mothcompute

great osdev actually 🏳️‍⚧️

@mothcompute

real internet bug

guys help im frozen in time

i post more on my FediPub Activityverse: @mothcompute@vixen.zone it is where i talk about all my fun projects

log inask
mothcompute
@mothcompute
cathoderaydude
@cathoderaydude
cathoderaydude
@cathoderaydude

hahaha fuck usb fuck thunderbolt fuck intel

i have never used thunderbolt before, because I knew it was a joke that you couldn't rely on. i have ignored all suggestions that i should solve this problem or that one with a thunderbolt peripheral because i knew it would be a waste of my time and literally any other solution would be more sustainable. i have been fully vindicated.

Thew
@Thew

my favorite example of this

particularly the "99% of usb cables are counterfeits" bit, is trying to plug in a cintiq

cathoderaydude
@cathoderaydude

i am constantly marveling at how unclothed this particular emperor (meaning USB, not @thew) is. i have twenty years of experience that say no, actually, all USB cables are fake except for the incredibly stiff 8"-long one that came with your device, and that will stop working after 6 months when a solder joint loosens up. people will tell me they don't have these problems, and that is so hard for me to believe that i've actually wondered if, for some unscrutable reason, they're lying.

Xylaria
@Xylaria

oh my FUCKING god USB is trash

doing high level tech support has made me hate every single motherfucker involved in determining cabling standards because they all FUCKING SUCK AT THEIR GOD DAMN JOBS

I primarily deploy Macbooks, thank fuck, because if you get something labeled TB4, you can generally just plug it into the fucking Macbook and it works.

Every SINGLE time I leave that realm, I start screaming in horror when things break. Stupid fucking docks are the absolute worst offenders - there's not a SINGLE dock manufacturer I trust as far as I throw them to have their cables and standards right. And I can chuck a dock quite a ways, ask me how I know.

Don't get me started on USB-C speeds. Did you know that there are iPhone 15s that WILL NOT CHARGE off USB-C cables because they're the wrong spec, despite there being ABSOLUTELY NO WAY TO KNOW WHAT SPEC THE CABLE IS, AND THE LABELING BEING FUCKING WRONG.

And the opposite is true too - Got an older USB-C device that takes the low-power standards? GOOD FUCKING LUCK GETTING A NEW CABLE, your choices are either gambling on knockoffs having the exact right voltage or getting a manufacturer-provided cable and hoping, because guess what? The new higher power cables DON'T HAVE A STEPDOWN MODE FOR OLD DEVICES FOR SOME FUCKING REASON, so they just don't negotiate with them at all. For. Some. Fucking. Reason.

I don't know the standards and specs, they're all written like garbage and have the dumbest most confusing names in fucking existence, and none of them matter because NO ONE FOLLOWS THEM ANYWAY.

It's the wild goddamn west out here and holy shit, I'm starting to miss dedicated power bricks, and that's saying something. AT LEAST YOU KNEW WHEN THEY WERE DIFFERENT.

cathoderaydude
@cathoderaydude

yeah exactly it's almost like things were better when incompatible devices (usually) didn't fit together

You must log in to comment.

in reply to @cathoderaydude's post:

plumpan
@plumpan

old TB support being dropped on a bios update is REAL shitty. I don't think there's any TB3 firewire adapters so that just straight up kills firewire support, unless TB1/2>3 adapters fix this (doubtful). Good to know.

login to reply
UnderpaidVillain
@UnderpaidVillain

I don’t know if it’s Dell’s fault or Intel’s fault, but they are actively making their USB3 support worse with newer generations. I have a Precision 7740 for work that plugs into their dual cable dock (which is barely above trash quality on a good day), and it does mostly fine if I don’t look at it wrong.

If I plug that same dock into a Precision 7770, it loses the ability to actually put the machine to sleep. Sure, the monitors and fans turn off, but the processor doesn’t. It just sits there, getting hot with no fans running in the hopes that someone wakes the machine up before the processor causes Windows to blue screen from a thermal event.

login to reply

in reply to @Thew's post:

MisutaaAsriel
@MisutaaAsriel

It's more or less because Apple was the only one who actually cared about making Thunderbolt work.

Point & Case: I owned an AMD eGPU from Sonnet when I was still using a 2017 MacBook Pro. Plug it in to macOS, it works. Software support varied, but that's a different issue.

I dual boot the machine with Windows, because I want to play PC games.

It doesn't work. Not enough memory it says, which is absurd because this machine has 16GB and barely any of it is being touched.

Specifically, Intel, in their infinite wisdom, decided that Thunderbolt devices in their chipset driver were mapped in 32-bit address space for the particular chipset this Mac used.

Now, anything POSIX, from Linux to Mac, don't care what address space Intel says to use; they will map drivers accordingly based on the RAM available. So if you're trying to run a GPU over it, they won't care if the first 4GB is full as long as you got enough RAM elsewhere.

Windows doesn't. It trusts the hardware when it says "I can only be mapped to the first 4GB and drivers are limited to [X]GB!" like the fool it is. I had to add an EFI pre-loader to my system partition to patch the chipset to allow use of the full 16GB of RAM with external Thunderbolt devices, then hand off to the Windows bootloader.

And there's more issues like that too. Not to mention that, at least once upon a time, You couldn't add Thunderbolt to AMD & older Intel systems. Intel refused to allow the TB3 controllers to be put onto a PCIe card — Motherboard only, modern Intel chipset only. — TB3 PCIe cards were just a glorified port, and you had to hook a little extra cable up to a special port on the motherboard that connected it to its controller.

Icing on the cake is, unlike TB 1 & 2, Intel was hella stingy with the licensing with v3. On their website, they once listed how they would provide access to the specifications, materials, and purchase information for Thunderbolt & its controllers to hobbyists and educational institutions for non-commercial purposes. Im thinking "Thunderbolt enclosures are expensive; surely making my own would be cheaper, right?" So, I contact them, and things start off well, but the minute I mention I'm looking at Thunderbolt 3? "Sorry, we do not provide access to materials, documentation, or other information regarding Thunderbolt 3 at this time for the educational and hobbyists markets. Ciao!"

It's probably why TB3+ shit is so fucking expensive.

login to reply
MisutaaAsriel
@MisutaaAsriel

That is hilarious if true. I assumed it was just corporate greed1 trying to extract licensing fees out of everyone and force Intel supremacy right around the time eGPUs & other high-performance external PCIe devices were getting popularity.

If it was more so just "we can't reliably make these things and our drivers are shit" that'd be much more amusing, albeit also more mundane.

  1. At the time, TB3 enclosures were, like, 2x the price of their TB2 brethren.

login to reply
Mayabotics
@Mayabotics

That they had previously made the offer to provide documentation and pricing to hobbyists and education is what makes me think that way. If they intended to keep it unavailable for licensing reasons, they wouldn't have made the offer to begin with.

The exception being if someone said/posted something without getting clearance from legal, but generally, I would be very surprised they made an offer like that without having the necessary 'take it or leave it' NDA & end use agreements drawn up by legal to be used with hobbyists.

That suggests to me that they thought that hobbyists and education putting those chips into a design wouldn't be a drain on their application support engineers' time or risk diverting significant amounts of chips.

When chip vendors are having yield issues, they start doing things like restricting who is eligible to purchase, and requiring customers to contract with specified design partners for board design and prototype production before the chip vendor will authorize the sale of parts needed for full-scale manufacturing.

The other thing that just occurred to me as an alternate possibility, is that export controls have been ramped up over the last several years. Doing the necessary end-use & end-user due diligence is much harder & expensive for potential hobbyists and education customers.

login to reply
MisutaaAsriel
@MisutaaAsriel

Maybe I wasn't clear initially but the gist was they would only provide access to information for versions 1 & 2. The website never specified which versions you'd get access to; they merely implied you'd get access to all of them. So I contacted them in hopes of getting documentation for Thunderbolt 3 (the one good for GPUs) only for them to straight up send me to a different department that insta-rejected me the moment I mentioned it was for the third generation stuff.

And that was probably the weirdest part. Thunderbolt 3 was not the normal Thunderbolt people's department. It was like it was under lock and key at Intel, as if it were the liquor cabinet in culinary school.

You might be right though.

login to reply
rosieposie
@rosieposie

i literally got a larger cintiq - which separates out all the functions into individual wires - because i couldnt stand the CONSTANT cable problems on the smaller one. i gave the smaller cintiq to a friend (for free because it is a hunk of junk without that cable!!!) hoping that they could solve the problems and they COULD NOT. usb is HELL

login to reply
lapisnev
@lapisnev

I haven't run into this problem with USB cables. I've had them outright fail but I've never had the cable be the reason a connection is downgraded. I've seen a flaky USB 3.0 controller that would lock up and reset if you tried to push an entire 480 Mbps...

I HAVE run into problems with Firewire cables funny enough. If you have a shit FireWire cable devices will blink for a split second when you plug them in and then not work. It's always the four-pin cables. If your Firewire camcorder cable is flaky then it will work some of the time or in one computer but not another, and changing the cable magically fixes it. There are no compliant Firewire 4-to-6 cables you can buy new today, you must find them in vintage computing lots on eBay and the like, thick and beige with chokes at both ends like mama used to make.

login to reply

in reply to @cathoderaydude's post:

JNP-3R
@JNP-3R

The amount of times people are like "It's a good cable. I just bought it!" Jeez...

Sometimes it's like folx aren't willing to perceive issues. Like, even with other pervasive issues, but tech where the average user (and even some savvy ones) won't be able to validate claims makes it harder. Someone lies to them and they just repeat it, thinking the issue must be with them, or their device, or the config, or somehow be isolated to this instance, even though this is the 15th time this has happened.

And when you ask someone if they had the problem it gets forgotten and memory holed. The issue is unpleasent and not in front of them, so it doesn't matter, but they do remember the factoids they were told a while back, and wow these nice cables are cheap, what a deal!

This one's lucky to only need USB to do USB 2.0 things, but even then there's enough rooting around for even like, a working headphone adapter, that it understands

login to reply
FoxBall
@FoxBall

I must be the outlier here then, I’ve literally never had any A type, B type, or C types fail on me, even one’s I’ve accidentally stepped on and bent. I’ve had multiple mini types and several micro connectors fail however. Honestly the least reliable type of cable for me is 3.5mm headphone cables. They are practically consumables! The shortest lifespan I’ve had is 2 weeks. Every failure is the same: one audio channel becomes flaky and angle dependent, then drops out, making the other go weird as well.

login to reply
petra
@petra

On the one hand I'm one of those people who mostly doesn't have these problems. On the other hand I go through life just assuming that a given USB thing will only ever be working at 2.0 speeds no-matter what it claims and anything faster is a happy accident, so maybe I'm just dealing with this shit it in a different way

login to reply
Mayabotics
@Mayabotics

What most people need external USB cables for, they neither need nowhere near the full bandwidth of even USB 2.0, or it's an application where they can't tell that transfers are much slower than they should. (If it just doesn't work, they think that the cable, or more likely the device, is just straight up broken.)

Like SD cards, HDMI, and DisplayPort cables, the only way you can be reasonably confident of getting a USB cable that actually fulfills your requirements without commissioning a manufacturer to produce them, is to find a manufacturer that provides compliance documentation for their products and then buy from an authorized distributor that sells those exact product SKUs in low unit quantities. (And even with all that, fakes still can get into the distribution channel.)

I'm still just a bit bitter, even though it's been years since I've last had to do acceptance inspection and testing for work.

login to reply
sirocyl
@sirocyl

I've been having the particular issues of:

  • type C ports completely loosing their grip or losing connectibility, where wiggling even slightly causing rapid connect/disconnect cycles
  • type C PD chargers (captive cable) having said cable freeze up and stiffen into an awkward shape, or lose connection randomly from stress, and no longer work
    • in at least one case, this has blown a hole in a "PD" charger cable. 🔥
    • in another case, the strain relief came completely undone from the endpoint plug and became an obvious hazard when it began sparking and causing the connected PC to blip off
  • random - almost completely random - type C cables not being able to carry PD for some devices, meaning my 9001-watt Type C PD++++ superbrick will only pump 5w over this cable. price nor OEM nor advertised feature set matter; it is entirely a crapshoot
    • the only exception is cables labeled "thunderbolt", which cost $120/meter at minimum and still have all the reliability and lifecycle issues of non-TB cables, just that they're not as replaceable when they inevitably fail
  • also random cables not being able to carry DP alt-mode, or USB3-compatible cables falling back to USB2 when the SS pairs come undone in the cable or the plug ends
login to reply
cathoderaydude
@cathoderaydude

i do not consider USB-PD real and i am astonished anyone does. the idea that you could trust it is so absurd. every single smartphone i have ever owned since 2011 says "Charging Slowly" if you plug it into anything other than the brick and cable it came with.

login to reply
cathoderaydude
@cathoderaydude

i consider laptops that charge off USB ports to be on par with teslas. people pointing at it going "see, see, it works" and i'm just speechless that that seems like an adequate justification for buying one in light of. you know. everything else

login to reply
sirocyl
@sirocyl

with you there. everything should have the dell/HP standard nineteen volt barrel jack. for smaller laptops and tablets, use the other dell/HP standard nineteen volt barrel jack.

I am currently using the steam deck's charger to keep my t490s running, as it's the only post-2013 laptop I have that currently works, barring chromebooks and repurposed e-waste. both of my OEM type C laptop bricks for this have gone stiff or busted the cable, and they're $75+ for a replacement, as opposed to a steam deck charger which is still $30 and still pricey, but works for now

login to reply
cathoderaydude
@cathoderaydude

it's completely absurd to me that they moved away from the barrel jack. this was a solved problem. we were INCHES away from having all laptops standardized onto Nineteen Volts Four Point Seven Two Amps Two Point One Millimeter Center Pin. we were so close.

login to reply
sirocyl
@sirocyl

I think my solution, honestly, to type-C PD charger woes, is to test on a diverse suite of devices (steam deck, thinkpad T490s, nintendo switch and pro controller, a six year old flagship LG phone, and some mediatek sludge) and literally throw out/refund any non-OEM charger that either refuses to charge a device, or doesn't charge all of them without "slow charger!" warnings.

the OEM ones go in a drawer as a ward against all my proper PD chargers blowing up.

as for cables, it's the same test, except now also carrying video and usb3 - if it fails that, it gets a red mark and gets set aside for low-impact uses.

I keep a bound up 7-inch tb3 cable in my everyday carry, because it's a known-good. I rarely use it until my other options are exhausted.

login to reply
cathoderaydude
@cathoderaydude

it's cool how if you buy a lightbulb and screw it into a socket it works, and if you buy a gallon of gas it doesn't make your engine seize, and if you buy a AA battery it runs your penlight, but if you buy a USB cable you have to become a standards validation facility. definitely a real technology and not a fraud perpetrated on the public

login to reply
sirocyl
@sirocyl

the standards they validate:

  • 5v 2a usb 2.0 (maybe)
  • gracefully degrades from (i.e., doesn't support) usb 3, pd, alt-modes, tb, e-marker, or anything but the above
  • puts power and data on the correct pins (usually)
  • doesn't set itself on fire or damage the device (usually, with many slippages.)

the standards I validate:

  • the rest of the fucking owl

login to reply
A comment has been hidden by the page which made this post.
modulusshift
@modulusshift

frowns in constant engine knocking for the last week

But to be fair I have a weird Swedish car now, it's just picky AFAICT. Not going back to that gas station for sure tho.

Anyway USB is still tons worse. I'm fascinated at the kinds of voodoo people learn about USB through trial and error, and unlike most of computers I have absolutely no standing to tell them they're wrong.

HDMI cables, they work or they don't, you probably don't need the top speed they offer anyway, why the fuck buy gold plated ones, it's a scam.

USB cables, brought to you by "ba DUH, DUH duh" (windows disconnect/reconnect sound if it isn't clear) Who knows, maybe the diamond ones are actually in spec, worth a shot! Still cheaper than buying the 6" Apple TB4 cable!

login to reply
DasterinTempus
@DasterinTempus

The last 3 phones I have had (Google Pixels!!!) I have replaced in like a year because the USB-C charging port completely fails in like 9 months if you plug it in and out frequently. Like if you use it in your car, an advertised feature of both my phone and car.

This time I just got one that wirelessly charges so I can continue using it with its nonfunctional port.

login to reply
EpsilDelta
@EpsilDelta

Specific instances where I can remember I've had issues with USB:

The plastic piece that holds the connectors together around the pins decoupled on a case I had. Made the port basically unusable.

I've had to rummage around for the "right" cable countless times if something needs a certain amount of power.

Speaking of power, my motherboard tends to not do USB 3 speeds in certain ports if something else is connected to the same USB rail. I understand why this happens, but... It shouldn't? The spec should account for this???

My cables tend to be reliable, but I've had some failures.

Overall, it's nice how universal of a standard it is, because ports should be standardized, but it's got a lot of issues that need to be fixed.

login to reply

in reply to @cathoderaydude's post:

Xuelder
@Xuelder

No but seriously, I worked with a lot of legacy hardware, and good god I never see these problems on Serial or Printer ports. Now, this is probably because these devices came out in mid/late 90s when we got this shit down pat, but still. The most annoying part was when the regular users would keep complaining that the serial port connected device was broken when it always, without fail, would be the Serial Port to USB dongle that was broken, the actual serial cable is as old as I am and finally just died thanks to normal degridation, or they plugged it into a bad port, etc. And of course we can't get desktop workstations, so no pci-e cards. USB makes me want to become a primativist.

login to reply