Mesa hm2/hm2_7i96s.0: error finishing read

bensttech, on your 8300 do you have:

1. AMT disabled in the BIOS
2. IRQ coalescing disabled

Here is another data point:

Computer: Lenovo Thinkcentre M710q Tiny i3-7100T
BIOS:
EIST, hyperthreading, virtualization disabled
Everything under Advanced->Intel Manageability disabled
Only C1 enabled (but enabling C1E and C3 as well makes no difference)

Running 5.10 kernel:
With 1 glxgears and one YT video running I get "hm2/hm2_7i96s.0: Smart Serial port 0: DoIt not cleared from previous servo thread" and/or the error finishing read almost instantly - in fact this screenshot was a bit tricky to get since things fail so fast:




Same test but with 4.19 kernel:
Here I don't get any DoIt or error finishing read messages during the minutes I had it running.

I get the impression the problem has to do with graphics: without the YT video and glxgears running things seems a lot more stable even on the 5.10 kernel (but perhaps not stable enough, haven't tested thoroughly). I notice I get spikes on shifting focus between windows for example (this of course can be noticed with latency-histogram as well for example).

Looking into /var/log/Xorg.0.log I find this info about my graphics card (and driver?):
(II) modeset(0): glamor X acceleration enabled on Mesa Intel(R) HD Graphics 630 (KBL GT2)

Right, but 5.10 works OK on my 8300
That's why I was asking about bensttech's 8300,
as it seems possible to run 5.X kernels without these huge
latency spikes. That is, what is the difference between my
8300 and bensstechs?

Ah, I see. Yes that's interesting! Perhaps you two should have a video call or something to try to figure that out. Barring obvious differences in hardware there is also the possibility of different firmware versions affecting things. Package name firmware-linux I think, as well as maybe intel-microcode.

tommylight wrote:

I doubt it will become standard, ever, as it has to stop plenty of functions used for power saving, and since the era of PC domination is over, laptops and tablets rule now.
Using Preempt-RT on laptops a lot, so i know they use much more power and heat a lot more with it. On most when RT is in use running LinuxCNC the processor will remain at max speed (not turbo) all the time.

I don't know. This article says otherwise.
www.phoronix.com/news/520-Maybe-Real-Time-PREEMPT_RT

Progress was slowed due to lack of funds, but Intel bought the main developer so there is progress again... and the number of patches are steadilly diminishing as the code is pushet to the main line kernel. Its very close.

PCW wrote:

Right, but 5.10 works OK on my 8300
That's why I was asking about bensttech's 8300,
as it seems possible to run 5.X kernels without these huge
latency spikes. That is, what is the difference between my
8300 and bensstechs?

When I looked around for a used 8300, there were two types. Conventional desktop and an All in one. I wonder if that is the difference?

Public announcement, Tower, mini tower, midi tower, SFF, USSF, etc with the same model number, will have differences in electronics on their boards, even the chipset will most likely not be the same diferentiated by a h, or m, or ex, or ......
At one point in time, intel had 14 of the "same" chipset with the same model number but different letters at the end, differences being in bus speed (back then normal), to power consumption (this is still very relevant), built in raid, 2 or 4 or 6 sata, usb 2/3 ports, etc etc.
Slowly but surely forgetting all that cr@p Intel did.
They did good things too, they pushed the power saving on Linux forward as it was terrible 20 years ago, they pushed the "boot in under 10 seconds" from a bloody CD rom, etc.
short version:
same model = not same performance for some things

rodw wrote:

When I looked around for a used 8300, there were two types. Conventional desktop and an All in one. I wonder if that is the difference?

Well, actually there are 5 types: All-in -one with touch, All-in one without touch.  Desktop 'tower', Slim desktop, and ultra slim (USDT). There's actually quite a bit of difference between the USDT and the desktops. Then there's also many possible microprocessors that can be used for any of those.

(Gee, Tommy beat me to it...)

JPL wrote:

(Gee, Tommy beat me to it...)

Sowwyyyy !