On Tue, Jul 4, 2023, at 16:51, Christian König wrote:
Am 04.07.23 um 16:31 schrieb Arnd Bergmann:Ok, good.
On Tue, Jul 4, 2023, at 14:33, Christian König wrote:Yeah, those are unfortunately very well known to us :(
Modern AMD GPUs have 16GiB of local memory (VRAM), making thoseI'd still like to understand this part better, as we have a lot of
accessible to a CPU which can only handle 32bit addresses by resizing
the BAR is impossible to begin with.
But going a step further even without resizing it is pretty hard to get
that hardware working on such an architecture.
arm64 chips with somewhat flawed PCIe implementations, often with
a tiny 64-bit memory space, but otherwise probably capable of
using a GPU.
What exactly do you expect to happen here?We have tons of x86 systems where we can't resize the BAR (because of
a) Use only part of the VRAM but otherwise work as expected
b) Access all of the VRAM, but at a performance cost for
bank switching?
lack of BIOS setup of the root PCIe windows). So bank switching is still
perfectly supported.
After investigating (which sometimes even includes involving engineersMakes sense, the power usage is clearly going to make this
from ARM) we usually find that those boards doesn't even remotely comply
to the PCIe specification, both regarding power as well as functional
things like DMA coherency.
impossible on a lot of boards. I would have expected noncoherent
DMA to be a solvable problem, since that generally works with
all drivers that use the dma-mapping interfaces correctly,
but I understand that drivers/gpu/* often does its own thing
here, which may make that harder.
Arnd