[Arm-netbook] Somewhat off-topic updates about Ryzen, PolarFire, HiFive and Elbrus (was Re: Current status)

[Luke Kenneth Casson Leighton]

---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68

On Sat, Nov 28, 2020 at 10:23 PM Paul Boddie <paul at boddie.org.uk> wrote:
>
> On Tuesday, 21 July 2020 23:36:28 CET Paul Boddie wrote:
> >
> > So, upcoming AMD products could be decent, but I would be wary about
> > stability for a while after their release. Following the Linux kernel bug
> > tracker can be informative:
> >
> > https://bugzilla.kernel.org/
> >
> > Searching for "amdgpu" is probably what you want to do. For quite some time,
> > people were having problems with the 3200G, and that made me worried about
> > the 3400G, but it seems that even within product families there might be
> > some parts that are better supported than others.
>
> Since then, my Ryzen system has worked better than my work laptop that has an
> Intel Core 7 CPU with integrated graphics that aren't used because there is
> also an Nvidia video card which occasionally does some very weird display
> memory "picture interference" thing.
>
> (I thought that might be the HDMI output playing up, having had to mess with
> Synopsys HDMI on the Ingenic JZ4780 and seeing the many weird things that have
> to be set up to enable the peripheral, but is probably some dodgy interaction
> between the binary firmware blobs and the baroque "Linux plus GNOME plus
> whatever is in-between" graphics stack.)
>
> The only weird thing I have seen with the Ryzen 3400G is this kind of message
> (with dmesg prefixes removed):
>
> Corrected error, no action required.
> CPU:7 (17:18:1) MC3_STATUS[-|CE|MiscV|-|-|-|SyndV|-|-|-]: 0x9820000000000150
> IPID: 0x000300b000000000, Syndrome: 0x000000002a000503
> Decode Unit Ext. Error Code: 0, Micro-op cache tag parity error.
> cache level: RESV, tx: INSN, mem-tx: IRD

my gaming laptop (also i7) i bought 2-3 years ago now had immediate
problems which i "resolved" by cranking down the PCIe bus speed to
level "2".  this stopped devices disappearing off of the PCIe bus,
such as "the entire USB subsystem" or "the NVMe SSD".

i suspect that the Reference Design (supplied by intel to laptop
manufacturers) simply failed to provide sufficient power, and
consequently would cause devices to go unstable and drop off the PCIe
bus if they ran above a certain speed, draining a certain amount of
power.

here were the options i ended up using in grub.cfg:

 ro  apparmor=0 nouveau.blacklist=1
nvme_core.default_ps_max_latency_us=0 acpi_backlight=vendor
acpi_os_name="Windows 2015" acpi_osi=! noapic pcie_aspm=off
mem_sleep_default=deep

there may have been more.

l.