ไม่สามารถเล่นวิดีโอนี้
ขออภัยในความไม่สะดวก
USB4 Isn't Cut And Dry - And That's A Bummer
ฝัง
- เผยแพร่เมื่อ 17 พ.ค. 2023
- Gordon tested the first true USB4 drive he's seen (ZikeDrive) and compared it to a couple USB 4.0 drives to see the difference in speed. Along the way he also found out that not every USB4 port is created equal - and realized it's all still a mess.
Buy PCWorld merch: crowdmade.com/...
Follow PCWorld for all things PC!
--------------------------------
SUBSCRIBE: www.youtube.co...
TWITCH: / pcworldus
TWITTER: / pcworld
WEBSITE: www.pcworld.com
#usb
In Gordon's voice: "It depends on what your definition of USB 4 is"
That sounds like Bill Clinton to me 😅
-Adam
@@pcworld It depends on what your definition of "is" is.
@@pcworldclose but no cigar as Monica said 😂
TLDR: Thunderbolt 3/4 only provides 32Gbps for data, the other 8Gbps of the often claimed 40Gbps on Thunderbolt is set aside for other protocols like Displayport over Thunderbolt. USB4 on the other hand can use the entire 40Gbps of its bandwidth for data and therefore can achieve higher transfer rates than TB3/4
This is incorrect. TB3 only allows 16gbps for data. TB4 is 32. USB4 *can* provide up to 40Gbps but the standard only **requires** 20gbps.
I think the magic word is "can" .
Reminds me of tons of internet articles that start with : XYZ might be the biggest etc etc
"can" and "might" : two words that have people make a lot of ass-umptions ...
@@itessentialai227 yes minimum spec and theoretical maximum are, sadly, not differentiated, usually through intentional deception.
would this also apply to external graphics cards?
@@6kbps yes, the limit is not "data" per se but PCIe. An external graphics card would use the PCIe bus to feed its data and thus would be subject to that limit across the connection.
I can't think of anything with a worse naming convention than USB. It's just been so consistently useless for multiple generations now.
AMD might be coming close ever since they came up with the new naming scheme for Ryzen 7000.
HDMI is trying
@@Mr.Morden How? Their naming is on point!
The deciding board wrote in new changes as recently as this year to have the USB type 1,2 ,3 ,4 and max theoretical speed written on the connector. eg USB 2.0 1 Gbps or USB 4 20 Gbps. companies are expected to use their suggestions , some will ignore them.
@@zacthegamer6145 have you looked at the image they dropped for their mobile ryzen lineup? bruh
Shame how USB was such a nice straightforward thing initially and slowly morphed into confusion. Either way; this will be amazing for small form factor systems needing extra storage!
follow the money. the less transparency, more specs, more interoperability failures all equate to more sales. Its as simple as that. It causes people to buy more cables, newer peripherals, and upgrade entire systems. It also gives PC manufacturers reasons to re-release entire laptops because some of us just gotta have that 3.1 _gen 2_, can't be left in the dust with gen 1.
@@mcpr5971 Some of the recent changes are due to extreme frequencies that the data rates are going to in the higher USB standards. USB Type A connectors just don't make it as a connector for 40Gbyt data streams. Physics forces you to change the connector format. USB-C does extend the frequency capabilities up to at least 40Gbs (only for distances less than 4 feet). Maybe a fiber optic version of USB is in order here. Fiber Optic USB cables could extent the frequency range out to possibly as high as 1Tbs, and good for a much longer distance. The future of USB will demand a different format for the connector as the frequencies go higher and higher.
We need USB 5.0 to have all the same cords, all be the same on every motherboard and actually be CONSISTENT
It's cool to see that decades later and multiple retroactive renames the only thing you can expect to work reliably is USB 3.0 and 2.0, and that Asmedia are still the only decent USB drive controllers
The best part is that by offical naming convention, "usb 3.0" doesn't exist anymore and is meant to be called "usb 3.1 gen 1" yet everyone calls it usb 3.0 because it's by far the better name
@@oggilein1 Technically speaking the official consumer name for USB 3.0 is and has always been "USB SuperSpeed 5Gbps"
While the faster transfer speeds 10gbit/s and 20 Gbit/s are supposed to be called "USB SuperSpeed+" with or without the gbit speed.
Meanwhile USB 2.0 is "USB Hi-speed" as opposed to USB 1.1 that is "USB Full speed" and 1.0 that is "USB Low Speed"
Which is very good and clear and that's why nobody ever used any of that because wtf is hi-speed and superspeed, who is bigger? Numbers, how they work?
The names they have been playing silly buggers with for 15 years are the "internal" in-industry names which they think nobody should use directly to customers.
USB spec board are morons
@@oggilein1 They actually changed it again, USB 3.0 is now called USB3.2Gen1
For anything after USB 3.0, ignore the dot and whatever number is after it. The important part is the "Gen."
For consumer purposes anything after the dot doesn't matter. Gen 1's speed is the same whether it's USB 3.0, USB 3.1, or USB 3.2.
So I just say "USB 3 Gen 2" for anything USB that's 10Gbps. If it's 20Gbps, I'll say "USB 3 Gen 2x2" (even thought that "2x2" still sucks).
@@killerb255 I've personally never needed to care about the difference between the USB 3 5/10/20 Gbit so I just call them all USB 3.0 for brevity.
If I had to, I would probably use the bandwith number because that actually means something a person can grasp instead of talking gens and 2x2 and whatnot
Oh the good old days of boot/maximum PC, when both Gordon and me had hair......😭😭
Well this is disappointing that the two USB4 ports on my X670E Taichi, a $500 board, are not able to reach the full 40Gb/s.
USB is starting to get as confusing as SCSI, maybe a new standard will come along to save us
At least they could agree on 2 physical connectors for USB by 2024 instead of the 20+ that SCSI had / has. For the electrical side i do agree i do with the USB-IF would make it easier for the consumer to tell which one is on their device. Just saying "everything has to be supported to be USB 5" would be good. The part of "everything" is the important part in tat but i fear that companies will get involved and they will get a curve out for theirself. To which i hope they will be forced to stay on the USB4 branding.
I cannot believe that, "We didn't say it was 'USB4', we said it was 'USB 4'," would fly in any court, if the USB consortium were to actually sue. The judge would say, "pull the other one, it's got bells on!"
This and given how close it is to the trademark, the usb if has the obligation to sue.......or that trademark is worthless
Excellent video. Thank you! And yeah.... data transfer speed marketing vs data transfer speed realized is definitely not the same😂
Just a side note. Doesn't mean those drives are bad. They are still the top 3 nvme enclosures and no doubt PCWorld carefully picked them for that reason. All of them are top-end and you can easily edit 4k video off of them. What is more worrying to me is how much worse everything else in the $80+ range is.
They are still pretty bad. They overheat in long sustained writing(2 tb+). They simply don't cool the usb/tb chip at all which make the enclosure disconnect and/or they don't have the thermal design to cool the ssd making it throttle.
My quest for fast backup pretty much pushed me to enterprise stuff... That's pretty annoying. I can't believe how peoples accept so low reliability from usb/tb enclosure and nvme drives.
Also, it's hard to believe how lightly those device are tested by professional reviewer. It's like no one do backup on anything else than spinning rust.
Handy to know about the space in the naming. Thank you.
5:06 Asus laptops and a beta BIOS that enables USB4 support… where have I heard that one before. Oh right, the Zenbook S13 with Ryzen 7 6800U, where Asus originally promised USB4 support, but then did a 180. I have a Zenbook S13 and I’m still mad about that.
2.8 GB/sec over a wire is still pretty darn good. You tested with Intel laptops and an AMD motherboard, what about an Intel motherboard with Maple Ridge + z690/z790? Also, what about on a Macintosh? Apple claims to have "USB4" ports, but how fast are they when connected to a drive enclosure with the new AsMedia usb4 chip?
On Mac is slow, but you can't compare benchmark. Just wait for real life tests.
@@nnnnnn3647 mac has the same 2800 MB/sec speed using the acasis enclosure (which is using titan ridge). The question is how will the asmedia usb4 controller perform on mac? will there be a similar performance uplift (vs the acasis) ?
Not only that, but you got ASUS who have not delivered on there promise of introducing USB4 via a Bios Update on the Asus Zenbook S 13 OLED equipped with a Ryzen 7 6800U which Gordon reviewed almost a year ago. Gordon was given a Bios Update directly from ASUS (not yet released to the general public) with USB4 enabled, but still nothing to this day. The Laptop has been fantastic, just this one minor omission that is frustrating...
Don't know if Gordon or the Team have any further information in relation to this, if so would be of help. Thanks
Similar story with the 2022 ROG line-up. Yes, G14, Flow X13, etc. got a beta BIOS, but it's still only beta and there is a new BIOS 3.18 (for the G14 at least) which doesn't even have a beta equivalent (3.17 is the latest beta with USB4). If I recall correctly Zephyrus G15 2022 didn't got a beta BIOS with USB4 yet at all.
It seems all 2023 model got the USBC4. I have 2022 flow x16, only has beta bios. The only issue I read is beta can't sleep your USBC4 device properly, so if you don't use sleep much the beta bios is as good as the official one
I didn't even know USB 4 was available. Everything I own still just brands with USB 3, and I haven't heard of any USB 4 devices either. Maybe I'd have run into some if I was looking for portable SSDs, but I'm not.
wait for usb5 - it is on the way along with wifi7 - sure they are bound to screw it up a bit but respins and bios updates should eventually unlock the full potential of nvme
All this time I've never had a USB drive with a type C connection let alone anything faster than 100 MB / sec lol
Thrat hair joke cracked me up!
I'll never forgive the USB IF for the naming conventions.
Ah, motherboard manufacturers trying to cut corners… like I've *_never_* seen that before! 😅 Sadly, it ain't quite as easy as it once was to fix though. At least, not unless you can get full PCIE expansion cards running the full spec bandwidth.
Compatibility doesn't mean the same max speed just the exchange language will function for the transfer. Some ports like for thunderbolt don't change connection contacts for the USB arrangement standard so there goes another slow down. What is good is being able to have an enclosure that can still handle old and new connections.
So by the time we have Intel 15th gen we will finally have proper USB4 drives and motherboards etc supporting it correctly(on a mainstream level).. 😅
Thanks Gordon for a very thorough and well-done video!
every usb since 3.0 is still a mess...
USB-IF cursed and blessed the community when they divorced the USB-C connector standard and the data transmission standard.
Blessed because USB-C is easy to plug in.
Cursed because USB-C can be usb2.0, usb 3.x, usb 3.genWTF, usb4, DP, TB3, TB4, display signal(s), tunneling/no-tunnels, etc
USB 3 is also really easy to pull out, which may be good for some situations but it's definitely not good whenever you're charging your phone while using it.
I get it's not a very common use case, but man I wish it clicked at least a little bit rather than slithering out of the socket in silence..
USB-C is shit, it breaks easily (compared to normal USB-B aka the classic usb ports or even USB-A aka the printer USB connector), the connector can get clogged or rusted easily (the contacts are tiny) and on top of the purely mechanical issues they added the randomness of whatever the fk interface is supported by the host and the guest device is now up in the air, also now there are dozens of different cables that may or may not support the features or the power delivery you need. And also it's fun when there are competing charging standards too for power delivery.
The only thing USB-C should have been was a replacement for micro USB connectors, that's it. In phones and other small devices it's OK and admittedly less bad than the micro usb. Everything else should have used normal USB ports instead
With USB-C we are literally back to the days of "only using the cable provided by the device manufacturer" which is a load of bull
I think USB 4.0 / PCIe 4.0 is a bit missleading , 40 Gbit is basically 4 Lanes PCIe 3.0 NVME Performance . I dont see a big Difference between 3200 vs 3800 = yes 20 % , but only with big Files = continued writing and even thats only half true because if the DRAM and SLC Cache limit is reached , writing speeds will go down drastically , if you use QLC it can be even lower than HDD speed ~ 50 -100 MB sec . But yes , if an new Controller from AS Media is faster and cheaper , why not use it ? Yes USB naming scheme is braindead with its renaming of existent Standards , they simply should have been added the Speed = USB3 5 GBit , USB3 10 GBit , USB3 20 GBIT , USB4 20 GBit / USB4 40 GBit and you would have known what you can expect .
PCIe is not a problem. It THAT PARTICULAR device and how they've always worked that you're pointing out (i.e. solid state drives in general)
PCIe is used for far more than storage you know.
It's also because while TB3/4 claims 40Gbps, it only allows data transfers to have 32Gbps (which is the speed of PCIe 3.0x4) while the other 8Gbps are set aside for other protocols like displayport over Thunderbolt. From this video it looks like USB4 can actually utilize that entire 40Gbps for data and therefore the USB4 drive in this video achieves faster speeds
@@post-leftluddite TB3 allows 24.8 Gbps for data transfers.
TB3 allows 16gpbs for data.
USB4 mandates 20gbps but can support 40Gbps if fully implemented.
There are many SSDs that can write to TLC at 3000Mb/s - or at least flush the SLC cache fast enough that you never go into the really-slow folding mode (emptying SLC cache to TLC while also writing in. new data to SLC).
QLC is actually not that slow. It's the folding mode that is the killer. That's why cache is very misleading, but also meaningful if you don't constantly write huge amounts of data.
>> they simply should have been added the Speed = USB3 5 GBit , USB3 10 GBit , USB3 20 GBIT , USB4 20 GBit / USB4 40 GBit and you would have known what you can expect .
They shouldn't have bothered with the numbers, except in tech literature. Just say USB 5gbit, USB 20Gbit, USB 40Gibt - and that's actually what they're starting to do now with cable compatibility.
@@nowandrew4442 With the Acasis TBU401 TB3/USB 3.2 Gen 2x1 enclosure I can reach 3100 MBytes/s, so it’s close to 24 Gbits/s.
Thanks for opening our eyes
Really well explained and presented. Thank you
USB 3: Hold my beer!
USB 3.1, the other 3.1, 3.2, etc.
I have a USB m.2 enclosure velcroed to my laptop lid. I installed Fedora on it and it runs great for lightweight use web, email, etc.
what bandwidth on the enclosure and what speed on the drive can you get out of it?
i've been looking at a 10Gbps enclosure and in theory it should be able to run at 1.25GBps which would easily be saturated by a cheap m.2 but from videos i've seen getting a 1GBps transfer rate out of a 10Gbps interface is already pretty good
@@deneguil-1618 It's running on a Ryzen 3050U 2 core CPU in a super cheap $250 laptop. The disk a 128GB SATA m.2 disk that came with the laptop. The USB-C port is 3.1 GEN 1 (625MB/s).
In KDiskMark I get 124MB/s read for RND4K Q32T1 and 44MB/s write.
15MB/s for RND4K Q1T1 read and 15MB/s write.
400MB/s for sequential read and 178MB/s sequential write.
So questionable controllers abound, on top of naming snafu.
I feel the uptake of USB4 to be oddly slow too. Wondering if some of these controllers were made before final ratification of the spec (outside of the blatant thunderbolt 3 controllers being paraded as 4 taking advantage of the idiotic naming conventions)
Wait, Gordon with hair? Surely that's an AI generated image.
Busted
Ohhh I hope you never fall into the hole that is made of USB-cables...
Gordon is love.
What about the HyperDrive us4 ssd enclosure or the satechi usb4 ssd enclosures
In the beginning there was RS232. There were many flavors of RS232, and if you did anything wrong, you could burn up the RS232 connection, and get nothing. Chaos ensued, and various more complex solutions were concocted to overcome this chaos. Then came Other flavors of RS232 ( RS485, and others) where the risk of burning something up was eliminated, and things were better, but not perfect. Then came the glory of USB. Everything was designed to work together, and the universe was good, and chaos was greatly reduced. USB 1, USB 2, and USB 3 came with all their communication goodness without too much chaos. Recently something has happened to the glorious simplicity of USB, and strange names have been coined that don't really reflect the actual capability of newer USB standards. So now chaos has returned to the quiet USB universe. Maybe it's time for a new standard beyond USB. But we need to try to recapture the elegant simplicity of USB 3.0 (past 3.0, and things get chaotic).
I went through much of the same research on my systems. The key to finding where the bottleneck was starting with the M.2/nVME interface and recognizing it is based on PCIE, so you have to be aware of whether the support is Gen3 or Gen4 speeds (Gen3 being 3500mBs and Gen4 being 7000mBs). It doesn't matter how fast the wire to the enclosure is rated at but rather what gen PCIE interface is tied to the M.2. I use Samsung drives almost exclusively and Samsung Magician will query the drive to tell you the speed of the interface that it is plugged into. I put a Gen4 drive into an enclosure (supposedly rated as USB4 ) and the drive reported it was attached to a Gen3 interface and would only support half of its rated throughput-no where close to the 40gbs rate that it was sold as. When I went back to the manufacturer they admitted they had NO product that would support a Gen4 drive at it's rated speed. HTH
I bought the zike drive and put a crucial t500 in it
But it does NOT work on the thunderbolt 4 port on my acer spin 5 dp513-55N (its not even detected). But it works just fine over usb.
My advice from what I could find check your drivers first! Because the newest tb4 drivers aren't the entire story. Apparently your NVM (just nvm) drivers needs to have st least 46 that's what zike told me.
What NVM-Firmware did/do you have? My Zike didn't work with 28.0 on my old AMD/AM4-Plattform (ASUS Crosshair VIII Extreme), but now i updated my BIOS/UEFI+TB4-Controller and NVM-Firmware 38.0 works!
Thanks for Captioning the video!!
Its stange , after all these it still feels the stopgap interface of estata still feels like the most stress-free and most likely to work vs usb 3,XX update, it feels strange how messy usb3 and its off shots come off, heck it was kind crazy around 2010 it was to find native usb3 support on stuff.
I think USB4 is pretty cut and dry, what isn't is the market around it skirting the specific spec with slightly altered namings. The way to prevent this is for USB-IF to trademark the alternate names for the sole purpose of disallowing them in marketing. Furthermore, USB-If needs to find some legal language that says companies that use these alternate namings, or anything lesser than the actual USB4 spec, face some kind of repercussion for walking the line of false advertising. I've said it before, and I'll continue saying it, but USB5 needs to come fast, and it needs to fix every open-ended issue USB4 currently has, where USB5 needs to be _the_ spec and _the only_ spec with no room for interpretations or bodged implementations for compatibility's sake; USB5 needs to be absolutely cut and dry across the board and across the industry. Which also means that USB5 needs to clean up any other PD issues that still exist, and needs to clean up various tunneling and alt mode nuances and just make an outright unified standard for it. Another thing that needs cleaned up is the braindead naming schemes and implementations, again USB5 needs to be a singular spec, no more generations, no more single or dual lane, no more symmetric or asymmetric with different nominal speeds, there needs to be one speed full stop, with major increases being a new spec; for example, the gen 2 single lane to gen 3 single lane jump being a doubling, the gen 3 single lane to gen 4 jump being a quadrupling on sym or a hextupling on asym, I know the spec is moving fast but USB4 Gen 4 should've been USB5, and USB5 should've been asym for purposes of simplicity.
I also wish a future PD spec could allow for daisy-chained one-cable setups, where there's power overhead for downstream devices, for example if I have a laptop, a display, and downstream peripherals, I want want them daisy chained to a singular wall wart, but as it currently stands I am limited to what individual ports can put out and overheads are so limited to where that setup requires at least two if not three wall warts. At the same time, I also want this implementation to carry data as intended, say I add some external drives, an eGPU, etc., into the mix, I still want a daisy-chained solution where there's virtually only one wire from one endpoint to the other. Essentially, I want a future USB spec to not only be the universal serial bus, but also the universal power bus, and not that but also the universal power and data backbone for setups, where a virtual one-wire implementation is possible, where a secondary branch wire is only needed in the event of a temporary device being added to the ecosystem. I say 'virtual one-wire' due to the fact that there are multiple wires, but a device is the common end point for two cables and thus ties those two loose ends into one extension; at least this is until we get a spec that allows for a singular true wire with multiple plugs, like a hub but rather than a block with multiple wires it's literally just a wire that meets multiple devices at the device with no additional cables needed, and honestly this might be the solution for the idea of a single-cable backbone as this could at least provide a discrete power rail, though flexibility takes a hit. Honestly I just want the spaghetti of wires to go away, especially as wireless isn't catching up in some areas (such as displays) and wireless still needs a temporary cable for power needs; for example I currently have six individual wires in my setup, two power and four peripherals, with one additional wireless peripheral with its own charging cable, yet this could easily be consolidated into a singular wire if USB and PD allowed for it, albeit that additional wireless peripheral would be a case of where a branch is needed for the sake of simplicity, and this issue gets exacerbated when I add a pre-USB4 Type C hub into the mix as it requires its own power source if your host doesn't have ~15w output or you're adapting to a Type A port. I just want things to be clean without the need for ample cable management, because even with the best cable management you'll run into issues if anything ever needs to be changed.
You can't claim something as your trademark and never use it yourself AND forbid others from using it because then the thing was technically never your trademark to begin with (because you never engaged in TRADE using that MARK). Trademarks are "use it or lose it", as well "as defend it or lose it" (I'm aware China allowed trademarks to be registered with no intent to use other than to sell it, which lead to rampant "trademark squatting" in that country, but they have introduced laws to close that loophole). USB-IF can indeed go after the usage of "USB 4.0" for trademark infringement and I am surprised they don't because "USB 4.0" is causing market confusion with "USB4".
Not to mention that there is no way to "test" the USB4 port itself. I have fried two enclosures just plugging them in. Thank goodness that the NVMe was not damaged.
Hrm so what does that "40" mean next to the USB 4 ports on the Asus board? If it were Gbps that would be 5GB/s right?
Yes 🤔
Thunderbolt 3 is already very fast for most needs. With speed matching the internal drive of my macbook. I already do some of my work and run apps stored in a WD black inside a tb3 enclosure.
Had no idea there was this issue. LTT seems to have covered it a while back, but I wasn't in the market for one back then and ignored it. Bought a USB3.1 NVME enclosure instead - didn't really need speed and it only cost $25.
To fix it, we can just have USB4.0, USB4.1gen1, USB4.1gen1TBlite, USB4.3betamax, and USB4.2x2at3.2x1speed
USB4 and Thunderbolt 4 are already outdated. We need more bandwidth especially for eGPU docks with NVMe storage, memory card readers, dual 10G AVB NICs and other extra ports. We really need PCIe 5.0 x4.
There's occulink, currently it is at the Gen 4 state but it is as close as what you can get for full pcie bandwidth.
@@lucretius8050 Oculink is cheaper than Thunderbolt but not as widely adopted as USB. Also one area of weakness is power delivery. Trying to find information online about Oculink is a pain.
And hasnt this stuff already been tested when thunderbolt originally came out claiming 40gpbs. People were testing thunderbolt enclosures like the G-Drive to see if they were getting what they paid for. Im even more confused than when i hit play.
Sheesh - there are so few fully functional AM5 motherboards (ones with a PS/2 port, to use God's Keyboard, the IBM Model M) -- now we have to look extra-hard for motherboards with true USB4 also??
I've been using an NVMe enclosure really fast on USB-4
Hello, I can't find any video on youtube that compares Satechi with ziketech. Have you already tried the Satechi model? Please help us, we need an expert.
More Framework videos would be nice.
Thank God USB 3 got sorted out...
That's a bummer, I've been watching for a good deal on the Orico
well that sucks, i too have the asus x670e gene motherboard like gordon and was going to populate the two USB4 ports with with these USB4 drives after i had populated the three onboard nvme's
It does have to be noted that the 40Gbps spec in the USB-4 standard is "upto" 40Gbps, not a guarantee of 40Gbps.. There are SO many factors that can reduce that - like the cable you choose.
Meanwhile, I'm beside myself if I can get 200MBps... so...
Correct, the USB4 spec only mandates 20Gbps, not 40gbps.
How does it function on the newest gen of Macbook Pros with M1 Pro and M2 Pro? these support USB4 and thunderbolt 4
They 'support' USB4 only because the port is electrically compatible and backwards-compatible. USB4 is a later revision that includes all TB4 abilities. But a TB4 port is not fully capable of full USB4 operation. Just compatible with the devices, but will likely run at USB 3.2 speed - 10Gbs. Possibly 20gbps if you get lucky (which is the minimum spec of USB4).
This is very insightful. Thanks
FML, I just picked up a lenovo X1 nano with 11th gen Intel chip. It has TB 4 ports, but it looks like they cap out at 32gbps... Thank God the laptop was a good deal on a mega discount
USB 4.0 and below are compatible, can the USB 3.2 20g USB 3.2 GEN2x2 20Gbps port run up to 20g? Or is it the same as the Thunderbolt hard drive box that slows down to 10G at the 20G port?
Could you please check compatibility of USB4 with Thunderbolt 3 audio interfaces (Antelope Audio, UAD Apollo and so forth) ?
Weridly enough is that USB3.2 gen2x2 (type-C) 20Gbps at the front performs better than USB4(Type-C) 40Gbps at the back (Especially Write speed).
I am using ADATA NVME3.0 SX8200 1TB with SSD HYPERDRIVE NEXT USB4 box and X670E ROG GENE.
.USB3.2 gen2x2 (type-C) 20Gbps: R/W ~2000/2000
.USB4 40Gbps: R/W ~2000/1200
I still dont get it. It literally says on the IO shield for thunderbolt 3 that its 40gbps. so everyone literally lied to us?
What. A. Mess.
Also Gbps and GB/s (which is used in PCIe versions) is very confusing, at the beginning of the video I thought (huh, if it's a USB4 drive it should be 40GB/s but no, it's 40Gbps which is something completely different. Annoying
I have questions. I bought a usb c hub with sd card reader and I notice the sd card is not reading or writing the mac speed .
ASUS ThunderboltEX 4, GIGABYTE GC-MAPLE RIDGE, ASRock THUNDERBOLT 4 AIC 2.0, and i'm sure most other Thunderbolt 4 expansion cards, all use PCIe 3.0 x4 which should only be capable of 4 GB/s while advertising 40gbps speeds which confuses me. i'm hoping some proper USB4 expansion cards come out for PCIe 4.0 x4 slots
the folks behind USB need to get their shit together... this is horrible for consumers
What would you like me to do mam?
Will you benefit from a USB4/Thunderbolt4 cable that can transfer 40 Gbps at 15 feet, instead of the 6 feet?
by this point, why do we not get Networking over USB?
40Gigs is so much faster than anything you can get in
LAN stuff and cheaper to boot . .
You technically can if you have the drivers to do it. You could build a router to do it with the internet. The issue is that USBC works best at short ranges. Ideally less than 2 meters (6ft ish) and gets significant degradation and then failure beyond that. The reason RJ45 is still used for networking is because the vast bulk of it is now bought by enterprise for server use where fiber isn't necessary. There's little market for a consumer solution on this when ISP speeds are up to a gbit in a best case scenario and most people's devices are wifi connected.
They need a thunderbolt 5 with PCIe 5.0x4 so that it has 128Gbps...that would open up a whole new world of possible I/O devices and accessories, like eGPUs without a loss in performance.
How about if using pcie 4x cards add on
Root cause of this issue is Thunderbolt
Whenever Apple steps in, things get ugly
That is the merge of USB and Thunderbolt
Dammit where’s mine? Spose you were early bird…
at this point USB is not a standard, more so a port than can operate with many different pin configurations. There should be wider contact USB C standards with notches to prevent non PD cables to go in and what not. but still offering the niche anyside is right to plug in.
inevitably, we need a bigger port to supply bigger power interfaces and enough differential pairs to have extremely high bandwidth connectivity. or heck: optical connections.
lets go down the usb port-hole and make more interfaces. we already this far
It is standard as at minimum every USB device will operate in 2.0 mode regardless what cable you use. USB-PD requires cables with chips with metadata without it you only get standard USB voltage, if something fried your device it means host or charger don't follow that standard
so a drive can identify as a USB4 drive but it needs to call it self USB 4.0 drive
4:05 lol
So, how much time will it save me transferring that doc file?
Sigh, just shut down USB and keep developing Thunderbolt
USB has been a mess since 2.0, change my mind.
I can't remember last time I used my external SSD stick, would be nice if all computers had 10Gbe lan ports by now, with optional 25Gbe or 100Gbe, instead we barely start seeing 2.5Gbe onboard, it's pathetic when you realize NVME PCIe 3.0 (~30Gbit) has been here for 8 years already and yet many enclosures barely go anywhere close to that, as seen in the video
the power consumption and heat dissipation required to transfer 25-100gbps over ethernet is not practical for consumer market.
The BW isn't the problem but how to maintqin signal integrity at low power is the problem.
The usb type c grows into 40Gbps or 20Gbps, but what about our ethernet? we just has a realtek 2.5G ethernet without issues, why, maybe we should say who stop/slows down the ethernet speed grows?
If you’re tunneling USB 3.2 Gen 2 (USB 10Gbps) data through a Thunderbolt 3 / USB4 connection surely you can claim USB4, right?
USB 'C' is also a bit of a mess in my book, with multiple flavours of it, some do video, some do power delivery.... so unless you know the finer details of every port on your computer, your results of trying to plug in that latest bit of kit maybe somewhat disappointing.
That because USB hosts don't support those optional features. Note that those are extra features, alternate modes are to pack different ports to USB port for thin laptops and mobile devices that did this kind of shananigans with micro USB anyway and power delivery is mostly for chargers, and it requires power control circuitry for each port that's why motherboards have problem implementing. USBPD is also standard on its own having own spec
USB-C is a connector. Nothing else. There's no mess there in the slightest.
@@nowandrew4442 What he refering to is some cables don't have all pins connected and cable power specs. There cables that only hald of high speed USB pins connected, there once that only USB 2.0 pins connected and there even cables that only have pins for power. It's mainly to deliver low cost cables, if device don't need full speed.
@@ShadowriverUB USB cables are also quite a mess, that's true.
USB naming in general is an absolute nightmare.
You don't really look through what you get for your money.
USB with its Gen 1 to Gen 3 and their underscore 1, 2 also gives 3?
The drop in naming will also be determined for USB 4 ford law.
USB4 is confusing Gordon...
omg!! he said the magic word!! ''hopefully one day'' !! may be the frown-hopper usb institute will make a patch! till then! haven't heard about anything cooking yet!
I didn't even hear about usb4 being available
Great information! 3GB/s vs 3.8 shouldn't matter to most people so it is good to see lickety split speeds are a norm
hearing 3.x gb per seconds is like... I take the worst of those ;)
Just transfering 2TB over effectly USB 2.0 ;)
(just right now, since yesterday )
🤪
I have long given up on trying to understand or trust USB devices, or cables.
framework 😍
I don't even get USB 3.2 speeds with a z690 board
I don't want to be that guy but 3831.14MB/s is only 30.65Gb/s, none of them even hit 32Gb/s much less 40Gb/s
In a world where the cloud is the nectar of life, why would buy a USB 4 drive when they can go buy a Big Mac instead.?
Part of the problem is companies playing games to get out of paying the USB royalties.
what nvme can sustain 3 gig writes?
Pcie gen 4 drives, yes even nvme drives have another layer of tech versions.
Given how new that usb4 logo is.......tis a lil too early xd
Maybe next year
Usb for porno drive? Whait, what? Aaa... Usb 4 point 0 drive...