Razer Blade Stealth Laptop On Linux, Various Linux Laptop Performance Metrics

[Jedibeeftrix]

you say this is the 13", but the size of the bezel in the screenshot suggests this is the older 12.5" model...

[asyn]

Michael, if I mailed you my Akitio Node eGPU and GTX 1080 TI, could you run tests to try to get that working?

[asyn]

Michael, if I mailed you my Akitio eGPU with an nVidia 1080 ti, could you run tests on it and see if you can get it to work?

[DMJC]

Hi Guys,

I have a Razerblade Stealth on my desk right here I also have the razer core. I can confirm it DOES work. However I use a script to switch between the NVIDIA desktop card and the laptop's Intel card when I game. It requires me to bring down X11 to switch. I have posted details on Razer's forum, essentially the script replaces the libGL.so symlinks when it does the cutover and you lose access to the laptop's internal Monitor/HDMI port when I switch to the external GPU. But it does allow full performance on the external card. If you choose to leave the Intel graphics running, the external GPU acts as an accelerator, but it has to transfer the data back over the type-C bus which impacts performance. Happy to answer any questions people have regarding the setup. I will add that my setup appears to have an issue where the capslock key can crash the entire laptop, (confirmed by other people as well, this can be mitigated by using a 3rd party Razer Keyboard driver from github, and rebinding capslock to shift with setxkbmap -option caps:shift_modifier [this only works until X restarts then it needs to be re-run]). Also the NVME drive randomly disconnects, (appears to be related to power management, reboot reattaches it and no data has been lost yet *fingers crossed*). My gut feeling is that the Razer BIOS software is garbage. That, or an update is missing. My current BIOS seems to be the current version though
BIOS Information:
Vendor: Razer
Version: 6.05
Release Date: 01/26/2017.
Firmware flashing can be done from Linux, but you need to download a BIOS flashing tool from the net and it's a pain in the ass to build (has a kernel dependency).
My Distro is Debian Buster. I had Jessie and upgraded to Stretch, am running Buster now, but none of those upgrades have fixed those issues. Kernel is 4.12.0-trunk-amd64 GNU/Linux
I need to re-test with Ubuntu, from memory a live usb worked quite well. Kernel 4.13 did not fix the issues on Debian either.

Overall, it's a nice laptop, has issues and can be a bit of a lemon, the build quality of the hardware is nice, if the BIOS/firmware issues get shaken out it'll be a great laptop for gaming. NVME disconnects are very random, sometimes several in a day, sometimes not for weeks. This laptop is also affected by the Intel Hyperthreading warning from Debian, so there's another thing that could be impacting it.

[chimpy]

They used to many years ago, but not anything recent, besides trying to get me to their annual 'press' event where you can test hardware at their office, but usually doesn't work for my schedule etc etc.

It sounds like the Galagos would be right up your alley; same specs as the Razer and comes pre installed with Linux, so I assume it wouldn't have any of the issues that were plaguing the Razer.

[gregzeng]

Thank you Michael for a very worthwhile review; all tested hardware can still be purchased from retailers now. Will now calibrate & test my current notebook, to compare my results as well, as you suggest at the end (pg 9). The comparison with so many operating systems is very enlightening.

Standouts were surprising imho. The two RPM-derived Linux systems were slowest to boot. "Clear" Linux is a very new & unusual choice; being a slow cloud-application operating system. Antergos is one of the most popular Arch-based Linux choices. Both Antergos & Clear scored well in boot times (pg 5). Another surprise was the average or better performance seen with Ubuntu (Gnome?). If a lighter spin of Ubuntu was selected (Peppermint, 1,286,602,752 bytes, with the latest Linux kernel), I assume it would match the Clear Os results.

Gems hidden in your reports include:
(1) "the battery use of laptops/ultrabooks on Linux to be much greater than Windows. With these basic results, the average was right in line with Windows 10 ... " (pg 3).
This is a pleasant surprise for Linux users. Windows boot times, bench-tests & power usage might be higher however, with malware-protection, indexing and other background operations normally in use.

(2) "Using Linux 4.13 over Ubuntu 17.04's stock Linux 4.10 kernel did reduce the power usage a bit." (pg4).
This suggests that using later versions of the Linux kernel may suggest greater power savings, but is it significant enough?

Much more popular than Antergos (1,942,355,968 bytes) is Manjaro, which has seven (7) "community spins", and four (4) official versions. Perhaps the best version to test for speed, power, etc is the "Manjaro-lxde-17.0.2-stable-x86_64" version, which is a "tiny" (1,283,332,096 bytes) download, which is about twice the size of Clear Os (559,573,460 bytes), which only has (slow-access) "cloud" applications.

[asyn]

Hi Guys,

I have a Razerblade Stealth on my desk right here I also have the razer core. I can confirm it DOES work. However I use a script to switch between the NVIDIA desktop card and the laptop's Intel card when I game. It requires me to bring down X11 to switch. I have posted details on Razer's forum, essentially the script replaces the libGL.so symlinks when it does the cutover and you lose access to the laptop's internal Monitor/HDMI port when I switch to the external GPU. But it does allow full performance on the external card. If you choose to leave the Intel graphics running, the external GPU acts as an accelerator, but it has to transfer the data back over the type-C bus which impacts performance.

DJMC, could you share the script that you're using to switch back and forth between internal graphics and the eGPU? I'm thinking that this may have big implications for Linux users across all different laptops and distros. I'm currently having the exact same issue on a System76 Galago Pro with Manjaro (Arch-based). Installing the eGPU nvidia drivers causes the internal Intel graphics to die, and I can't get get both displays to work simultaneously for the life of me. If I could leave the Intel Graphics running like you describe, I think I'd be in heaven.

[tamashumi]

I've seen some successful stories about Razer Blade Stealth (13" 2017 i7-7500U) with linux. I could kind of advertise such myself, kind of...

I'm running Ubuntu 16.04 (currently upgraded to kernel 4.14) on this machine since about half a year as my main system (work on it daily).
On the first glimpse everything is fine. There are two very annoying things though.

1. i915 related issues, symptoms:
Built-in screen blinks a lot if an external display connected (HDMI) it blinks even more as crazy.
Hitting caps-lock twice crashes machine permanently with a lot of glitches on the screen. Even magical REISUB doesn't help, only holding power button for 5 sec to put the laptop down.

dmseg shows:
[drm:intel_pch_fifo_underrun_irq_handler [i915]] *ERROR* PCH transcoder A FIFO underrun

I've been updating kernel many times to the most recent. I've installed i915 drivers for Kaby Lake from Intel but the issue remains.

Although installing the i915 drivers took away initframs -u warnings about missing firmware, it was:
W: Possible missing firmware /lib/firmware/i915/kbl_guc_ver9_14.bin for module i915
W: Possible missing firmware /lib/firmware/i915/bxt_guc_ver8_7.bin for module i915

So this is gone at least but the issues described above I was able to overcome only with a workaround which I don't feel to be a proper solution.
The workaround is adding a following boot kernel flag: i915.enable_rc6=0

2. AHCI related issues
If only shutdown and boot up cycles are used, this doesn't happen often or not at all.
However on reboot laptop quite often foes sleep-like instead of restarting. Sometimes it happens still on going down, sometime on booting up (this includes starts from cold boot if it was sleeping in the previous session). Sleep-like means that the power led on the case flashes gently as it was in sleep mode but nothing can be done to wake it or shut down further apart from holding the power button for a 5 sec to give it power cut off.

Haven't found any workaround for this one. There are is something in the dmesg which might be related:
[ 4.074568] tpm_crb MSFT0101:00: [Firmware Bug]: ACPI region does not cover the entire command/response buffer. [mem 0xfed40000-0xfed4087f flags 0x201] vs fed40080 f80

The fact Razer doesn't seem to be releasing any BIOS updates doesn't help at all. Can't wait till this official linux support. It's been ages and soon I may just need to falling back for some XPS or something.

If anyone could confirm having similar issues or even better how to solve them it would be great.
Cheers!

[tamashumi]

I've seen some successful stories about Razer Blade Stealth (13" 2017 i7-7500U) with linux. I could kind of advertise such myself, kind of...

I'm running Ubuntu 16.04 (currently upgraded to kernel 4.14) on this machine since about half a year as my main system (work on it daily).
On the first glimpse everything is fine. There are two very annoying things though.

1. i915 related issues, symptoms:
Built-in screen blinks a lot if an external display connected (HDMI) it blinks even more as crazy.
Hitting caps-lock twice crashes machine permanently with a lot of glitches on the screen. Even magical REISUB doesn't help, only holding power button for 5 sec to put the laptop down.

dmseg shows:
[drm:intel_pch_fifo_underrun_irq_handler [i915]] *ERROR* PCH transcoder A FIFO underrun

I've been updating kernel many times to the most recent. I've installed i915 drivers for Kaby Lake from Intel but the issue remains.

Although installing the i915 drivers took away initframs -u warnings about missing firmware, it was:
W: Possible missing firmware /lib/firmware/i915/kbl_guc_ver9_14.bin for module i915
W: Possible missing firmware /lib/firmware/i915/bxt_guc_ver8_7.bin for module i915

So this is gone at least but the issues described above I was able to overcome only with a workaround which I don't feel to be a proper solution.
The workaround is adding a following boot kernel flag: i915.enable_rc6=0

2. AHCI related issues
If only shutdown and boot up cycles are used, this doesn't happen often or not at all.
However on reboot laptop quite often foes sleep-like instead of restarting. Sometimes it happens still on going down, sometime on booting up (this includes starts from cold boot if it was sleeping in the previous session). Sleep-like means that the power led on the case flashes gently as it was in sleep mode but nothing can be done to wake it or shut down further apart from holding the power button for a 5 sec to give it power cut off.

Haven't found any workaround for this one. There are is something in the dmesg which might be related:
[ 4.074568] tpm_crb MSFT0101:00: [Firmware Bug]: ACPI region does not cover the entire command/response buffer. [mem 0xfed40000-0xfed4087f flags 0x201] vs fed40080 f80

The fact Razer doesn't seem to be releasing any BIOS updates doesn't help at all. Can't wait till this official linux support. It's been ages and soon I may just need to falling back for some XPS or something.

If anyone could confirm having similar issues or even better how to solve them it would be great.
Cheers!

[JohnnyConservative]

I did not see the thunderbolt 3 port tested or discussed. How may PCIe Lanes? Is it functional underLinux (not functional / partially / fully)?

I am always disappointed by reviews that do not test the Thunderbolt 3 ports.