Dell XPS 13 9350 Late 2015

frank604 · 2015-11-20 04:44:39

@jyc, thanks for the further development of the issue. I'll try reloading snd_hda_intel and if it works... I'll add a reload script in system sleep for other to use.

edit: modprobe -rf snd_hda_intel results in fatal in use error. How'd you get it to unload?

Last edited by frank604 (2015-11-20 08:08:17)

pezzonovante · 2015-11-20 05:27:22

How's everyone's battery life and power usage under different kernels? Could you people share what the power usage is like for normal web browsing under 4.4rc1, linux-bcm4350 and linux-dell by @frank604? If you're using tlp or any other powersaving options/kernel parameters please state that too. Thanks.

EDIT: I would've shared mine, but my ordered unit hasn't shipped yet. I'm trying to choose in advance which kernel to install when it does arrive

Last edited by pezzonovante (2015-11-20 05:50:13)

bart_vv · 2015-11-20 16:51:03

@condensator, I also encountered that problem with touchscreen and sleeping. Didn't investigate it yet (I completely don't need a touchscreen, actually would buy a FHD variant if there was one with 16GB RAM), maybe I will in my spare time. Do you use libinput? Do you have TLP/usb autosuspend enabled?

mkey · 2015-11-21 11:18:54

I've got the i5 FHD 8Go variant and installed linux-mainline from AUR.
Full system freezes disappeared and i915 preliminary support kernel parameter could safely be removed.

I can confirm sound issue when plugging headphones.
Same behavior as reported by franck604: white hissing noise which disappears when setting up "Headphone Mic Boost" to 22.
I also tried to plug a mic/headphones combo adapter in order to have a proper output/input device plugged in: same behavior.

Otherwise, everything seems to work just fine. BTW I could not verify 3.1 USB port as I don't have such a device to test with.

Concerning power consumption, 4.4-rc1 with pcie_aspm=force i915.enable_rc6=7 parameters and tlp running (default configuration) leads me between a 5W and 10W discharge rate.
Average is 7W (browsing + music playing and screen brightness set to 1/3) which corresponds to approximately 8 hours battery life.

pezzonovante · 2015-11-21 13:55:04

@mkey: thank you very much for the detailed post. It's very helpful.

One question though, are you using the 256GB SSD or 128GB SSD version? I ask that, because the 256 GB and higher versions have the nvme SSD, while the 128 GB version doesn't. That could make a difference when it comes to power consumption.

condensator · 2015-11-21 17:18:54

bart_vv wrote:

@condensator, I also encountered that problem with touchscreen and sleeping. Didn't investigate it yet (I completely don't need a touchscreen, actually would buy a FHD variant if there was one with 16GB RAM), maybe I will in my spare time. Do you use libinput? Do you have TLP/usb autosuspend enabled?

Same here, don't really need it but would of course be nice for it to work
I use libinput. I have tried both with and without tlp. I haven't configured anything manually with regard to usb autosuspend.
I also realized that sometimes it does work after suspend -- but usually not.

mkey · 2015-11-21 21:14:22

pezzonovante wrote:

One question though, are you using the 256GB SSD or 128GB SSD version? I ask that, because the 256 GB and higher versions have the nvme SSD, while the 128 GB version doesn't. That could make a difference when it comes to power consumption.

It's the 256Go version that comes with NVMe controller.

frank604 · 2015-11-21 23:29:43

Guys, if you know how to patch, please help patch "patch_realtek.c" with different modes and help find the fix for our sound issue. It is possible, just takes a bit of time. I've already gone through a few hours of compiling with no luck but it'd go faster if we could setup a public document and cross off the ones we've checked for hour headset mode.

Edit: refer to the bugzilla bug report for more info

Last edited by frank604 (2015-11-21 23:30:08)

vilavil · 2015-11-23 00:03:40

Hi guys,
I've been following this conversation for a while and just compiled and installed the 4.4-rc1 kernel. I had to add nvme to /etc/modules to be able to boot and added the firmware from the archwiki for the wifi to work. Things are finally starting to work out!
I haven't played much with the powersaving settings but it seems like the general trend is 6-7h under Linux and 9-10h under Win10.
The funny thing is that I ran a comparison of power consumption using Intel RAPL domain and it seems like CPU + GPU is actually draining less power under Linux than Win10 (!) even though overall power consumption is quite a bit higher. Could it mean that the SSD is at fault?
Small writeup of the powerstat figures here : http://pastebin.com/44gbqcaQ

frank604 · 2015-11-23 03:46:58

vilavil wrote:

Hi guys,
I've been following this conversation for a while and just compiled and installed the 4.4-rc1 kernel. I had to add nvme to /etc/modules to be able to boot and added the firmware from the archwiki for the wifi to work. Things are finally starting to work out!
I haven't played much with the powersaving settings but it seems like the general trend is 6-7h under Linux and 9-10h under Win10.
The funny thing is that I ran a comparison of power consumption using Intel RAPL domain and it seems like CPU + GPU is actually draining less power under Linux than Win10 (!) even though overall power consumption is quite a bit higher. Could it mean that the SSD is at fault?
Small writeup of the powerstat figures here : http://pastebin.com/44gbqcaQ

Someone sent me this link and a quote from the first page:
It appears that Samsung's NVMe drives have much higher idle power consumption than the AHCI drives, even when using the same UBX controller. It's clear that our system configuration is not putting the 950 Pro or the SM951 with NVMe into a low power state when idle, but the cause for what that happens is not clear.

If this is true then during idle times our nvme pcie ssds aren't powering down and it could be worse in linux than windows which would explain where that extra 1W goes. This is all a guess. NVME is still quite new ...

vilavil · 2015-11-23 15:56:34

Interesting, that does seem to be a likely option and a good place to look, although 1W still would not fully explain the ~3W difference with the 9343 XPS 13. It would be interesting to find a person who owns a 128GB version with a non-NVMe controller to see where we stand.

After trying some of the tips for the 9343, the options "pcie_aspm=force i915.enable_fbc=1 i915.enable_rc6=7" or "i915.enable_fbc=1" had no noticeable effect on the power consumption. Installing tlp decreased idle power consumption by about 0.7W to 6.2W. That gain seems to be attributed to the PCIe ASPM powersave option, which when disabled gives me the previous power consumption, which is odd because the boot option was supposed to activate it.

Anyone else getting these errors at startup? No idea if they could be partly responsible :

[    1.485959] i915 0000:00:02.0: Direct firmware load for i915/skl_guc_ver4.bin failed with error -2
[    1.485989] [drm:intel_guc_ucode_init [i915]] *ERROR* Failed to fetch GuC firmware from i915/skl_guc_ver4.bin (error -2)
[    1.487822] [drm:i915_gem_init_hw [i915]] *ERROR* Failed to initialize GuC, error -5 (ignored)
...
[    3.065329] ACPI Warning: \_SB_.IETM._ART: Return Package type mismatch at index 0 - found Integer, expected Reference (20150930/nspredef-297)
[    3.065337] ACPI: Invalid package element [0]: got number, expecting [R]
[    3.065339] _ART package 0 is invalid, ignored

zetxx · 2015-11-23 20:22:56

no errors here:

─ dmesg| grep -i i915
[    0.000000] Command line: initrd=\initramfs-linux.img root=/dev/nvme0n1p2 rw i915.preliminary_hw_support=1 elevator=noop pcie_aspm=force i915.enable_rc6=7 i915.enable_execlists=0 i915.enable_psr=1
[    0.000000] Kernel command line: initrd=\initramfs-linux.img root=/dev/nvme0n1p2 rw i915.preliminary_hw_support=1 elevator=noop pcie_aspm=force i915.enable_rc6=7 i915.enable_execlists=0 i915.enable_psr=1
[    0.494149] [drm] Initialized i915 1.6.0 20150522 for 0000:00:02.0 on minor 0
[    1.565683] i915 0000:00:02.0: fb0: inteldrmfb frame buffer device
[    1.565684] i915 0000:00:02.0: registered panic notifier
[    2.211821] snd_hda_intel 0000:00:1f.3: bound 0000:00:02.0 (ops i915_exit [i915])

Last edited by zetxx (2015-11-28 17:09:42)

jyc · 2015-11-23 22:54:13

frank604 wrote:

@jyc, thanks for the further development of the issue. I'll try reloading snd_hda_intel and if it works... I'll add a reload script in system sleep for other to use.
edit: modprobe -rf snd_hda_intel results in fatal in use error. How'd you get it to unload?

Hm, it seems to be on/off. I got it to work once, but I just tried again and wasn't able to.

On another note, does anyone have weird beeping and buzzing noises coming from the motherboard area? I asked Dell and they claimed they are looking for a fix, but I was wondering if this was uncommon and I should return it.

frank604 · 2015-11-24 00:48:54

jyc wrote:

On another note, does anyone have weird beeping and buzzing noises coming from the motherboard area? I asked Dell and they claimed they are looking for a fix, but I was wondering if this was uncommon and I should return it.

Did you disable pc speaker?

jyc · 2015-11-24 00:53:46

frank604 wrote:

Did you disable pc speaker?

Hi Frank, thanks for the tip. Would I disable the PC speaker using the pcspkr module? Would doing that help, or am I supposed to not disable it?

mkey · 2015-11-24 06:46:53

jyc wrote:

On another note, does anyone have weird beeping and buzzing noises coming from the motherboard area? I asked Dell and they claimed they are looking for a fix, but I was wondering if this was uncommon and I should return it.

Same weird noise here. Did Dell tell you about the origin of this issue? It's sounds like electrical interference.

Last edited by mkey (2015-11-24 06:57:35)

ksim · 2015-11-24 07:15:46

jyc wrote:

On another note, does anyone have weird beeping and buzzing noises coming from the motherboard area? I asked Dell and they claimed they are looking for a fix, but I was wondering if this was uncommon and I should return it.

The same noise, i've read complains about it for previous generations also. looks like a common problem, it depends on cpu power consumption and IDLE state.

May be also depends on the model, I have 16Gb RAM variant.

Last edited by ksim (2015-11-24 07:16:37)

frank604 · 2015-11-24 10:28:26

Update: whitenoise bug fixed thanks to kernel devs. Patch will be included in 4.4-rc3 if all goes to plan. Until then you can patch it (attached to bug tracker). pkgbuild updated to include this fix on my github built for 4.4-rc1

zetxx · 2015-11-24 11:13:39

jyc wrote:

On another note, does anyone have weird beeping and buzzing noises coming from the motherboard area? I asked Dell and they claimed they are looking for a fix, but I was wondering if this was uncommon and I should return it.

Same noise here

SheepOnMeth · 2015-11-24 13:08:20

I have also the same noise.

Thanks frank, I am compiling the kernel with your PKGBUILD, I just added -j$(nproc) to the make line to use all the cores for compiling.

I have been using 4.4rc1 for a few days now, it seems that the pstate driver does not allow the change of max frequency. Most of the time the 4 cores are at 500MHz. It is a little to slow for my taste, when chromium render web pages it needs more.

I have been looking at the NVMe specification on http://www.nvmexpress.org/specifications/ . In the last specification update there is something about power level. Also I found this tool
https://github.com/linux-nvme/nvme-cli for getting information and sending command parameter to the NVMe controler. I did not find the magic command line to reduce the power consumption. The next step is looking into the NVMe driver itself.

Last edited by SheepOnMeth (2015-11-24 17:24:20)

zetxx · 2015-11-24 21:24:23

in case if someone needs windows key sudo xxd /sys/firmware/acpi/tables/MSDM

condensator · 2015-11-24 22:32:34

linux-mainline 4.4-rc2 has been working fine for me for some time now.

Last edited by condensator (2015-11-24 22:32:54)

Utini · 2015-11-25 14:34:32

Does anyone have "kernel parameters" for power saving which would also work on my Dell XPS 15 9530 ?

Thanks !

vilavil · 2015-11-25 17:23:17

The GuC errors in my previous reply

[    1.485959] i915 0000:00:02.0: Direct firmware load for i915/skl_guc_ver4.bin failed with error -2
[    1.485989] [drm:intel_guc_ucode_init [i915]] *ERROR* Failed to fetch GuC firmware from i915/skl_guc_ver4.bin (error -2)
[    1.487822] [drm:i915_gem_init_hw [i915]] *ERROR* Failed to initialize GuC, error -5 (ignored)

were due to the fact that I loaded nvme as a module instead of compiling it in the kernel. This caused i915 to be loaded in initramfs before the disk was mounted, hence the error that the firmware was not found. The right way to fix this is to recompile the kernel with nvme compiled directly into it with the line CONFIG_BLK_DEV_NVME=y. Now the 4.4-rc1 and 4.4-rc2 both work nicely. To be able to compile them I had to remove the following snippet

-	case IOC_LIBCFS_PING_TEST: {
-		extern void (kping_client)(struct libcfs_ioctl_data *);
-		void (*ping)(struct libcfs_ioctl_data *);
-
-		CDEBUG(D_IOCTL, "doing %d pings to nid %s (%s)\n",
-		       data->ioc_count, libcfs_nid2str(data->ioc_nid),
-		       libcfs_nid2str(data->ioc_nid));
-		ping = symbol_get(kping_client);
-		if (!ping)
-			CERROR("symbol_get failed\n");
-		else {
-			ping(data);
-			symbol_put(kping_client);
-		}
-		return 0;
-	}
-

from the file drivers/staging/lustre/lustre/libcfs/module.c in the kernel directory.

I will also have a look at the nvme driver to see what could be done to enable power saving mode.

frank604 · 2015-11-25 20:00:58

vilavil wrote:

The GuC errors in my previous reply
[    1.485959] i915 0000:00:02.0: Direct firmware load for i915/skl_guc_ver4.bin failed with error -2
[    1.485989] [drm:intel_guc_ucode_init [i915]] *ERROR* Failed to fetch GuC firmware from i915/skl_guc_ver4.bin (error -2)
[    1.487822] [drm:i915_gem_init_hw [i915]] *ERROR* Failed to initialize GuC, error -5 (ignored)
were due to the fact that I loaded nvme as a module instead of compiling it in the kernel. This caused i915 to be loaded in initramfs before the disk was mounted, hence the error that the firmware was not found. The right way to fix this is to recompile the kernel with nvme compiled directly into it with the line CONFIG_BLK_DEV_NVME=y. Now the 4.4-rc1 and 4.4-rc2 both work nicely. To be able to compile them I had to remove the following snippet
-	case IOC_LIBCFS_PING_TEST: {
-		extern void (kping_client)(struct libcfs_ioctl_data *);
-		void (*ping)(struct libcfs_ioctl_data *);
-
-		CDEBUG(D_IOCTL, "doing %d pings to nid %s (%s)\n",
-		       data->ioc_count, libcfs_nid2str(data->ioc_nid),
-		       libcfs_nid2str(data->ioc_nid));
-		ping = symbol_get(kping_client);
-		if (!ping)
-			CERROR("symbol_get failed\n");
-		else {
-			ping(data);
-			symbol_put(kping_client);
-		}
-		return 0;
-	}
-
from the file drivers/staging/lustre/lustre/libcfs/module.c in the kernel directory.
I will also have a look at the nvme driver to see what could be done to enable power saving mode.

linux-mainline from aur has this patch. For those that are affected with your issue. I never needed this patch though (? or do I ?). Was this causing you to not boot? Or did you have video issues?

Arch Linux

#26 2015-11-20 04:44:39

Re: Dell XPS 13 9350 Late 2015

#27 2015-11-20 05:27:22

Re: Dell XPS 13 9350 Late 2015

#28 2015-11-20 16:51:03

Re: Dell XPS 13 9350 Late 2015

#29 2015-11-21 11:18:54

Re: Dell XPS 13 9350 Late 2015

#30 2015-11-21 13:55:04

Re: Dell XPS 13 9350 Late 2015

#31 2015-11-21 17:18:54

Re: Dell XPS 13 9350 Late 2015

#32 2015-11-21 21:14:22

Re: Dell XPS 13 9350 Late 2015

#33 2015-11-21 23:29:43

Re: Dell XPS 13 9350 Late 2015

#34 2015-11-23 00:03:40

Re: Dell XPS 13 9350 Late 2015

#35 2015-11-23 03:46:58

Re: Dell XPS 13 9350 Late 2015

#36 2015-11-23 15:56:34

Re: Dell XPS 13 9350 Late 2015

#37 2015-11-23 20:22:56

Re: Dell XPS 13 9350 Late 2015

#38 2015-11-23 22:54:13

Re: Dell XPS 13 9350 Late 2015

#39 2015-11-24 00:48:54

Re: Dell XPS 13 9350 Late 2015

#40 2015-11-24 00:53:46

Re: Dell XPS 13 9350 Late 2015

#41 2015-11-24 06:46:53

Re: Dell XPS 13 9350 Late 2015

#42 2015-11-24 07:15:46

Re: Dell XPS 13 9350 Late 2015

#43 2015-11-24 10:28:26

Re: Dell XPS 13 9350 Late 2015

#44 2015-11-24 11:13:39

Re: Dell XPS 13 9350 Late 2015

#45 2015-11-24 13:08:20

Re: Dell XPS 13 9350 Late 2015

#46 2015-11-24 21:24:23

Re: Dell XPS 13 9350 Late 2015

#47 2015-11-24 22:32:34

Re: Dell XPS 13 9350 Late 2015

#48 2015-11-25 14:34:32

Re: Dell XPS 13 9350 Late 2015

#49 2015-11-25 17:23:17

Re: Dell XPS 13 9350 Late 2015

#50 2015-11-25 20:00:58

Re: Dell XPS 13 9350 Late 2015

Board footer