USB Cables

I assume that to achieve the highest speed the two ends have to negotiate. Is it possible that there is some issue with negotiating between the two ends? Or that the built in intelligence in the connectors is faulty?

Cheers

Dave R

For C to C cables the trick is to buy 'Thunderbolt' cables as those definitely have the extra pairs wired or they wouldn't work at all for Thunderbolt. For A to C cables you either have to buy a tester or see if devices come up as USB 2 or 3.

Theo

(who has just binned a microUSB cable that had no data connection at all)

Has the A->C cable got the extra 5 pins wired? Does it have any 5 Gbps or 10 Gbps marking on it?

Look for the "Superspeed" or "Superspeed+" logo as this should ensure the extra pairs are in place. They may instead be marked USB 5Gb/s or USB 10Gb/s. As mentioned above, Thunderbolt also requires these extra pairs.

I know next to nothing about the various USB standards, but I'm utterly unsurprised that the USB-C cable works best.

One of the issues is that loads of these cables are sold just for charging, and nobody really expects them to be used for transferring data.

We were just discussing this over in uk.comp.sys.mac.

For cables with a USB A plug capable of USB 3 speeds, the USB A plug should have a blue (rather than white or black) insert.

USB C cables capable of USB 3 speeds may (but don't always) have an "SS" logo on the plugs with 5 or 10 next to it indicating the maximum speed in Gb/s.

-- Richard

An update as I am having some issues.

My Asus Z170-K mobo has on the back:

ASMedia USB3.1 Controller 2 x Type A connectors

It also has 2 Blue Type A connectors on the front.

I have tried a variety of USB cables including the new one that is USB 3.1 with following results:

A variety of SSD drives and an NVMe in caddies work from all the USB sockets. The Terramaster DAS works from the type C socket on the back and the Type A socket on the front. It doesn't recognise the drives when using the Type A (USB 3.1) ASMedia controller on the back, the fast one!

Before my brain dies on me has anybody come across a similar situation with an ASMedia controller?

I am thinking of getting a USB 3.2 PCIe card with a Type C connector but it the Terramaster is this picky I wonder ho good it is as a back up?

Just to close this off. It seems the Asmedia USB 3.1 controller isn't up to spec as it works fine on my Z790 using my new cable.

I think I will buy an add in card to get the extra speed.

Try one of these.

(2) USB Type-A using USB 3.2 Gen 2 (10 Gbps) [this is red-tab on some motherboards]

PCI-Express 3.0 x4 Slot (can run on PCIe Rev 2 as well)

The purpose of owning one of these, is to ensure that "at least the blue tab connector delivers the 500 I expect".

Do not believe any over-promised result when dealing with older hardware.

USB 3.1 Gen 1 – SS, 5 Gbit/s signaling rate over 1 lane, 8b/10b encoding (raw data rate: 500 MB/s); UASP will be less USB 3.1 Gen 2 – SS+, 10 Gbit/s signaling rate over 1 lane, 128b/132b encoding (raw data rate: 1212 MB/s). UASP will be less

inadequate 4 lanes buffers??? PCIe Rev1.1 8b10b 0.25GB/sec 1GB/sec ~0.5GB/sec Likely "doesn't bench good for our 500 wish" PCIe Rev2 8b10b 0.50GB/sec 2GB/sec ~1.0GB/sec Should do a good job on our 500 wish PCIe Rev3 128b/130b 0.985GB/sec 3.94GB/sec ??? Most likely to be able to use both ports

The deal (as I understand it, and without evidence), goes like this.

When PCIe came out, the company that made PCIe bridges and switches, produced a technical document about PCIe. It showed how the buffer size in the Intel hub, could have an impact on the "wire rate" we were promised. Intel used some sort of small buffer at the hub in the CPU, which could have an influence on performance. To simplify matters, I don't show you the whole curve, and merely use a derating factor of "50%" on old Intel hardware. I do not know when this was fixed, but some evidence suggests that around PCIe Rev3 times, Intel did *something* to fix this. We're getting better than the 50% value, today.

So anyway, the idea is, you buy a card with red-tab capability, and expect when it is jammed into the x4 slot in your older Intel, that it will be able to sustain the desired blue-tab rate. That's why I bought the card above. It doesn't even sit in the PC right now, as I don't have the right kit for benching it :-) The 4930K has 42 lanes, even if they're rather slow lanes.

There are probably better cards than that, but there is an abyss in the USB standards, between USB3.2 2x2 and USB4, where some USB4 hardwares do not do USB3.2 2x2, and it is up to the consumer to do the research and figure out what USB-C they should really be buying for some home project. Maybe if you had a USB4 port, you would need "yet another hardware-crap" to intercede with your peripheral.

For example, this would use the AsMedia chip that's been out for a year or a bit more. Does it actually do 40Gbit/sec ? We need someone to bench it and *prove* it does that. It might only do 32Gbit/sec for example. How many boards have x4 slots (off PCH) that are PCIe Rev4 ?

Any time they build add-in chippage, they always short-sheet the bus connection on it. It makes me grind my teeth. (If it is soldered to the outside of the PCH, it is slow, if it is inside the PCH it might be OK.)

Anyway, at least I can show you my philosophy, of "smothering any I/O gremlins" by using more lanes than the job needs. I can shove that x4 card into my

Paul

I generally like startech products

Wouldn't surprise me if that is the USB3 chipset on the existing M/B

I went for this: