hot take: USB 3 / USB type-C sucks. speeds over 1 Gbps? many motherboards only route PCI-e 3 x1 to the USB controller so nope Ethernet alternate mode? doesn't exist Thunderbolt alternate mode? exists but Thunderbolt chipsets are fucking unobtainium without public docs
I don't know a good solution for that kind of data rate, but I know a design that gives an external peripheral easily plugged in an external port full memory access and ring0 code-exec is an idiotically bad one.
-
-
"Option ROMs are bad" sure, regardless of Thunderbolt "OS/firmware should enable IOMMU" sure, regardless of Thunderbolt "Thunderbolt is awful" I lost you
-
Aren't option ROMs and lack of enforced IOMMU (i.e. a policy where nothing functions until OS sets up IOMMU) part of the host spec for Thunderbolt?
-
nope, Intel recommends using IOMMU https://firmware.intel.com/sites/default/files/Intel_WhitePaper_Using_IOMMU_for_DMA_Protection_in_UEFI.pdf …
-
also, with Falcon Ridge and later you don't even need IOMMU, AFAICT http://lkml.iu.edu/hypermail/linux/kernel/1705.2/02534.html …
End of conversation
New conversation -
-
-
My understanding is that from a capabilities/design standpoint USB3 is much less horrible, but can't reach the data rate you want.
-
no every version of USB is an overcomplicated piece of shit in USB 3 they fixed their bizarro physical layer (it uses 8b10b now, which is still dumb, but much better than the incredibly moronic variable width encoding it used to have) but everything else is just as bad
-
Wait, USB isn't using 64b/66b?
-
I think USB 3.1 uses 128b/130b, like PCIe 3.0?
-
yeah, I was talking about 3.0, I forgot 3.1 is a thing
-
Doesn't 3.1 also have a "Gen 1" (still 5G) and "Gen 2" (10G)? This branding is confusing and terrible.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.