New LG Gram 17

There are reflections where there is a lot of light. Playing darker games is difficult in that kind of light with the reflection. I don't have too many issues in general usability in brighter areas though since the screen content is typically brighter. Granted, I don't take my laptop outside a whole lot if at all. I do have a window behind me though. I'm used to using a MacBook Pro 15 with a glass screen, it's a least a TAD better, but not much.

For me personally, I don't really care much about glossy screens. I actually like them more because the content looks better without a matte layer in the way. But this is purely opinion and everyone's use cases are different.

I would say if you can't stand MacBook type glass panels you probably will have mostly the same problem with the Gram 17 unless you add a matte layer.

 

Thanks for the feedback, I actually had the LG gram 15 and absolutely hated the reflections, I also work often outside.

 

Yeah, if you work outside a lot you should either get a notebook with a matte panel or get one of those glare reduction layers.

 

I tried outdoor use with my anti-glare thing and I find it at the border of being usable (at noon, cloudy sky). I wouldn't want to use it outside, the screen is not bright enough to work comfortably. I would immediately look for a shady spot.

 

Today morning I have received following email with a request to post a link here. Strange is a have zero problem to use stock Fedora 29 & 30 on my gram 17.

>>>
Hey Nalim,

I don't have an account on the Notebook Review forum; I tried creating one earlier but someone has to approve my account before I can post anything. It's the only site I've seen the issue being actively discussed, so I wanted to share a workaround I have:
https://github.com/dhedlund/kernel-patch-lg-gram-17

If you would be so kind, could you share that link in the LG Gram 17 forum thread to give other people on the thread, or via google searches, hope that there's a solution.

Now that I've been able to narrow down the issue to a specific commit, I'm going work on a proper patch this weekend. Because of how long it takes Linux distros to adopt the latest kernel changes and how long it takes install images to also get updated, this issue is going to be seen by people for a while.

Cheers,

Daniel Hedlund
danielATdigitreeDOTorg
<<<

 

Wow, awesome! Thanks for this.

 

Interestingly, I was able to boot into Pop! OS from the USB after first booting into Deepin then rebooting and booting off the USB. It seems like something about booting into Win 10 first breaks ACPI in Linux. I have a feeling something is being changed in the firmware on boot.

Edit: Apparently this is not the case. But it seems to work somewhat randomly. One time I'll see the ACPI errors, the next everything seems fine. Might be some race condition.

 

TL;DR: ACPI error workaround as of May 13th, add `pci=nommconf`.

Small update on the endless ACPI errors. The issue appears to be related to trying to increase the io size allocated to a specific pci hotplug device on the thunderbolt 3 bridge. What I'm seeing is consistent with the results of the git bisect I did earlier that added a call to reallocate unassigned bridge resources; a change introduced in that commit tries to optimize the layout of memory and io across a bridge, and allocates more resources to hotplug buses that would otherwise go to waste.

When the above is performed in combination with MMCONFIG/MCFG (memory-mapped config space/table provided by motherboard (firmware, bios?), some kind of corruption occurs that immediately begins spamming the ACPI interrupt handler with junk events; this may be due to memory corruption/offset errors, but I'm not sure yet. They don't appear to be valid events; they seem to be randomized and each event has flags set on them that should make it impossible to reach the code paths that are generating the errors due to guards on the dispatch side. ACPI errors have their event's dispatch type set to none, which should never happen...it's not the GPE XX that's the problem, they're just bogus.

I'm still trying to narrow the issue further to see if it's a kernel bug or on Intel/LGs side. Unfortunately, I won't have more time for several days. Most likely the issue is with the MMCONFIG table, or an assumption the kernel is making that it can grow into specific memory addresses that are actually reserved.

Please let know if this works, or if there are any other issues being seen as a result.

 

Thanks a lot for this work Daniel! Much appreciated. I spent a bit of time digging through this code after you noted the commit that caused the problems, but you've clearly understood the issue better than I.

I blew away my Linux partitions yesterday in anticipation for WSL2 solving the reasons I wanted to install Linux in the first place, but I'll give this a shot this week and see if it works around the problem.

I still don't QUITE understand why it seemed to be random whether I could boot successfully or not, I'm guessing ether something in the BIOS was being changed on successive boots into different OS's, or it's simply some asynchronous order of operations causing an intermittent race condition. Perhaps the amount of time it takes for a driver to init TB3?

 

Youre doing the lord's work. As far as I can tell, everything works and is stable with pci=nommconf.

I appreciate any further insight you can give into this, if eg its on linux's side or lg/intel, as you mention (although I think linux acpi considers anything a bug which doesnt work if it works with windows). I'll keep checking back in case the kernel param is not needed in the future.

 

I can vouch for

Code:

pci=nommconf

working fine for Manjaro Deepin edition.

Thanks!

 

Hello,

I just bought a withe LG Gram 17 from Korea (ebay.co.kr).
Is is a smaller version, i3, 8GB RAM, 128 GB, no Thunderbold. 1.380 USD. In comparions ...
maybee a little bit expensive. But so far a very nice notebook for me. I have installed Windows Pro.
I was not able to use "LG Update Center for LG Notebook.(Ver.1.0.1810.1701)" from lg.com.
But there was a "DNA" partition with software, drivers, to install everything.

I am a Web Developer and use Docker. I've been working with Ubuntu 18.04 recently.
I would like to install 18.04 or maybee 19.04 with fractional scaling the next days.
(my upgrade attempts to 19.04 were not very successfull on my old Notebook ...)

I had read that Lg products are better supported with new linux kernels (4.2-5).
And found this list: https://github.com/torvalds/linux/blob/master/Documentation/laptops/lg-laptop.rst
Reader Mode and Silent Mode is supported with new linux kernel? I like these
Modes very mutch.

Best Regards from Germany
Simon

 

19.04 works fine here so far. fractional scaling dont work well with firefox, phpstorm, keepassxc ... so I step back to bigger font sizes and zooms for the browsers. I booted a few times with no problems. (I do not have thunderbold and other thing migth be different). best regards

 

I don't have the LG_LAPTOP driver loaded, so I'm missing those features (panel brightness and the few critical ones work fine). I might compile a new kernel just to see if I can get the touchpad switch working, I usually have it turned off.

 

I think I also do not have more than Panel Brightness, Speaker Loudness via FN Keys.
But I think my fan_mode works. Manually writing 1 to /sys/devices/platform/lg-laptop/fan_mode.

 

Despite the performance and thermal issues mentioned in the thread, would you still recommend this laptop?
The screen size / resolution and small frame makes it my ideal laptop with no practical alternative at the moment.
Unfortunately LG does not sell laptops in my country so I have to import it. I wonder if it's worth the hassle given the mentioned performance and thermal issues. I do not need top performance for my usage but in case of support issues it would be very unpractical to return or get serviced.

 

It is very very silent and has a top screen. With hdpi, fonts looks mutch better than on full hd, its a big win when you work with a lot of text.
I would not recommend the laptop for outside. Inside its good, and with lower light great. I do not have any performance or thermal issues with i3. Programming work,
(my java IDE is demanding - here I can sometimes hear the fan), videos, big excel files (4-5 MB) no problem so far. I do not need more and do not have any problems so far.
If you want safty regarding returning the laptop, buy it from amazon. USA or Spain. I am happy with this laptop.

 

Without hesitation. It's the best laptop I've owned. Perfect union of screen size and weight. Decent performance. Good battery life. With an eGPU it can game like a boss. I really like it.

 

Got mine from Amazon in the post today, they sent it outside the US. Box came pretty bashed up, just a piece of crumpled paper inside to absorb knocks, the LG box inside the Amazon box was good though despite sliding all around the Amazon box and the laptop was up and running perfectly in no time.

 

Yes, If you want to go for 17'' and truly portable, there is no alternative. Performance is ok even with the thermal issues, I'm just OCDing about it.

I did try some more stuff in the meantime to improve thermal performance, but still haven't found the silver bullet. I will make another mod post soonish. The thing I'm concerned about with this machine is hitting 90 degree celsius quickly all the time. I know the CPU should be able to handle it, but I'm still worried about shortening its lifetime.

 

I will test Google Stadia when it is available

 

I'm in Europe and the only European Amazon store that sells this model at the moment is the Spanish Amazon store. I don't have a creditcard so that's an extra hoop to jump through, that's why I asked if it would be worth it.

The price is pretty good at the moment on the Spanish amazon store (€1530 including tax, excluding shipping) so I'll look into getting one this week on a borrowed card.
Thanks everyone for the positive replies.

 

hmm, sometimes with Ubuntu 19.04 I have the problem, that the touchpad stops working for a moment,
and one time, I have to restart the computer, because touchpad did not come back.
"/var/log/syslogs" shows many errors like this, when I grep for "Touchpad"

...
May 20 14:57:52 lpg17 /usr/lib/gdm3/gdm-x-session[1009]: (EE) event5 - 04CA00B1:00 04CA:00B1 Touchpad: libinput bug: unexpected scroll state 3
May 20 14:57:53 lpg17 /usr/lib/gdm3/gdm-x-session[1009]: message repeated 3 times: [ (EE) event5 - 04CA00B1:00 04CA:00B1 Touchpad: libinput bug: unexpected scroll state 3]
May 20 14:57:53 lpg17 /usr/lib/gdm3/gdm-x-session[1009]: (EE) event5 - 04CA00B1:00 04CA:00B1 Touchpad: libinput bug: unexpected scroll event 4 in area state
May 20 14:57:53 lpg17 /usr/lib/gdm3/gdm-x-session[1009]: (EE) event5 - 04CA00B1:00 04CA:00B1 Touchpad: libinput bug: unexpected scroll state 3
...
May 20 17:21:10 lpg17 /usr/lib/gdm3/gdm-x-session[1009]: (EE) event5 - 04CA00B1:00 04CA:00B1 Touchpad: kernel bug: Touch jump detected and discarded.
May 20 17:21:10 lpg17 /usr/lib/gdm3/gdm-x-session[1009]: See https://wayland.freedesktop.org/libinput/doc/1.12.6/touchpad-jumping-cursors.html for details
May 20 17:24:36 lpg17 /usr/lib/gdm3/gdm-x-session[1009]: (EE) event5 - 04CA00B1:00 04CA:00B1 Touchpad: kernel bug: Touch jump detected and discarded.
...

Usually, its no problem. https://wayland.freedesktop.org/libinput/doc/1.12.6/touchpad-jumping-cursors.html
said "When you encounter the warning in the log, please generate an evemu recording of your touchpad and f
ile a bug. See Reporting bugs for more details." But I do not really understand, how to submit a proper bug
with Evemu.

 

you basically do this: https://wayland.freedesktop.org/libinput/doc/1.12.6/reporting-bugs.html#evemu

then submit it. The full instructions are there.

 

How long does it take to charge the laptop? Does it have a quick charge feature like some of the other Ultrabooks?

 

I do not believe it does. It has great battery life, but I don't believe it has any quick charge type capability. At least it isn't a noted feature from LG. I use it plugged in almost all the time and never bothered to note how fast it was charging otherwise. The included barrel plug charger is 48W, as well you can use the USB-C port to charge it. So if you wanted to carry around one of those 45W power delivery batteries that would probably work fine.

 

As far as I can tell the Gram 17 never charges with more than 32W. I measured charging over USB-C and with the included power brick and 32W was the max wattage I saw with an empty battery. As the battery is 72Wh, a full charge takes 2.5 - 3h. It would be nice if it had fast charging, but on the other hand even if you charge for just 1h it'll last you for many hours of usage, it's really a power sipper. And unlike other thin and light laptops, charging does not lead to performance issues due to heat from the charging.

So I wouldn't mind having faster charging, but so far I never ran into troubles because of the speed. The thing I wished was different was the charging connector. It should not be there, but instead there should be another usb-c.

 

A word of warning. I am still trying to get the gram 17 to run at the full performance potential with reasonable temps and did various mods and I had a big mishap. Maybe this post will prevent this for other people: At some point I started to think there is just not enough pressure from the heat sink and the way it's mounted to the cpu socket. So I started adding copper shims and using more force on those heat sink screws and that lead to a big accident.

Suddenly temps got really high and after investigating I found out one of the CPU screw sockets came loose. I thought the laptop is done for, but then I realized those things are just soldered onto the motherboard. There is no support bracket on this machine!!! This means the CPU heat sink screws cannot take a lot of force as it's the motherboard itself that's holding things together. So don't use too much force on the CPU sink screws! Luckily I was able to reattach the screw socket and resolder the connection, but it was nerve-wrecking and I had to remove the motherboard completely (which is btw. quite easy).

The lack of pressure really is a problem, but what would be needed to solve it was a custom bracket.

 

OH MAN. That's scary. Yeah I never use a lot of force fastening a heat sink, big no-no in any circumstance. You also have the rare times where doing that can chip or crack a die, super rare but I've seen it. With this weak sauce assembly in the Gram 17 it probably would never happen...

 

Here are some images of the horror, as well as a high-res shot of the bottom side of the motherboard.




Motherboard:

 

Its curious that they've removed the labeling from some of those components over by the TB3 port.

 

If anyone is interested in the Gram 17 it seems to be $150 off everywhere right now, including straight from LG. $1699 down to approx $1549.

That would basically pay for two of those PCIe 2x 500GB Western Digital Blue SSD.

 

They are labeled, but on the photo light reflections hide the information.

 

Ahh yeah, I can barely see it on the rightmost component. lol, sorry.

 

I'm wondering if the $150 off sale on these means they are going to refresh it with 10th Gen i7 U series CPU imminently.

I Was just randomly reading the manual and it says it takes 3 hrs to charge fully.

 

Last month LG dropped the price 100USD then put it back up to 1699USD. I missed that window and paid full price. Still happy tho as great laptop.

 

Two newbie questions:

1. Anyone using those keyboard covers such as Leze - Ultra Thin Keyboard Cover for LG Gram? I need something as will working in dusty environs occasionally I think. Wondering if these covers can be left on when carrying or will touch/scratch screen? The screen seems rather soft and at risk of scratching if not looked after?

2. I read the laptop can be charged using USB C port? Can I use any phone charger for this or what are the requirements? Until now I haven't touched it as I don't know.

Cheers

 

1) no keyboard cover in use here, but I have the screen protector. I see some keyboard dirt on that, but no scratches. I think the keys are too soft to scratch the screen.

2) Yes, but charging with a regular phone charger will be slow (most have 10W or less). You need an usb-c charger that can deliver 30W. I use one from a HUAWEI matebook x pro and also have an aftermarket one from LVSUN. Both of these could charge with 60W, but on the LG Gram17 they use the 30W mode.

 

OK. Thanks. So my Huawei Mate 9 phone super charger says its 4.5v/5a which I suppose is 22.5W or a bit too low for the LG Gram 17.

 

I finally found time to code the little tool I wanted to control the keyboard backlight and add a timeout to it. Thanks to TPFanControl's source code I now have c++ code to read and write the EC registers directly using the WinRing0 library. I made all this into a system tray application. Just run it as admin (don't run it without admin rights.. it'll burn cpu trying to write to the EC) and the backlight will time out after some inactivity. To get the backlight to turn on again at night, just touch the trackpad. It adds a tray icon with a "GH" icon that allows to change the backlight and exit the app.

It uses the system inactivity timer (also used for screensaver etc.) and it should pick up manual changes to the backlight. It should be light on CPU and not get in that way. I will keep it running on my system and it might improve a bit over time but I would be curious whether this works for anyone else. In the future I might put the code on GitHub, but first I'd have to do a cleanup session . Finally I can watch movies at night without having to turn off the backlight manually!



This tool will probably work on all 2018/2019 Gram Models but it is only tested with the 2019 LG Gram 17! On incompatible models, writing to these registers directly might cause big troubles.

-- Attachment removed, v1.01 is on thread page 26

 

Which version of Win10 should I now have? When I go to settings it says I'm up to date with 1809 but when I just installed Win10 on a VMWare virtual machine on the same laptop it loaded 1903?

Can anyone help me understand why please?

 

Microsoft is rolling it out iteratively, you'll get it eventually. If you want to force it you can download the update assistant from them to update it, I forget where it is exactly just google it.

 

For anyone that cares, LG is having a sale on the entire Gram line again and the Gram 17 is $200 off this time at $1499. If i would have known this was going to happen I would have waited a few months. No regrets though, for sure.

 

Some more pictures from my HW cooling modding attempts. I bought some copper sheets and chipquick low temp solder paste as well as various heatpipes. I am now using what is basically a toaster oven to solder things together. The first thing I found out was that 1mm thick heatpipes cannot be heated to 140 degree celsius. They become round again. However 2mm and thicker work just fine.

I modified the original heatpipe assembly with a copper plate soldered to my extra fins. On top I added 3 2mm thick heatpipes and on the cpu contact area a 0.3 mm shim. I baked it and got this:


To be able to install the thing I had to remove some more material from the chassis where the fins go.




Using these mods I got quite a big performance improvement. The shim unfortunately broke my motherboard cpu mounting posts as I posted earlier, but since then I found a better solution I'll cover in a future post.

However this heavily modified assembly is not enough to run the machine at its full performance. It definitely is enough to run at 15W, but I was aiming for 25W at the time and this only got me to about 24W.

Using a thermal camera I found that the heatpipe and fins only get to about 65 C even if the CPU is at 90. And the fan maybe just does not move enough air.

 

After this I did something completely different. I added a copper sheet to the underside of the laptop and a thermalpad on top of the CPU. This way heat could go directly to the copper sheet. This allowed the machine to turbo at 30W for quite some time. Then of course the bottom would get quite hot. I used a 0.5mm thick copper sheet and this added almost 100g of copper to the machine.




I removed this extra sink again because I felt the laptop gets too heavy with it. However If someone just wishes to turbo for longer, adding some copper weight on top of the CPU could be a solution. It does work and prevents the CPU from going to 90 degree almost immediately under heavy load.

 

Later I received a more powerful fan (from a DELL XPS) and started doing experiments with that. I 3d modelled and printed a tool to bend flat heatpipes (known to be almost impossible).



I constructed a heatsink assembly from scratch using another array of fins and a 3.5 mm thick heatpipe. I baked things together in the oven.



I then tried to run the dell fan and this new heatpipe assembly, but that's where I hit another roadblock. Somehow the LG Gram 17 fan controller got extremely confused by the dell fan. Maybe I broke it somehow, but the result was very odd fan behaviour where it would slow down at maximum heat. I guess the controller reads the speed of the fan and expects values in the correct range or somehow recalibrates itself with that. Or maybe the pwm signal is custom tailored for the specific fan and just doesn't work on the dell.

So I had to come up with another plan. Here's what's inside my machine now:


It's the dell fan, but it is driven by a micro arduino (adafruit trinket). The fan speed is controlled by reading the temp using a 10kohm thermistor attached to the underside of the heatpipe. The system fan port is used for 5V power only now. If anyone is interested I would be happy to share code and stuff. It was particularly tricky to get the trinket to output the correct 25kHz pwm signal for the fan.

Overall this works rather well, but it's still not enough to cool 30W - more like 28W or so. The dell fan is louder and it is too thick and cannot pull in air from the underside. So I am already building another solution at the moment using very flat dual fans. I'll post more once it is done.

What I really would like to achieve is getting fan control from the OS. Unfortunately there don't seem to be any internal USB connectors that could be rerouted and I don't want to sacrifice any of the usb ports. So what I am playing with at the moment is a bluetooth low energy arduino and I'd like to add fan control over bluetooth in the future. Then the real CPU temp could be used to speed up the fan.

I am also building another heatsink assembly where the heatpipe is a lot closer to the cpu surface. In the pictured one, everything sits on a 0.8 mm copper plate and there's also a 0.3 shim. I think this limits the heat transfer to the heatpipe a bit. You can also see my new mounting solution using a screw from the bottom. The resoldered one broke again...

 

In other news, I'll update the helper I wrote soonish. I am seeing some problems with it. At some point my keyboard stopped working! I'm probably hammering the EC with too many read requests or maybe it's a problem when the machine wakes from sleep.

 

Oh.. and I forgot to brag with my Cinebench R20 score. Using the pictured mods I achieve 1901 CB.

 

Very nice!

How did you secure the copper shims/plates and the heatsink fins to the heatpipe? Just regular solder (or is there a thermal solder?)?

Does the fan actually require ~25k Hz range for PWM? I remembering doing something similar with a Noctua, [EDIT: nevermind. We adjusted the duty cycle from 20% to 100% at 980Hz, you can guess this was a regular old AVR3028 Arduino].

Though that is a standard PC fan, not a custom notebook fan. It's probably safe to assume you already probed the fan range using a scope

EDIT2: Whoops. Found the Intel ATX spec sheet. 25kHz is correct. I'm surprised the my responded correctly.

 

I used chipquik low-temperature solder (melts at 138 C). Good heatpipes survive temperatures up to 150 C.

The fan I am currently using does seem to require 25 kHz. Different fans have different levels of pickiness when it comes to pwm speed. But if something in the audible range is used, noise is often an issue. This dell fan just can't process low freq pwm signals at all and turns at random speeds. I have some other fans that work with default arduino pwm speeds, but a hum is audible (I guess from turning the motor off and on again).

Luckily the ATTiny85 in the adafruit trinket (as well as other higher-clocked variants) can supply precise 25 kHz pwm signals. It does require timer register programming and in this case it got rather tricky as there is only one timer with the necessary capabilities on the chip and it's normally used for system timings. Full featured arduinos could use an extra timer and then things would be easy but initially I went for the trinket to get low power consumption. I could redo the current state with just an attiny85v and then it would use almost no power for the fan control.