[PLUG] Determining if a newer version of a Slackpkg is available

Fri Mar 15 01:55:55 UTC 2019

I have first hand experience in the CPU/computing industry.

It takes a lot of effort and activities to bring new product architecture
to the market. Many of these activities are sequential and just cannot be
done faster.

This is especially true when the resources are limited. Remember, AMD is
relatively small company, and they were not in the position to spend money
they did not have yet when launching there products.

It is not obvious to observe the development and verification time line
when looking at well oiled experience with current Intel architecture. They
have much deeper and wider product development pipeline than anybody else.
So, the development and verification can be done incrementally, often ahead
of product release.

You can get a glimpse of Intel's product pipeline depth when observing
events such as Spectre - discovered in summer 2017, still being worked on.
My guesstimate is that we will see reasonable product solution (not a
patch) some time in 2020 time frame. That is three years worth of
somebody's hard work.

In that light, I would think that AMD's Linux kernel upbringing is in line
with the industry. You can see their long term investment by open sourcing
their drivers. In theory, that should significantly speed up their kernel
contributions as the new products become smaller incremental changes.

This is hard work.

Just my 2c,
Tomas

On Thu, Mar 14, 2019, 4:30 PM Ben Koenig <techkoenig at gmail.com> wrote:

> I was under the impression that you were referring to the CPU hardware
> itself with the statement
> "If such a trivial feature such as power management for Linux can be
> overlooked at a CPU
> launch date I can only assume that CPU still has quite a few bugs to
> knit out."
>
> You are right about the launch day issues, but there's no hardware
> problem here. When a driver sucks, everyone wants to blame the
> hardware. That's not really how it works.... I have 2 Ryzen
> processors, a first gen 1700x and a Raven Ridge 2700U which is a
> mobile ryzen/vega APU. The hardware itself seems fine, but drivers
> have definitely been an issue. Drivers can be fixed, a CPU not so
> much.
>
> Case in point: Official support for Raven Ridge processors was added
> in kernel 4.19.
> - I wasn't able to run halt or reboot
> - sleep/suspend wasn't working
> - GPU would consistently lock up while idle
>
> After 4.19.10...
> - everything works great.
> It was like flipping a light switch... something changed in the kernel
> and after 10 bugfix updates it works as advertised. It's almost as if
> AMD is being intentionally slow.
>
>
> On Thu, Mar 14, 2019 at 11:33 AM Thomas Groman
> <tgrom.automail at nuegia.net> wrote:
> >
> > On Wed, 27 Feb 2019 07:42:39 -0800
> > Ben Koenig <techkoenig at gmail.com> wrote:
> >
> > > On 2/26/19 10:13 PM, Tom wrote:
> > > > On Tue, 26 Feb 2019 18:33:14 -0800
> > > > Ben Koenig <techkoenig at gmail.com> wrote:
> > > >
> > > >> On 2/26/19 7:48 AM, King Beowulf wrote:
> > > >>> On 2/25/19 9:52 PM, Ben Koenig wrote:
> > > >>>> Considering how long it's taken nvidia to fix, I think you are
> > > >>>> allowed to be cranky.
> > > >>>>
> > > >>>> I see a bunch of new downloads on their website, did they finally
> > > >>>> fix it?
> > > >>>>
> > > >>>> Or do I need to place an order for a Vega 56?
> > > >>>>
> > > >>>>
> > > >>>>
> > > >>> You are gainfully employed now. go for ot!
> > > >>>
> > > >>> I haven't tested the Nvidia updates yet for the new Slackware
> > > >>> kernels. In progress.
> > > >>>
> > > >>> -Ed
> > > >>>
> > > >> Bought a core3D sound card instead. Vega will have to wait.
> > > >>
> > > >>
> > > >> Nvidia + Ryzen has been resulting in some lock ups for people. My
> > > >> uptime has been capped at 4 days with nvidia-418.30 :(
> > > >>
> > > >> I'll know by Saturday if 418.43 fixed anything. FWIW my laptop
> > > >> (ryzen 2700U) has been up for 22 days. I even left it in suspend
> > > >> for a solid week and it came back up like a champ.
> > > >>
> > > >>
> > > >> This is on -current, but if my system is still running at the end
> > > >> of the week I'm gonna call the driver "stable".
> > > >>
> > > >> -Ben
> > > >>
> > > >> _______________________________________________
> > > >> PLUG mailing list
> > > >> PLUG at pdxlinux.org
> > > >> http://lists.pdxlinux.org/mailman/listinfo/plug
> > > > I have heard that even some time after the initial launch of AMD
> > > > Ryzen that suspending didn't even work on Linux. If such a trivial
> > > > feature such as power management for Linux can be overlooked at a
> > > > CPU launch date I can only assume that CPU still has quite a few
> > > > bugs to knit out.
> > > >
> > > > Regarding Nvidia,
> > > > https://invidio.us/watch?v=IVpOyKCNZYw
> > >
> > >
> > > Careful with that video on public lists. The systemd community might
> > > try to have you banned in accordance with their rigid CoC.
> > >
> > > It also might behoove you to understand the difference between a
> > > driver bug, and a hardware flaw.
> > >
> > >
> > > AMD is trying to prove that the software stack DOES have an affect on
> > > the functionality of the overall product. This affect can be observed
> > > at both the OS and application levels. We've spent decades assuming
> > > that a "faster Intel CPU" will "make your computer faster". This is
> > > not true, and IMO the best way to dispel such fallacies is through
> > > proof of concept.
> > >
> > > Behold:
> > >
> > > 1) CPU runs like shit
> > >
> > > 2) Software is updated
> > >
> > > 3) CPU runs great
> > >
> > > 4) STUNNING CONCLUSION:
> > >
> > > CPU was not shit.
> > >
> > > Software was shit.
> > >
> > >
> > > AMD is a hardware company. They make great hardware, and this has
> > > always resulted in driver headaches. The reason I look the other way
> > > is because AMD is a hardware company, so I don't actually give a shit
> > > if their software sucks.
> > >
> > > Application of Intel power management algorithms to an AMD processor
> > > caused problems? Really?!
> > >
> > > <sarcasm>I HAD NO IDEA THAT WAS A THING</sarcasm>
> > >
> > > _______________________________________________
> > > PLUG mailing list
> > > PLUG at pdxlinux.org
> > > http://lists.pdxlinux.org/mailman/listinfo/plug
> >
> > It's _IS_ a driver issue if the machine can't enter certain power
> > states. And there is nothing 'Intel' about S power states either. S1,
> > S2, S3, S4 etc. I wasn't talking about software (at least not
> > userspace) but AMD's lack of driver support in Linux for their own
> > flagship CPUs.
> > _______________________________________________
> > PLUG mailing list
> > PLUG at pdxlinux.org
> > http://lists.pdxlinux.org/mailman/listinfo/plug
> _______________________________________________
> PLUG mailing list
> PLUG at pdxlinux.org
> http://lists.pdxlinux.org/mailman/listinfo/plug
>