can maintain that speed consistently these days. Thunderbolt is just gonna be hurky-jerky in its data delivery. You literally only need google 'external GPU over Thunderbolt' to see it isn't that impressive, a literal direct visual performance test. PCIe it is not. By the time it is actually affordable, it will have been greatly exceeded by a new standard you haven't even heard of yet.
But if the industry really cared about your transfer speeds, USB would have been optical from 2.0 onwards. Or ditch USB altogether. A TOSLINK cable would be faster for digital data transfer. It was faster 20 years ago than USB and would be faster today if they had kept developing it for binary data transfer.
@@Lurch-Bot Oh look who can't detect sarcasm and thus wasted a bunch of time writing paragraph after paragraph of salty rants.
Get some help dude, this is unhealthy.
I even have Oculink devices here (to be fair I work at a data center more or less)
Valve index uses oculink and uses it for power as well.
Am I the only one who thought that the thumbnail was of DisplayPort?
1:15 You said it wrong Riley, it's not Bandwidth, it is Bandwdith
4:56 All hail MrYeast!
1:16 Bandwdith 🙃
And 4x8=64 😂.
IMOH, best use case of OCuLink is link between light-weight business laptop computer or mini PC and external powered GPU (like RTX4080) enclosure. Gaming laptop is too expensive, too heavy, too bulky and short on battery time. Thunderbolt 3 or USB4 port are too expensive for a light-weight budget business laptop. OCuLink can be add to such laptop with minimal additional cost.
The main benefit of OcuLink is performance and price of the eGPU enclosure. Thunderbolt/USB4 costs like $20 on the host side, so not too expensive.
You do not want a port that enables anybody with the right tool to pwn you, on a laptop, especially not a business laptop. You need a port with somekind of protocol, and setup and drivers, and that gets you to USB or Thunderbolt.
@@Carewolf You don't understand what you're talking about. Thunderbolt is literally PCIe with more steps and it doesn't need "setup and drivers". Thunderspy and Option ROM attacks are proof that Thunderbolt isn't safe. Thunderbolt 4 allows any device to connect by default. Secure Boot and IOMMU are bulletproof protections from these kinds of attacks, and both also work on OCuLink and are enabled by default. Also it's much easier to pwn someone with USB, all it takes is a cheap microcontroller that acts like a keyboard. Thunderspy needs $300 worth of bulky equipment and in some cases disassembling the laptop.
@@null-nl5su The attacks on Thunderbolt at all attacks that circumvent the built-in protections. You are advocating for a connection with NO built-in protection.
@@Carewolf You didn't understand what I wrote AT ALL. Option ROM attacks don't circumvent anything and don't work when Secure Boot is enabled. Thunderspy does circumvent Thunderbolt security levels but it requires dissasembly and an SPI programmer with default settings and Thunderbolt 4 also has "no built-in protection", it allows any device to connect by default. Thunderspy doesn't work when IOMMU is enabled and it is by default.
Fun fact: OCuLink is what was used on the CV1 Oculus Rift headset cable
2:51 skip ad
Linus should check out some oculink egpus it has more bandwidth than thunderbolt 4
PCIE actually has a hot plugging feature.
It's just not a thing in consumer devices.
So Oculink could definitely do hot plugging if a manufacturer is willing to pony up for the feature.
I've read some tech reviews for current gen GPUs where significant reduction in card throughout and performance has been observed when running over OCuLink. Seems the tech needs a bit more debugging to get some GPUs to performance parity on some boards.
Never heard about this one!
Thanks!
OCuLink sounds like a cable for the oculus
Question - HDMI 2.1 vs DP 1.4 - Let's say I get the Asus PG32UCDM (yum), then I get a 5090 when they become available. What method of connection would be better? Wouldn't the higher bandwidth of HDMI 2.1 be preferable? or is there some downside I'm missing?
Nice video. I knew what it was. I always though that OcuLink could of been "Fixed". The Fiber optic connecetion with the fiber optic cables was also a under-used option. I loved being able to have a portable CD-2-MD recording setup with the Fiber Optic functionality between the CD and MD. I even did a few CD to MD transfers for friends and family.
Very, very different sort of fiber. The old audio TosLink used a plastic fiber with basic LED and phototransistors transducers, one way. Very very cheap, very low bandwidth - the main reason to use optical was as a way to escape the curse of ground loops and differentials, the hated electrical problem that is usually the source of that humming noise in the speakers you can't get rid of. Optical components for high-speed data communication are a lot more expensive.
@@vylbird8014 An example of the cost difference is that a 5m standard DP cable is like $20 while the same length cable but with optical fibre instead of copper is $130. Source: I found out that 5m DP cables cannot carry 3440x1440@100Hz without issues while my optical one can do so with ease.
@@emu071981 Yeah. It's the same story with Thunderbolt: The early versions were intended to run over short-range optical cables. Intel initially called the standard Light Peak. Once the engineers got to work on it though, it was eventually found that designing copper connections capable of handling the bandwidth was actually a lot easier than trying to make cheap enough optical interfaces - and there was no demand for five-meter-long cables in the intended use cases. anyway. So the optical element was removed and it became a copper-only technology.
You didn't mention the most important element. The latency is much lower.
its influence the frame rate to a big extent
And it is very possible that the moment a cable become a smart cable, it gain latency. So it possible that even thunderbolt 5 can't really perform better than oculink
around 1:50 ... sure a passive cable is cheaper, than a TBT host and device chipset, which gears the PCIe Gen3x4 and two DP ports into two 20Gb/s lanes (40Gb/s on TBT5?)
I actually have heard of oculink.
A part of me wishes this was canon, because I've really wanted something to explore the pre-city age, but it's probably for the best to keep this as an alternate universe.
I actually knew about this cable, when I owned a steam deck the only way to use an external graphics card was via one of these cables, ofc I never went through with it though, it was way better for me to just sell the steam deck and get a laptop with far better specs while maintaining portability.
Was there an adapter for this kind of thing?
Steam Deck never had OCuLink by default, unless you're referring to the M.2 NVMe to OCuLink adapter.
@@Kerbiter yep, talking about the adapter.
@@_Rerr_ indeed there is.
Thunderbolt has broader adoption because it's the ONE standard out there that actually means a damn thing anymore, now that the USB group has completely given up even pretending to care about any sort of standards compliance. If you see thunderbolt you know what you're getting, it's that simple. If you see USB it's a crapshoot.
but I have heard of it before :( I used it to conect a raid controller with a dock for U.2 SSDs
Occulink doesn’t support power delivery, does it ? And no hotplugging either which limits its usefulness for certain devices. But still it is a good alternative to have. Where’s the pcie 5 version ?
No, no power. Not sure about hot plugging. PCIe technically supports it, and OCuLINK is basically just a bunch of PCIe lanes on a cable… so possibly.
@@zgriggs it doesn’t hot plug I tested it. That’s where thunder bolt has it beat. It also doesn’t tunnel usb or DisplayPort. But unlike thunderbolt, there’s no need for a thunderbolt or usb4 host controller. So occulink has merit especially for eGPUs. But for other types of media like storage not so much because what if you need to plug and plug the storage while the computer is still running. No hot plug support is no bueno.
Faster not bottlenecked full PCIE speed hot swappable OcuLink would be a holy grail for the future eGPU PC setup!
This connector should be on the next generation steam deck, to allow true high performance when docked, would allow a steam deck 2 to fully replace desktop PCs for at keast some people
Best Techquickie in a long time
I think the most important thing for me, as a consumer, is ease of use. I can never go back to a connection that I have to orient correctly to plug in. The USB-C connection is such a great evolution and platform to grow from. However, ease of use isn’t as important if it is a set it and forget it connection such as a monitor. I will say that plugging in HDMI and DisplayPort is much more pleasant than DVI or VGA. So if we can make a mini OCuLink that doesn’t need proper orientation then I am all in replacing all the proprietary bottlenecks of Thunderbolt.
You do realize that, for the average consumer, whatever USB-C port they have is slower than most of their USB-A ports...
The average consumer buys thinking USB-C=Thunderbolt. The whole thing is a scam. A true next-gen high speed data transfer standard would use its own connector. And it would be optical. They are literally giving us today what we should have had 15 years ago.
I ordered the new thinkbook, and it comes with oculink. Look forward to its arrival
i am expecting some oculink accessories to appear for the framework 16's expansion bay.
In Enterprise, OCuLink is nowhere near the popularity of Slim-SAS (SFF-8654) for PCIe Gen4. Also, MCIO (SFF-TA-1016) is the dominant PCIe Gen5 connector. I don't see much future for OCuLink...
having a high performance GPU in a self contained unit that has the cooling and power required for the GPU and just needs one of these OCuLink cables to connect to the PC would be a huge advantage... Imagine being able to connect that thing not only to a PC tower but also to a laptop and finally have reliable high performance GPU performance and no temperature issues or melting connectors or other expensive issues along those lines
I would rewind back to 2016 and look up a company called Alienware 😆
@Techquickie - at 1:16, bandwidth is spelled incorrectly. :)
The problem with Thunderbolt is actually that the plugs can be easily unseated or unplugged.
You need to have that port and plug checked out. It's not easy to unplug USB-C cables from my laptop.
Looks like display port
Excited for Thunderbolt 5, the two 4k at 60hz is very limiting when you have three 4k monitors.
3:26 not even a problem, I really don't know why everyone is allergic to cables. Also I don't think video would be a issue considering the data can be sent in a different format and decoded on the monitor side.
I'm pretty sure the Asus XG Mobile connector is a way of combining OCulink with USB4.
The new Ayaneo Flip models come with an Oculink connector on the top. Seems like more devices could benefit from a similar approach (and would be a lot cheaper than the ROG propeitary connector in the Ally).
The miniPC community has heard of this
There is now a Thunderbolt 5 dock from Huber, aka HyperDrive Next Thunderbolt 5, specs are nice, 5GE Ethernet. @Techlinked, may be have Linus or the team take a crack at that one vs the TB5 standard. Nick Savage video on TB4 Apple cables vs the cheap cables, non trivial differences
yeey, more cables and adapters please, I still have room in my drawers (not really) 😅
I love how Techquickie teaches me something genuinely interesting and new everyday ❤
it’s you again. it’s goddamn you.
Which connection would provide the most stable and, fastest connection for an eGPU. OCuLink or M.2 NVME 4.0?
I've found displaylink and hdmi over usbc either has resolution problems or latency/frame drops. I've experienced this displaylink problem on both windows and mac
GPD, a company that has been making PC handhelds since before the Switch was even a thing, uses Oculing ports on their modern handhelds. They also sell an EGPU with an oculink port.
But the GPU that GPD is using is low enough performance that Thunderbolt 4 isn't a bottleneck.
@@rightwingsafetysquad9872
Thunderbolt 4 may not be a bottleneck, but you need an intel CPU to use it. AMD can use USB4, but USB 4 is equivalent to Thunderbolt 3 in terms of bandwidth.
You can buy various external GPU shrouds with Oculink too. The point is less about the GPD's EGPU and more about the fact that their handhelds have the port out of the box, which is rare in consumer devices.
@XzaroX Thunderbolt 4 is just Thunderbolt 3, but some previous optional features are now mandatory. USB4 and Thunderbolt 4 have the same bandwidth. And the only OCulink GPU expansion I've seen looks super sketchy.
LTT viewers are likely some of the only people in the world who actually know what GPD is lmao. Don’t have to explain them here xd
Daddy Intel have disappointed me again by not having Thunderbolt 5 integrated into their latest Meteor Lake 'Core Ultra' chips. If the MSI Claw came out with Thunderbolt 5, it would be a definitive feature since a single cable connection from eGPUs to Handheld Gaming Console PCs make the most compelling argument. OCulink has gained renewed interest because of its adoption on handhelds from likes of GPD. Propriety solutions exist like on the ROG Ally or the Surface Connect. I'm very excited for Intel to enable Thunderbolt Share, which is a peer to peer ethernet networking. I can imagine two handheld gaming console PCs connected by a single USB-C cable for local multiplayer gaming.
Is there any PCIe Riser using oCuLink interface? Rather using USB 3.0 that can only transfer X1 PCIe lanes.
Me: Is that big display port(from the tumbnail).
Pretty difficult to get excited about something that I cannot use.
Yep, the only pre-assembled cable I could find on digikey is £61 - I'll just keep using flat ribbons for short range and DP cables for longer range PCIe connections.
The issue is that, unlike pure image signals and HDMI/Display Port, external PCIe is itself a bit of a niche on the consumer market. Like, I don't even really need Thunderbolt.
Well not sure, that's like saying PCI-E is a bit niche. What's niche about having an extensibility port? Just because it's not used by an average user whose work doesn't involve using special hardware with their PC, for example, it doesn't become niche . Instead I would say that people are just too used to their laptops not being extensible that the idea of extensibility is a bit alien.
@@Kerbiter
Few people use external GPUs or similar.
@@Alias_Anybody not necessarily due to lack of demand, so far eGPU options been subpar (high penalty, compatibility issues because of another protocol on top) and costed a lot with Thunderbolt. It's not entirely fair to use such argument when the market supply wasn't there. People just got used to work that way and may be reluctant to changing their way.
Also meant professional equipment more, such as audio interfaces or whatever else that may use PCI-E.
0:02 Yes, he uses it to control his robot arms, legs, butt, etc
It's nice that in the video you make the argument for calling the elements their latin names. Yes, we should do that. English names are actual garbage. Latin, however, makes sense.
yeah thunderbolt 4 is already niche and has bandwidth for most application.
Minis forum i love you. I remember the day you came out, most people say you such sketchy ass brand.. but now.. you are making innovations left and right (most killing the overpriced product)
So it's better than a floptical drive? (also a real, obscure thing)
@1:15, might want to check your math...8Gb/s per lane times 4 lanes...does not yield 64 Gb/s...thinking you gave the 8 lane spec?
Only heard of this from serve the homes as a workstation or server link. To connect a graphics or ai engine external form factor.
Thinkbook 16+2024 will support the Oculink， it's a good sign
I don't think I have the emotional bandwidth for yet another connector standard. I'm so close to purging Micro USB from my life. I can't handle another one.
Whos doing math at 1:15?
Use a calculater bro
Is that the connecter used for the headset side of the Oculus Rift S cable?
Jokes on you. I've already heard of this connector
Only place I've seen OcuLink is on VR headsets, though in those cases it also supplies power to the headset so fair to say it's not real OcuLink, just the connector.
I had PCIe on a few laptops,
I think I could use it as a usbport?
With the image at :59 seconds if TB 3 and TB4 have the same pcie and 32 gbps speed and both cant transfer video....whats the differences? Seems the same with an updated name.
me: wtf is with these stock videos?
video: ...right after we thank [stock video vendor] for sponsoring this video!
Linus never fails to tickle my pickle
@techquickie: there is a typo in the word "bandwidth" @1:18
Is the optical component not still valid for longer cables?
tbh USBC/TB is finally at the 1 cable for all stage, from smartphone to laptop power AND picture delivery...
let's go with that
>DVI >SDI >HDMI >DP >DIVA >MHL
there's 6 digital video standards, let's make just all into one!
there is now 7 digital video standards