ChanServ changed the topic of #aarch64-laptops to: Linux support for AArch64 Laptops (Chrome OS Trogdor Devices - Asus NovaGo TP370QL - HP Envy x2 - Lenovo Mixx 630 - Lenovo Yoga C630 - Lenovo ThinkPad X13s - and various other snapdragon laptops) - https://oftc.irclog.whitequark.org/aarch64-laptops
agl has quit [Quit: ZNC 1.8.2+deb3.1+deb12u1 - https://znc.in]
agl has joined #aarch64-laptops
rz has joined #aarch64-laptops
chrisl has joined #aarch64-laptops
chrisl has quit [Ping timeout: 480 seconds]
laine_ has joined #aarch64-laptops
laine has quit [Read error: Connection reset by peer]
lucanis has quit [Ping timeout: 480 seconds]
chrisl has joined #aarch64-laptops
chrisl has quit [Ping timeout: 480 seconds]
tobhe has joined #aarch64-laptops
tobhe_ has quit [Ping timeout: 480 seconds]
tobhe_ has joined #aarch64-laptops
tobhe has quit [Ping timeout: 480 seconds]
chrisl has joined #aarch64-laptops
chrisl has quit [Ping timeout: 480 seconds]
hexdump01 has joined #aarch64-laptops
hexdump0815 has quit [Ping timeout: 480 seconds]
chrisl has joined #aarch64-laptops
chrisl has quit [Ping timeout: 480 seconds]
alfredo has joined #aarch64-laptops
alfredo has quit [Quit: alfredo]
chrisl has joined #aarch64-laptops
chrisl has quit [Ping timeout: 480 seconds]
<steev>
so, which one do i want, the x1e-78-100 or the x1p-42-100?
<steev>
the x1p-42 is 60 dollars less
<steev>
i'm assuming the x1p is purwa?
<steev>
oh interesting, they don't offer the 64gb ram with x1p
<steev>
alright, should be here 12th to the 14th
jhovold has joined #aarch64-laptops
chrisl has joined #aarch64-laptops
SpieringsAE has joined #aarch64-laptops
<SpieringsAE>
dangit I keep missing events, I just realised FOSDEM is in brussels this weekend, now it is too late to arange anything
chrisl has quit [Ping timeout: 480 seconds]
<steev>
ah crud
<JensGlathe[m]>
x1p4* is Purwa
<JensGlathe[m]>
x1p6* is HAmoa
<JensGlathe[m]>
x1e* is Hamoa
<steev>
i didn't wanna wait another 6 months to get 64gb of ram with purwa
<JensGlathe[m]>
Booting from type-c is getting increasingly catty
<JensGlathe[m]>
DevKit most of the time doesn't come up, Thinkbook similar
<JensGlathe[m]>
better use Type-A but it needs to be there in the dtb
<JensGlathe[m]>
steev: Did you buy a T14s?
<steev>
yeah
<steev>
with my f&f discount it was 1337, how could i not?
<JensGlathe[m]>
That's a sign
<JensGlathe[m]>
x1p42 seems to be quite efficient, need to do a native compile run - but as of now its not even properly installed, booted from type-a ssd
<steev>
i do wish they had 64gb option with it, but there was only 16 or 32
<JensGlathe[m]>
- working kb and touchpad on ThinkBook 16
<JensGlathe[m]>
If somebody wants to try a complete-ish dtb for the VivoBook S15, its also in there, testers wanted
<JensGlathe[m]>
to test: dp Altmode support
<JensGlathe[m]>
Its based on SpieringsAE 's patches
<steev>
nice
<smoorgborg[m]>
<steev> "i do wish they had 64gb option..." <- Yeah, then you could have joined the club of us that has to disable half of the ram to get stability 😂
<jhovold>
heh
<steev>
smoorgborg[m]: i will be joining that club :P
<smoorgborg[m]>
It's a really nice laptop, it has such a nice "feel". And I have a lot of ThinkPads - this one is 10/10 IMO. Did you get OLED as well?
<steev>
yeah, oled, 64gb but stuck with 256 because a 2tb is less than 1/2 what they want for a 1tb
<steev>
no wwan though, because i didn't wanna pay 300 for it
<smoorgborg[m]>
Me neither - I just wifi share my phone. But I am curious if the M2 connector is in there or not ... could be useful for something else :-) I don't think it's mounted though, but haven't taken mine apart (I went for the 1TB, as I bought when all RAM / storage configs were priced equally)
<steev>
i think they wanted 490 for the 1TB, and the SN770M 2TB is 180 on amazon
<HdkR>
smoorgborg[m]: Watch out, Lenovo vendor locks the wwan ports. Although on my X1E Lenovo the connector isn't even soldered
<steev>
hopefully they don't send me some other random laptop instead this time too
<smoorgborg[m]>
HdkR: I suspected that - also as far as I can see it's just the USB-port that is connected for the WWAN M2 port (makes sense since it's part of the USB sub board)
<smoorgborg[m]>
On the other hand, if it is just connected over USB they're probably not whitelisting, as a device could be connected to any port at bootup. Guessing here though.
<anonymix007[m]>
I believe the M.2 has the B key and the WWAN card supports both PCIe and USB. Having a second SSD would be more useful imho. But I'm not soldering the connector just to check this.
chrisl has joined #aarch64-laptops
klaus[m] has joined #aarch64-laptops
klaus[m] is now known as Treibholz[m]
chrisl has quit [Ping timeout: 480 seconds]
chrisl has joined #aarch64-laptops
srinik has joined #aarch64-laptops
chrisl has quit [Ping timeout: 480 seconds]
<Treibholz>
I'm currently waiting for my train to Fosdem and found a real big issue with my X13s: the keyboard does not warm my fingers. :-)
<JensGlathe[m]>
X86 ftw?
<kalebris>
Treibholz: compile kernels :) that usually helps with that :)
<kalebris>
but I agree, my everyday use of the x13s makes a very poor handheating device
<Treibholz[m]>
kalebris: I thought about running ollama with deepseek-r1:14b
<kalebris>
right, 6.14-rc1 didn't drop yet :)
<\[m]>
@steev jealous
<\[m]>
lol
<\[m]>
did you get oled?
<smoorgborg[m]>
T14s: I see the bus cable connecting the SUB board to MB is quite wide, so there might be PCIe in there too. Hmm, a connector then possibly would allow a secondary NVMe as suggested
<SpieringsAE>
my vivobook is quite good at scorching legs
<SpieringsAE>
by either A: the charging circuitry getting to like 50 degrees on the outside, or B: the fans not really spinning so when the cpu gets hot the exhaust gets very hot as well with its low flow
<\[m]>
Treibholz could you post if that has acceptable latency?
<\[m]>
HdkR they could support egpu but just no drivers? waw
chrisl has joined #aarch64-laptops
<JensGlathe[m]>
x1p42100 native kernel build run: 49m30s vs 35m on x1e78100 vs 29m30s on x1e001de
<SpieringsAE>
wouldn't we also need usb4 support for egpu? I believe we currently only have usb3 right now?
<Treibholz[m]>
\: I only have deepseek-r1:8b here. The answer for "How much is the fish?" took about 2 sek.
<JensGlathe[m]>
SpieringsAE: apparently, yeah. But how long do we have these devices now, half a year?
<SpieringsAE>
I kind of want to see what this npu can do, so far have not seen much of it
<SpieringsAE>
also I believe PCIE tunneling is not a mandatory usb4 feature, so might not have it at all
<SpieringsAE>
no idea though
chrisl has quit [Ping timeout: 480 seconds]
<SpieringsAE>
dont care much for an egpu, I just want to compile code fast
<JensGlathe[m]>
there is no software available. Even on Windows, this is all BS IMO (completely ignorant of non-opensource in this case)
<JensGlathe[m]>
the x1e'sdothat just fine
<SpieringsAE>
okay nice
<JensGlathe[m]>
I read that PCIE passthrough works with a Surface laptop (some 10GBE card?) on Windows
<JensGlathe[m]>
That ThinkBook has quite some fan, but it isnt whiney turbine like the Dev Kit
<HdkR>
\[m]: No, Apple can't support eGPU on Apple Silicon because of hardware limitations.
<SpieringsAE>
asus has quite a bit of fan too, it just doesn't work under linux lol
<SpieringsAE>
constantly spins at a low speed with no change whatsoever
<SpieringsAE>
on windows it gets quite loud
<SpieringsAE>
I should monitor the temps on windows some time
alfredo has joined #aarch64-laptops
rz_ has joined #aarch64-laptops
rz has quit [Remote host closed the connection]
chrisl has joined #aarch64-laptops
<travmurav[m]>
Jens Glathe: re kernel build times: do you happen to have same kernel build time but cross-compiled on some x86 box? So we can put the perf into the perspective :D
<travmurav[m]>
as in, I'm curious to what on x86 this roughly compares
<travmurav[m]>
but also 30 minutes is the full aarch64 defconfig I guess?
<JensGlathe[m]>
r9-3900 (12 cores 24 threads) build time cross-compiled is ~ x1e78100 native
chrisl has quit [Ping timeout: 480 seconds]
<travmurav[m]>
ohh
<travmurav[m]>
thanks
<JensGlathe[m]>
but you need >2x the power
<travmurav[m]>
yeah that's actually very interesting then
<travmurav[m]>
I guess those huge server cores do the punch
<travmurav[m]>
on x1e that is
<JensGlathe[m]>
probably, yeah.
alfredo1 has joined #aarch64-laptops
<JensGlathe[m]>
I find this downscaling approach odd, but if you actually have a server design that you squeeze into mobile devices, maybe that's the right approach
<travmurav[m]>
well it seems like for 8 elite they put MUCH smaller cores in
<JensGlathe[m]>
the x1p42 Purwa SoC appears pretty efficient
<travmurav[m]>
as in for phone ones
<travmurav[m]>
phone oryon is 1 core that is a bit smaller than x1e and more cores that like 25% size
<travmurav[m]>
hamoa however has 12 same size huge cores
<travmurav[m]>
(contrary to what we thought for 4+8, seems like they're the same)
<travmurav[m]>
at least according to some die shots
<JensGlathe[m]>
they have different wiring
<JensGlathe[m]>
the 4 efficiency cores are stepping 2
alfredo has quit [Read error: Connection reset by peer]
chrisl has quit [Ping timeout: 480 seconds]
srinik has quit [Remote host closed the connection]
chrisl has joined #aarch64-laptops
chrisl has quit [Ping timeout: 480 seconds]
<JensGlathe[m]>
30 minutes is the Ubuntu config limited to the Qcom platform. That‘s a bit deviating from defconfig.
<Treibholz[m]>
My personal benchmark, to compare performance on differenent machines is a python-script, which analyzes the headers of all mails in a Maildir (>100k mails). My overclocked OrangePi5+ does 700 msgs/sec (8 workers), my Workstation Ryzen5 5600X does 2280 msgs/sec (12 workers), the X13s does 1649 msgs/sec (8 workers). - I remember my T14gen3 (i7-1260P) doing about 1800 msg/sec in performance-mode.
<JensGlathe[m]>
in my repo its the ubuntu_x1e_defconfig
<Treibholz[m]>
It's also funny, when I run benchmark on the ryzen with 6 workers only, it does almost 2000 msgs/sec = Hyperthreading just increases the performance by 10%.
chrisl has joined #aarch64-laptops
srinik has joined #aarch64-laptops
<albsen[m]>
Treibholz: I've noticed that compiling the kernel with all 8 threads results in compilation errors on the x13s, now I use 4 and its stable
chrisl has quit [Ping timeout: 480 seconds]
<Treibholz[m]>
albsen: that's strange. this shouldn't happen.
<Treibholz[m]>
I mean, for the compiler the only difference between the X1C and the A78C should be the speed...
<JensGlathe[m]>
2 possible issues: trip points too high, thermal paste degraded / missing
<Treibholz[m]>
albsen: to validate Jens Glathe, try it outside, when its's cold :-)
<JensGlathe[m]>
I had these effects, too, but got mittigated with adjustments in the dtsi / dts
<JensGlathe[m]>
oh and applying new thermal paste
<albsen[m]>
certainly getting to hot
alfredo has joined #aarch64-laptops
<jhovold>
albsen[m]: have you verified that thermal throttling is working in your setup (e.g. that your distro enables all the needed drivers)?
kalebris_ has joined #aarch64-laptops
kalebris has quit [Ping timeout: 480 seconds]
<albsen[m]>
jhovold: no I didn't, I expected that the hardware does that by itself. what should I check? I'm running 6.10.0-rc3-20240621-g959d8c14d757 on x13s
chrisl has joined #aarch64-laptops
<jhovold>
yeah, there's some hw (fw) throttling of the cpus, but the kernel also reads the skin temperature sensor and throttles the cpu based on that
<jhovold>
otherwise you'll burn your fingers
<jhovold>
sounds like you build your own kernel, and if you use a config based on johan_defconfig you should be fine
<jhovold>
but it's conceivable that some distro may have missed enabling some of the drivers needed
chrisl has quit [Ping timeout: 480 seconds]
alfredo has quit [Remote host closed the connection]
SpieringsAE has quit [Quit: Leaving]
<albsen[m]>
jhovold: correct, I'm using a modified version of johan_defconfig
<albsen[m]>
it only happens if u run all 8 threads at max for some time. and I am in a hotter country as well which likely doesnt help :)
<bamse>
no polar bears roaming the streets, like where jhovold lives?
<bamse>
albsen[m]: but you shouldn't get compile errors from temperature issues...the temperature of the ddr and cpu cores would be throttled by hardware, linux doesn't have a chance to keep that in check
<bamse>
albsen[m]: and you have both cpu*_thermal and mem_thermal exposed as thermal zones, so you should be able to keep an eye on those
<Treibholz[m]>
btw. I sent a pull-request to htop, to see see the per core-temperature of the Snapdragon 8cx in the overview. https://github.com/htop-dev/htop/pull/1589
<Treibholz[m]>
maybe somebody wants to test, if it als works other (newer?) Snapdragons?
<bamse>
Treibholz[m]: cool!
chrisl has joined #aarch64-laptops
<bamse>
Treibholz[m]: fwiw, when you write commit messages, it's quite useful for the reviewer, or future reader of the git history if you describe the problem that you're solving...then the reader doesn't need to guess why the change was made or which problem(s) was actually solved
<robclark>
fwiw, at least the 7x, it seems like the EC reads the skin sensors and controls the fans on it's own? Idk if that is different from some of the other laptops?
chrisl has quit [Ping timeout: 480 seconds]
<Treibholz[m]>
bamse: thank you for this kind advice. I just wanted my local fix for myself not to rot on filesystem. But tf submitting a PR takes more effort, than finding AND fixing the problem, then people will loose their interest in submitting PRs.
<Treibholz[m]>
* bamse: thank you for this kind advice. I just wanted my local fix for myself not to rot on my filesystem. But if submitting a PR takes more effort, than finding AND fixing the problem, then people will loose their interest in submitting PRs.
<anthony25>
I don't think that the yoga 7x being hot is a problem with the fans, the fans speed up quite early, but the CPU isn't throttled
<bamse>
robclark: the xps13 magically spins up the fans as well...not sure what it's based on, and there's a few thousand lines of fan-related things in the dsdt...
<anthony25>
if the fans don't get enough air and I do a full kernel compilation, the laptop can crash
<bamse>
anthony25: in particular the cores would overheat way before the fan has any effect
<anthony25>
like immediate reboot
<anthony25>
bamse: what do you mean?
<bamse>
Treibholz[m]: it's a bit of a hazzle, but really nice to see things improving across the system
<robclark>
hmm, I haven't tried blocking the fans.. I don't think I've seen anything throttle.. the fans defn kick in when building kernel, but they aren't so loud. I guess if the cpu's aren't throttling, then the fans are effective too.
<bamse>
anthony25: i mean that the fan's effect on the core temperature is rather slow compared to how rapidly the temperature of the core can rise
<anthony25>
it doesn't really work this way, does it? the laptop has a heat capacity it can handle and dissipate without the fans
<bamse>
anthony25: sure, but plot the cpu core temperature at the time of starting a "make -j99"
<anthony25>
the crash doesn't happen right away, it's after 10min or so
<bamse>
anthony25: what i'm saying is that the cores should be throttled by the hardware in the soc, much much faster than what dissipation of the fans does
<anthony25>
when the chassis of the laptop becomes quite hot, so if you're compiling a kernel in your bed (heh, don't judge me :D), the airflow can be restricted and at one point it reaches the temp max and the firmware triggers a reboot
<bamse>
which temp_max? core temp_max?
<anthony25>
bamse: it makes sense, but I don't see the frequency being lowered
<bamse>
anthony25: that would imply that either we don't get thermal pressure interrupts or there's an issue with the readback... the kernel will request a frequency and the hardware will throttle that behind our backs
<bamse>
well, i presume the read back value wouldn't be reflected in the cpufreq stats even...
<anthony25>
bamse: sorry, I don't know how it's named on the x1e sensor, I meant the critical core temperature before "something" (the firmware?) triggers a hard reboot for safety
<bamse>
anthony25: but for that you should have time to see cpuN_thermal go beyond 100C
<bamse>
it seems rather to me, that hardware throttles the cores and fans are dissipating things...but as we continue to heat the whole machine, there's something else that get overheated
<bamse>
or we could have e.g. voltage votes that are inadequate and that shows as instability as the temperature rises
<anthony25>
I'll double check next time it happens, I didn't look at the temperature in real time, I just had htop running and didn't see the core frequency dropping
<anthony25>
they stay at 3.4GHz
<anthony25>
<bamse> or we could have e.g. voltage votes that are inadequate and that shows as instability as the temperature rises << that would make sense
<bamse>
doesn't seem like anyone has added debugcc support for this platform :(
<bamse>
anthony25: but as i said, i think the cpufreq vote is as 3.4ghz...that just means that we request the hardware to keep us at that frequency, not that we're running at that
chrisl has joined #aarch64-laptops
<anthony25>
ha sorry, I didn't know that, I thought that the core frequency reported was the frequency the cores were running at, not the cpufreq vote
<jhovold>
anthony25: there's no throttling yet based on skin temperature, the cpus still do some throttling in hw/fw, but you can still burn yourself
<anthony25>
ha ok, that's what you meant
chrisl has quit [Ping timeout: 480 seconds]
cyrinux has quit []
cyrinux has joined #aarch64-laptops
<bamse>
anthony25: the overall idea is that we let the hardware protect the cpu cores etc - because that needs very fast response - and then we use software deal with slower events, such as protecting the user from the kernel build burning their lap
<bamse>
anthony25: that said, this is obviously a rather simplified representation of the world, so more work needs to go into this
chrisl has joined #aarch64-laptops
<anthony25>
I see, thanks!
chrisl has quit [Ping timeout: 480 seconds]
srinik has quit [Ping timeout: 480 seconds]
chrisl has joined #aarch64-laptops
chrisl has quit [Ping timeout: 480 seconds]
chrisl has joined #aarch64-laptops
srinik has joined #aarch64-laptops
chrisl has quit [Ping timeout: 480 seconds]
laine__ has joined #aarch64-laptops
laine_ has quit [Read error: Connection reset by peer]
srinik has quit [Ping timeout: 480 seconds]
chrisl has joined #aarch64-laptops
chrisl has quit [Ping timeout: 480 seconds]
laine__ has quit [Remote host closed the connection]
laine__ has joined #aarch64-laptops
laine__ has quit [Remote host closed the connection]
laine has joined #aarch64-laptops
<JensGlathe[m]>
travmurav: cross-build times on 12/24 core VM (R9-3950X unconstrained power limits): 37m44s
<JensGlathe[m]>
That's quite impressive, and consistent with my previous measurements. The impressive part is that the x1e-78-100 beats this setup with 35m