Dell XPS 13 9350 Late 2015

robsmith11 · 2016-11-24 15:55:17

robsmith11 wrote:

Is anyone able to charge a phone via a USB 2 cable using their 9350?
I've tried two android phones and while both show the charging indicator, they continue to discharge at a normal rate.
The only power saving tweaks I've done are what was suggested by powertop. I disabled the USB auto suspend tweaks in case they were preventing charging, but that had no effect.

For anyone else experiencing charging issues: upgrading my BIOS from 1.4.4 to 1.4.10 seems to have fixed it. My android phone now charges fine using USB.

Last edited by robsmith11 (2016-11-24 15:55:39)

kubrick · 2016-11-25 16:30:04

Hi,

I have done many tests with the nvme patch and for me it doesn't make any measurable difference, I go down to 6.7W with the lowest screen brightness with the standard arch kernel or the nvme patch. Am I missing something?

smartctl -a /dev/nvme0
smartctl 6.5 2016-05-07 r4318 [x86_64-linux-4.8.10-1-ARCH] (local build)
Copyright (C) 2002-16, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Model Number:                       THNSN5512GPU7 NVMe TOSHIBA 512GB
Serial Number:                      66AS10UQT5ZV
Firmware Version:                   57DA4103
PCI Vendor/Subsystem ID:            0x1179
IEEE OUI Identifier:                0x0d0800
Controller ID:                      0
Number of Namespaces:               1
Namespace 1 Size/Capacity:          512,110,190,592 [512 GB]
Namespace 1 Formatted LBA Size:     512
Local Time is:                      Fri Nov 25 08:14:15 2016 PST
Firmware Updates (0x02):            1 Slot
Optional Admin Commands (0x0016):   Format Frmw_DL *Other*
Optional NVM Commands (0x001e):     Wr_Unc DS_Mngmt Wr_Zero Sav/Sel_Feat
Warning  Comp. Temp. Threshold:     78 Celsius
Critical Comp. Temp. Threshold:     82 Celsius

Supported Power States
St Op     Max   Active     Idle   RL RT WL WT  Ent_Lat  Ex_Lat
 0 +     6.00W       -        -    0  0  0  0        0       0
 1 +     2.40W       -        -    1  1  1  1        0       0
 2 +     1.90W       -        -    2  2  2  2        0       0
 3 -   0.0120W       -        -    3  3  3  3     5000   10000
 4 -   0.0060W       -        -    4  4  4  4   100000   50000

Supported LBA Sizes (NSID 0x1)
Id Fmt  Data  Metadt  Rel_Perf
 0 +     512       0         2
 1 -    4096       0         1
...

F.

robsmith11 · 2016-11-26 05:41:24

kubrick wrote:

I have done many tests with the nvme patch and for me it doesn't make any measurable difference, I go down to 6.7W with the lowest screen brightness with the standard arch kernel or the nvme patch. Am I missing something?

Just a guess, but maybe the power savings are dependent on the specific SSD model? Using the 256GB Samsung in my FHD, I definitely saw a 0.8-1.0W improvement with the patches.

smartctl 6.5 2016-05-07 r4318 [x86_64-linux-4.8.0-rc6-nvme] (local build)
Copyright (C) 2002-16, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Model Number:                       PM951 NVMe SAMSUNG 256GB
Serial Number:                      S29NNXAGB02546
Firmware Version:                   BXV77D0Q
PCI Vendor/Subsystem ID:            0x144d
IEEE OUI Identifier:                0x002538
Controller ID:                      1
Number of Namespaces:               1
Namespace 1 Size/Capacity:          256,060,514,304 [256 GB]
Namespace 1 Utilization:            21,009,117,184 [21.0 GB]
Namespace 1 Formatted LBA Size:     512
Local Time is:                      Sat Nov 26 05:34:54 2016 UTC
Firmware Updates (0x06):            3 Slots
Optional Admin Commands (0x0017):   Security Format Frmw_DL *Other*
Optional NVM Commands (0x001f):     Comp Wr_Unc DS_Mngmt Wr_Zero Sav/Sel_Feat
Maximum Data Transfer Size:         32 Pages

Supported Power States
St Op     Max   Active     Idle   RL RT WL WT  Ent_Lat  Ex_Lat
 0 +     6.00W       -        -    0  0  0  0        5       5
 1 +     4.20W       -        -    1  1  1  1       30      30
 2 +     3.10W       -        -    2  2  2  2      100     100
 3 -   0.0700W       -        -    3  3  3  3      500    5000
 4 -   0.0050W       -        -    4  4  4  4     2000   22000

Supported LBA Sizes (NSID 0x1)
Id Fmt  Data  Metadt  Rel_Perf
 0 +     512       0         0

=== START OF SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED

SMART/Health Information (NVMe Log 0x02, NSID 0xffffffff)
Critical Warning:                   0x00
Temperature:                        34 Celsius
Available Spare:                    100%
Available Spare Threshold:          50%
Percentage Used:                    0%
Data Units Read:                    1,499,882 [767 GB]
Data Units Written:                 2,145,117 [1.09 TB]
Host Read Commands:                 27,811,257
Host Write Commands:                28,466,677
Controller Busy Time:               383
Power Cycles:                       1,034
Power On Hours:                     824
Unsafe Shutdowns:                   154
Media and Data Integrity Errors:    0
Error Information Log Entries:      80

Error Information (NVMe Log 0x01, max 64 entries)
Num   ErrCount  SQId   CmdId  Status  PELoc          LBA  NSID    VS
  0         80     0  0x0011  0x4016  0x000            0     1     -
  1         79     0  0x0013  0x4016  0x000            0     0     -
  2         78     1  0x0061  0x4016  0x000            0     0     -
  3         77     0  0x0011  0x4016  0x000            0     0     -
  4         76     0  0x0011  0x4016  0x000            0     1     -
  5         75     0  0x0004  0x4016  0x000            0     1     -
  6         74     0  0x0004  0x4016  0x000            0     1     -
  7         73     0  0x0004  0x4016  0x000            0     1     -
  8         72     0  0x0004  0x4016  0x000            0     1     -
  9         71     1  0x0007  0x4016  0x000            0     0     -
 10         70     0  0x0004  0x4016  0x000            0     1     -
 11         69     0  0x0004  0x4016  0x000            0     1     -
 12         68     0  0x0004  0x4016  0x000            0     1     -
 13         67     0  0x0004  0x4016  0x000            0     1     -
 14         66     0  0x0004  0x4016  0x000            0     1     -
 15         65     0  0x0004  0x4016  0x000            0     1     -
... (48 entries not shown)

kang · 2016-11-26 06:45:19

i7/iris/qhd w/ Samsung 512gb here, i get just at 6w w/ wifi on, keyboard backlight off, screen backlight low, 4k. switching to 1080 it goes to 5.5 or so
without the nvme patch its about 0.7w more, consistently.

all measurements done when idle

Leofree · 2016-11-26 15:58:53

I can't boot arch using UEFI mode. It seems that the bios cannot pass parameters to vmlinuz-linux.
Puzzled

kgizdov wrote:

There's a new BIOS - 1.4.10

kgizdov · 2016-11-26 19:57:26

@Leofree, double check your config and kernels. I can boot in UEFI fine.

damige · 2016-11-27 01:41:50

just to let you guys know, i have made a little tool that locks the fanspeed:
https://github.com/damige/xpsfan

Its possible to run the laptop in a silent fanless mode.
If this tool is handy i might put it on AUR.

Please read the readme carefully, i'm not responsible for any damages because of this, use at your own risk.

OdinEidolon · 2016-11-27 13:19:28

damige wrote:

just to let you guys know, i have made a little tool that locks the fanspeed:
https://github.com/damige/xpsfan
Its possible to run the laptop in a silent fanless mode.
If this tool is handy i might put it on AUR.
Please read the readme carefully, i'm not responsible for any damages because of this, use at your own risk.

Is this supposed to work in the XPS15 too? I'd love an AUR package

Thanks!

damige · 2016-11-27 15:09:41

OdinEidolon wrote:

damige wrote:
just to let you guys know, i have made a little tool that locks the fanspeed:
https://github.com/damige/xpsfan
Its possible to run the laptop in a silent fanless mode.
If this tool is handy i might put it on AUR.
Please read the readme carefully, i'm not responsible for any damages because of this, use at your own risk.
Is this supposed to work in the XPS15 too? I'd love an AUR package
Thanks!

the xps_fan-lock / xps_fan-unlock should work on the 15 aswell. (maybe even more dell laptops) the silent mode is tuned for the XPS13 thermal properties, and i have not accounted for dedicated graphics.
In any case it should work but your mileage my vary.

edit: maybe try it out and let me know

Last edited by damige (2016-11-27 15:10:04)

nos1609 · 2016-11-27 20:35:03

damige wrote:

just to let you guys know, i have made a little tool that locks the fanspeed:
https://github.com/damige/xpsfan

Just tested on the new bios, runs pretty cool with it, but have two improvement proposals for further developement.
There is a necessety in a shutdown/reboot hook which automatically sets the xps_fan-original mode, without it windows goes crazy with fan speeds.
The #2 option could be an automatic swich to xps_fan-silent option on boot. Also would be nice to add the turboboost disabling option completely, not just cutting frequenccies. Sorry if I misunderstood the technical concept, and TB is disabled already, I'm not a coder;) I think it's a great alternative to that tool for osx which also makes my xps running cooler and quieter.

By the way thanks a lot for the realization of such a cool feature! Keep on!

Last edited by nos1609 (2016-11-27 20:36:46)

Leofree · 2016-11-28 10:04:33

I have tried several methods.
I want to use EFISTUB and put vmlinuz-linux and initramfs-linux.img in the ESP. I disabled secureboot and legacy boot, and added a boot option like 'vmlinuz-linux initrd=initramfs-linux.img root=/dev/nvme0n1p2' . (/dev/nvme0n1p2 is my root partition)
But it didn't work. If I used efi boot option like 'vmlinuz-linux' without any parameters, it could boot but stuck at mounting rootfs.

I wrote the boot script to the startup.nsh, after uefishellx64_v2.efi loaded, the script run automatically, but failed with a message 'No bootable device found'.
However, If run the startup.nsh manually, It would boot successfully.

Is there something wrong with me?

kgizdov wrote:

@Leofree, double check your config and kernels. I can boot in UEFI fine.

Last edited by Leofree (2016-11-28 10:05:15)

Jiaan · 2016-11-28 17:25:13

Leofree wrote:

I have tried several methods.
I want to use EFISTUB and put vmlinuz-linux and initramfs-linux.img in the ESP.

@Leofree, check out this discussion page , you should use a bootloader like systemd-boot.

Jiaan · 2016-11-28 17:30:46

(tongue twsiter) damige damages xps fan with xps_fan.

Last edited by Jiaan (2016-11-28 17:33:37)

damige · 2016-11-28 20:41:56

nos1609 wrote:

damige wrote:
just to let you guys know, i have made a little tool that locks the fanspeed:
https://github.com/damige/xpsfan
Just tested on the new bios, runs pretty cool with it, but have two improvement proposals for further developement.
There is a necessety in a shutdown/reboot hook which automatically sets the xps_fan-original mode, without it windows goes crazy with fan speeds.
The #2 option could be an automatic swich to xps_fan-silent option on boot. Also would be nice to add the turboboost disabling option completely, not just cutting frequenccies. Sorry if I misunderstood the technical concept, and TB is disabled already, I'm not a coder;) I think it's a great alternative to that tool for osx which also makes my xps running cooler and quieter.
By the way thanks a lot for the realization of such a cool feature! Keep on!

Thanks, and you are right, this setting is persistent unless you turn the laptop completely off. A normal reboot wont change this setting.
Turbo boost is disabled in this tool when running silent mode too. If you want just a toggle to turn turbo on or off this is easy to make. (echo "1" | sudo tee /sys/devices/system/cpu/intel_pstate/no_turbo) to turn off. echo 0 to turn back on.
Though i don't recommend only limiting turbo when running silent mode. More needs to be limited to make sure it does not overheat.

I will look into improving this tool some more , i have a few things i want to test when i have more time.

Jiaan wrote:

(tongue twsiter) damige damages xps fan with xps_fan.

Oh dear, better call for warranty (and dont mention my name )

Asriel · 2016-11-28 21:23:39

I have tried many kernels for my 9350 FHD - finally get this combination. It is 4.9-rc2 with some patches up to 4.9-rc7 and strange version of i915 driver based onb some very old patches by Rodrigo Vivi. It works for me fine - psr, fbc, guc submission - everything works, almost no flicker (sometimes it happens in firefox but in this case just need to move cursor avay from the place where it is). DA200 dock in HDMI does nit work on high res but other HDMI-USB-C do.

And what is most imprtant - power usage at idle without firefox-skype-viber etc opened at low screen brightness is 2.11 whatts, with all those network power hogs up it is around 3.5-4.

I do not know how it works and all my attempts to apply the same patches to newer kernel anf get same results fail, so give it as is.

that are my grub settings

pcie_aspm=force i915.enable_psr=1 i915.semaphores=1 i915.enable_fbc=1 i915.enable_rc6=7 i915.enable_guc_loading=1 i915.enable_guc_submission=1

The only point - I use SATA ssd drive, not NVMe - 1TB sata upgrade was half a price of NVMe , so not sure that NVMe will be same effective bit the NVMe patch from Andy applied as well over 4.9-rc standard NVMe.

Here is the link to the kernel sourse (100Mb) - maybe it would help somebody maybe not. If you try to build this kernel please leave your old one as an alternative - it boots fine on my laptop but I have not tried it on NVMe ssd-s.

https://yadi.sk/d/BC-g7LR_zhvqu

kang · 2016-11-28 22:44:02

Kernel 4.8.11 re-fixes the issue with the display port disconnect and the modesetting DDX \o/ (@belette)

damige · 2016-11-29 07:35:11

I have been measuring the power usage of the gpu and i am quite amased in how much (compared to the cpu) it uses power and generates heat.

In my case im running plasma and when i scroll rapidly in a terminal window it boosts the gpu frequancy a lot(up to 950mhz in my case)
this results in the temp rising 10c within ~10 seconds. Just by scrolling in a terminal!

When locking the gpufreq to 300mhz i notice no slowdowns in normal use and batt/power consumption seems better when activly browsing, etc.

Also because of the max TDP when gaming it runs the cpu at about 1.6ghz and gpu at about 750mhz. Some games benefit by more CPU or more GPU. Pillars of Eternity for example runs 5fps more when gpu is downclocked to 600mhz. (cpu rises to about 2.0-2.2)

I have made 2 scripts that i bound to hotkeys that adjust the max frequency in steps. This allowes for easy searching for optimal gpu/cpu balance when gaming.

Edit: they are part of xps_fan. Just something im fiddling with.

Last edited by damige (2016-11-29 07:38:10)

OdinEidolon · 2016-11-29 08:39:44

Asriel wrote:

I have tried many kernels for my 9350 FHD - finally get this combination. It is 4.9-rc2 with some patches up to 4.9-rc7 and strange version of i915 driver based onb some very old patches by Rodrigo Vivi. It works for me fine - psr, fbc, guc submission - everything works, almost no flicker (sometimes it happens in firefox but in this case just need to move cursor avay from the place where it is). DA200 dock in HDMI does nit work on high res but other HDMI-USB-C do.
And what is most imprtant - power usage at idle without firefox-skype-viber etc opened at low screen brightness is 2.11 whatts, with all those network power hogs up it is around 3.5-4.
I do not know how it works and all my attempts to apply the same patches to newer kernel anf get same results fail, so give it as is.
that are my grub settings
pcie_aspm=force i915.enable_psr=1 i915.semaphores=1 i915.enable_fbc=1 i915.enable_rc6=7 i915.enable_guc_loading=1 i915.enable_guc_submission=1
The only point - I use SATA ssd drive, not NVMe - 1TB sata upgrade was half a price of NVMe , so not sure that NVMe will be same effective bit the NVMe patch from Andy applied as well over 4.9-rc standard NVMe.
Here is the link to the kernel sourse (100Mb) - maybe it would help somebody maybe not. If you try to build this kernel please leave your old one as an alternative - it boots fine on my laptop but I have not tried it on NVMe ssd-s.
https://yadi.sk/d/BC-g7LR_zhvqu

Interesting. What does GuC do and why should we enable it? Do you have a link?

belette · 2016-11-29 09:01:21

thanks kang ! will update, make test and let you know

kang wrote:

Kernel 4.8.11 re-fixes the issue with the display port disconnect and the modesetting DDX \o/ (@belette)

robsmith11 · 2016-11-29 13:45:23

@damige
In the readme, you mention the intel gpu tools won't work with the Xorg modesetting DDX driver, but it seems to be working for me.

As a test, I dropped the frequency from 1000MHz to 300MHz while running the fish tank webgl benchmark in firefox. FPS went from 40ish to low 20's and power usage went from 19W to 12W.

Last edited by robsmith11 (2016-11-29 14:20:36)

nos1609 · 2016-11-29 14:06:48

Successfully activated full touch support in firefox on 9350.

First you need to modify your .desktop shirtcut file to set the env value. Open

/usr/share/applications/firefox-nightly.desktop

find lines tagged Exec and modify to the following

env MOZ_USE_XINPUT2=1 firefox-nightly %u

. You'll have to reproduce this step each time you get your firefox upgraded

In firefox addressbar type

about:config

Set browser.gesture.pinch.in to cmd_fullZoomReduce
Set browser.gesture.pinch.out to cmd_fullZoomEnlarge
Set browser.gesture.pinch.threshold to 150
Set apz.overscroll.enabled to true

Didn't get yet why it stops working after holding the finger to open menu, but clicking, scrolling and zooming works successfully till that.

Last edited by nos1609 (2016-11-30 10:09:39)

rgzfxf6bhu · 2016-11-29 14:35:43

nos1609 wrote:

Successfully activated full touch support in firefox on 9350. Anybody needs this info posted here?

Sure! Would be nice, because I still use 'Grab and Drag'.

Last edited by rgzfxf6bhu (2016-11-29 14:36:12)

damige · 2016-11-29 16:53:18

robsmith11 wrote:

@damige
In the readme, you mention the intel gpu tools won't work with the Xorg modesetting DDX driver, but it seems to be working for me.
As a test, I dropped the frequency from 1000MHz to 300MHz while running the fish tank webgl benchmark in firefox. FPS went from 40ish to low 20's and power usage went from 19W to 12W.

Strange! I tested this a few months back and had no success. I'll edit the readme.

Keep in mind that unless you locked the cpu freq chances are that when you lowered the gpu the cpu went up in freq consuming more power. (So maybe not accurately showing the real impact)

Asriel · 2016-11-29 19:38:02

OdinEidolon wrote:

Interesting. What does GuC do and why should we enable it? Do you have a link?

Not sure completely but how I understood GuC is an intel closed source firmware that controls the load distribution between graphics cores so I believe it also do some power saving there. But it is closed source so nobody out of intel knows exactly.

I had experimented with settings and for me this kernel and this boot parameters gave the best result in battery life. And also fan is almost all the time stopped without any special fan control - it runs only when compiling something or doing video conversion. But just web-browsing, openoffice documents, mail - no fan noise at all.

OdinEidolon · 2016-11-29 19:44:11

Asriel wrote:

OdinEidolon wrote:
Interesting. What does GuC do and why should we enable it? Do you have a link?
Not sure completely but how I understood GuC is an intel closed source firmware that controls the load distribution between graphics cores so I believe it also do some power saving there. But it is closed source so nobody out of intel knows exactly.
I had experimented with settings and for me this kernel and this boot parameters gave the best result in battery life. And also fan is almost all the time stopped without any special fan control - it runs only when compiling something or doing video conversion. But just web-browsing, openoffice documents, mail - no fan noise at all.

Thanks, I will try this on my 9550.

Arch Linux

#1126 2016-11-24 15:55:17

Re: Dell XPS 13 9350 Late 2015

#1127 2016-11-25 16:30:04

Re: Dell XPS 13 9350 Late 2015

#1128 2016-11-26 05:41:24

Re: Dell XPS 13 9350 Late 2015

#1129 2016-11-26 06:45:19

Re: Dell XPS 13 9350 Late 2015

#1130 2016-11-26 15:58:53

Re: Dell XPS 13 9350 Late 2015

#1131 2016-11-26 19:57:26

Re: Dell XPS 13 9350 Late 2015

#1132 2016-11-27 01:41:50

Re: Dell XPS 13 9350 Late 2015

#1133 2016-11-27 13:19:28

Re: Dell XPS 13 9350 Late 2015

#1134 2016-11-27 15:09:41

Re: Dell XPS 13 9350 Late 2015

#1135 2016-11-27 20:35:03

Re: Dell XPS 13 9350 Late 2015

#1136 2016-11-28 10:04:33

Re: Dell XPS 13 9350 Late 2015

#1137 2016-11-28 17:25:13

Re: Dell XPS 13 9350 Late 2015

#1138 2016-11-28 17:30:46

Re: Dell XPS 13 9350 Late 2015

#1139 2016-11-28 20:41:56

Re: Dell XPS 13 9350 Late 2015

#1140 2016-11-28 21:23:39

Re: Dell XPS 13 9350 Late 2015

#1141 2016-11-28 22:44:02

Re: Dell XPS 13 9350 Late 2015

#1142 2016-11-29 07:35:11

Re: Dell XPS 13 9350 Late 2015

#1143 2016-11-29 08:39:44

Re: Dell XPS 13 9350 Late 2015

#1144 2016-11-29 09:01:21

Re: Dell XPS 13 9350 Late 2015

#1145 2016-11-29 13:45:23

Re: Dell XPS 13 9350 Late 2015

#1146 2016-11-29 14:06:48

Re: Dell XPS 13 9350 Late 2015

#1147 2016-11-29 14:35:43

Re: Dell XPS 13 9350 Late 2015

#1148 2016-11-29 16:53:18

Re: Dell XPS 13 9350 Late 2015

#1149 2016-11-29 19:38:02

Re: Dell XPS 13 9350 Late 2015

#1150 2016-11-29 19:44:11

Re: Dell XPS 13 9350 Late 2015

Board footer