Macbook Pro Late 2013 Retina

chutsu · 2014-04-28 20:45:55

Ok I got a problem with Flash, I tried installing flashplugin using pacman. But its very unstable, it often crashes at the start of a youtube video. Has anyone here had similar experiences?

I've got Model 11,2. Using pulseaudio.

hrod · 2014-04-30 01:48:29

chutsu wrote:

Ok I got a problem with Flash, I tried installing flashplugin using pacman. But its very unstable, it often crashes at the start of a youtube video. Has anyone here had similar experiences?
I've got Model 11,2. Using pulseaudio.

- you could try flashplugin-beta from aur
- or you could use google-chrome (stable/beta/dev) from aur which has flash included
- or you could use chromium-pepper-flash from aur if you are using chromium from pacman

gofvonx · 2014-05-01 09:50:23

Sc0rian wrote:

sh4nks wrote:
Is it normal that everytime I upgrade the kernel, I need to reinstall the broadcom driver?
yup.. have to admit on the new kernel the broadcom driver has became super unreliable.

If the macbook (without ethernet adapter) is my only machine running arch, what is the recommended procedure to upgrade the kernel and obtaining the driver again (since one needs to build the package with the new kernel).

Downloading the broadcom-wl.tar on MacOS, unpack it and running

 makepg -s

ends in "Downloading hybrid-v35..." which does not work without internet.

sh4nks · 2014-05-01 20:53:03

gofvonx wrote:

Sc0rian wrote:
sh4nks wrote:
Is it normal that everytime I upgrade the kernel, I need to reinstall the broadcom driver?
yup.. have to admit on the new kernel the broadcom driver has became super unreliable.
If the macbook (without ethernet adapter) is my only machine running arch, what is the recommended procedure to upgrade the kernel and obtaining the driver again (since one needs to build the package with the new kernel).
Downloading the broadcom-wl.tar on MacOS, unpack it and running
 makepg -s 
ends in "Downloading hybrid-v35..." which does not work without internet.

I am keeping the driver package, so that I just need to install it with "pacman -U" if a new kernel comes out.

baronmog · 2014-05-01 22:33:40

gofvonx wrote:

Sc0rian wrote:
sh4nks wrote:
Is it normal that everytime I upgrade the kernel, I need to reinstall the broadcom driver?
yup.. have to admit on the new kernel the broadcom driver has became super unreliable.
If the macbook (without ethernet adapter) is my only machine running arch, what is the recommended procedure to upgrade the kernel and obtaining the driver again (since one needs to build the package with the new kernel).
Downloading the broadcom-wl.tar on MacOS, unpack it and running
 makepg -s 
ends in "Downloading hybrid-v35..." which does not work without internet.

Do you already have wireless working with your current kernel? If so, you can replace the broadcom-wl package with the broadcom-wl-dkms package. After installing it, follow the directions at https://wiki.archlinux.org/index.php/Dy … le_Support .

After that, the dkms daemon will rebuild and install the wl.ko driver whenever it detects that you've booted with a kernel that doesn't have it.

regulament · 2014-05-02 11:36:39

To be noted that on my system I had broadcom-wl-dkms installed and when I upgraded to 3.14.2 wireless broke( never had any problems before). Reinstalling broadcom-wl-dkms did not fix the problem.

I did install broadcom-wl and it worked perfectly.

baronmog · 2014-05-02 16:46:16

regulament wrote:

To be noted that on my system I had broadcom-wl-dkms installed and when I upgraded to 3.14.2 wireless broke( never had any problems before). Reinstalling broadcom-wl-dkms did not fix the problem.
I did install broadcom-wl and it worked perfectly.

The dkms breakage is because the 3.14.2 kernel was built with the 4.9 gcc and used a flag, fstack-protector-strong, that was introduced in 4.9. However, the stable repos were still serving up gcc 4.8, so compiling kernel modules was broken after the kernel update (see https://bbs.archlinux.org/viewtopic.php?pid=1411426). However, as of this morning, the stable repo is serving up gcc 4.9, so that should be fixed.

For future reference, you can downgrade an updated package by pointing pacman at the package cache in /var/cache/pacman/pkg/.

pacman -U /var/cache/pacman/pkg/linux-*3.14.1*

is the command I used yesterday to downgrade to a kernel and kernel-headers package that I could get wireless working with.

regulament · 2014-05-02 18:36:33

Oh, so that's why that happened. Unfortunately I needed my notebook for work and I couldn't pursue and see why or how it broke. Thanks for the info.

gofvonx · 2014-05-03 08:08:27

baronmog wrote:

Do you already have wireless working with your current kernel? If so, you can replace the broadcom-wl package with the broadcom-wl-dkms package. After installing it, follow the directions at https://wiki.archlinux.org/index.php/Dy … le_Support .
After that, the dkms daemon will rebuild and install the wl.ko driver whenever it detects that you've booted with a kernel that doesn't have it.

It works, really helpful. Thank you!

regulament · 2014-05-05 07:33:39

I just started experiencing a weird issue. From time to time( twice since Friday, three times in total) the screen goes all black, with some pixels on the left on( no clear pattern).

I've got no idea why or how this is happening. It happened during a pacman update one time and while editing a text file in Sublime the other time. I could not switch to another TTY, no key-bindings worked( tried to start a terminal to reboot the machine) and the only thing that worked was holding the power button to reboot the machine.

Any idea where I should look for problems? And how? It's the first time I am trying to debug things on linux. I tried searching, but I have no idea where to start.

Last edited by regulament (2014-05-05 07:34:10)

fixel · 2014-05-05 20:27:05

regulament wrote:

I just started experiencing a weird issue. From time to time( twice since Friday, three times in total) the screen goes all black, with some pixels on the left...

Same issue.. maybe intel drivers?

regulament · 2014-05-06 08:58:53

Could be. But again, I have no idea where to start searching for the cause of the problem.

fixel · 2014-05-06 14:34:43

There was an intel-dri update released today, I haven't had any issues since upgrading.

EDIT: Nope, still causes the issue.

Last edited by fixel (2014-05-07 11:20:46)

heipei · 2014-05-07 09:39:49

Sc0rian wrote:

I have no "default controller"
If I run "stress" on a base arch install my cpu hits over 100c, with fan sitting at 1200rpm..

Same here, my CPU is above 90c and the fan is doing somewhere between 1200 and 2200 RPM. I know that the max temperature for my i5-4258U is 100c, but 95c is cutting it awfully close in my opinion...

I certainly don't want to play around with the fan control. The easiest way for me to avoid these temps is to use no_turbo and/or reduce the max performance in

/sys/devices/system/cpu/intel_pstate/

vanja_z · 2014-05-12 11:37:36

Hey guys, I'm getting around twice the power consumption using Arch compared to OS X when idle and under light use. Takes 8 mins to drain 1% under OS X and 4 to 4.5 mins under Arch depending on power settings. I've gone thru all steps in the wiki for power saving with very little noticeable difference. I made a thread but nobody replied, do you guys have any advice? Can anyone confirm or deny my power usage? Any idea what could be using all this power?
https://bbs.archlinux.org/viewtopic.php?id=181374

regulament · 2014-05-12 13:40:53

Power consumption is not measured by how long it takes to drain 1%. And you should explain what programs you have running, what brightness, etc. A web browser in OS X might drain more power than one in Linux or either way around and so on.

Install powertop and go from there. See exactly what is eating up all that power. Not just what state the CPU is in.

There are some posts about power usage in this thread, read that.

vanja_z · 2014-05-13 03:00:45

Wow, so you managed to craft a hostile, dismissive and useless reply without even bothering to comprehend my post.

Power consumption is not measured by how long it takes to drain 1%

Yes, yes it is. Some basic electronics shows that the unit of power (W) is proportional to percentage drain (mA H) times the cell voltage, roughly a constant (V) divided by time (min). Since there is no cross platform or accurate way to measure watts there is nothing wrong with measuring the percentage drain over time to compare power consumption.

And you should explain what programs you have running, what brightness, etc. A web browser in OS X might drain more power than one in Linux or either way around and so on.

Again, I did explain all that in the thread, I have NO PROGRAMS running during these tests. I'm talking about idle power consumption so that should be pretty easy to deduce without me explicitly stating it (although I did in the linked thread).

Install powertop and go from there. See exactly what is eating up all that power.

Again, yes I have already said that I have done that. I have gone thru all the steps of enabling power save for audio, blacklisting blutooth, PCI power management, etc. Everything in the wiki and everything under powertop "tunables", those steps improved it from 4 mins to 4.5 mins per percent. Still much worse than 8 mins in OS X.

ewaller · 2014-05-13 03:23:28

First off <moderatorMode>Be good to each other</moderatorMode>

Second <electricalEngineerMode>
Power is voltage * current. Current is measured in mA.
mAH are milliamp hours and represent charge. i.e. Coulombs.
maH * Volts has an unusual unit of Coulomb Volts. I have no clue what that would be used for. [Edit: Yes I do, it will work out to be energy]
dV/dt might be interesting depending on the function of the battery voltage versus charge, like if it were linear, or some monotonic function; but...

Lithium polymer and Lithium Ion batteries have almost no slope across most of their discharge curve. I have seen some that are even non-monotonic over their discharge curve. In other words, as the battery discharges, there are regions where the voltage can actually increase.

Modern batteries have battery supervisor circuits that integrate current in and out of the battery to keep track of the charge. The processor communicates with this chip using a low speed bus such as SPI or i2c. The real question is, does Linux use the same algorithm to convert what it learns from that chip to percentage as does Apple in OS/X? Are you sure? Or are we comparing Apples to Penguins?

Edit, BTW. mWH is energy, but it has little to do with the battery voltage. Modern laptops will use a buck converter to drop the battery voltage down to the power rail voltages. Those converters are pretty efficient (90%+) and pretty much mirror the output power to the input power * the efficiency. So, as long as the battery stays above its minimum discharge voltage, so the current delivered to the rails is different than the current drawn from the battery. The current from the battery is the power divided by the current battery voltage.

Last edited by ewaller (2014-05-13 06:35:40)

vanja_z · 2014-05-13 06:09:18

Dude, I don't think I'm the one being disrespectful. I feel giving a flippant and unhelpful response without even bothering to properly read my question is disrespectful, anyway.

Yes, mAH represents charge but you divide that by time taken to draw that charge and you get back to mA (current) multiply that by battery voltage and you get power draw. It's really simple and I don't know why you people are saying that I can't measure the power usage by measuring the time taken to use a certain percentage of battery.

Lets say I use 1000 mAH in one hour then the current draw was 1000 mA. Now if the cell voltage was roughly constant at 8 V, then the power draw was 8W. Obviously made up numbers for convenience.

EDIT. OK, I'm not saying it's a 100% accurate way of measuring power but in this case I don't think it accounts for a 200% difference between the two OS.

Last edited by vanja_z (2014-05-13 06:11:22)

vanja_z · 2014-05-13 06:15:14

ewaller wrote:

The real question is, does Linux use the same algorithm to convert what it learns from that chip to percentage as does Apple in OS/X? Are you sure? Or are we comparing Apples to Penguins?

I'll try and investigate more but like I said I've checked a few points in the discharge cycle (30%, 50%, 80%) and the readouts under both OS agree within 2%.

ewaller · 2014-05-13 06:32:49

Yeah, I was just coming back to reconsider what I wrote; mA * h * V is actually Energy. The problem is that V is a function of both time and current -- and it is really non-linear
Divide that by t, and we are back to power again. If the battery voltage were constant, and the percentage remaining was an accurate representation of remaining capacity, yeah, it would work. I just do not know how those percentages are generated in OS X or in Linux. Assuming, for the sake of argument, the OSes use the same algorithm, I am still not convinced that a percent at 90% of battery life represents the same amount of energy as a percent at 10% of remaining life. If you were to control the experiment so the test was performed at the same charge state for both OSes, I might start to say, okay. But, I still think the telemetry is lying to us and that x% in OSx and x% in Linux do not represent the same charge state.

Could they be off by 100%? I bet they could. Do I think Linux draws more than OSx? You bet. Apple designed the hardware and the software to optimize the battery life on that machine.
How do we minimize the error? Integrate it over a much longer sample period. Try this. Run from a full charge down to a half charge according to Linux. Reboot to OSx, and see where the meter indicates. Run it all the way down and see how much longer you can run on the initial charge. Then lets play with the ratio of the times and the extent to which the OSes agree upon what half charge is.

Edit. we were posting at the same time, you did much of this ....

Edit 2: Signing off. Good night.

Last edited by ewaller (2014-05-13 06:38:47)

vanja_z · 2014-05-13 07:28:39

Thanks ewaller, I hear what you're saying and I agree. I think I've dealt with all of the experimental uncertainty you mentioned by doing an 80 minute test which went from 50% to 30% on Arch and 50% to 40% on OS X. Even if there is say 3% discrepancy in the battery percent readout and a 1% error due to the step of battery % and all of the error is in favour of Arch it still doesn't explain the whole difference. I've also done a bunch of shorter 20 min test over the range 100% to 30% and it is very consistently 4 to 4.5 mins per %.

Another note is that the laptop is juuuust slightly warm to touch when idle under Arch but stone cold under OS X.

I'll keep looking.

regulament · 2014-05-13 09:32:01

Yeah, sorry for being so flippant. It' just that I find it rather odd that your machine uses twice the power of OS X when mine uses around a third.

After optimizing everything with powertop and reading all the posts about power saving both on 10,x and 11,x, with full brightness, full keyboard illumination, wireless on, bluetooth off I get 10.5 W and about 8h of battery life. OS X with no programs running and same settings, I get 16 W drainage( installed some app to get the some basic values and I did the calculation to get 16). All of this on 11,2.

And I stand by it. I have 6 hours of classes and at the end of those six hours I still have about 1:30 hour of juice. This is with a slightly lower brightness, wireless on, keyboard illumination on, Sublime Text 3 running, about 3 terminal instances and dwb or Opera( I tend to close the browsers if I don't use them because they do eat a lot of power. I yet have to do test to see which one is the best from a power usage point).

LE: Yes, the machine is warmer on Linux for me too. I think it's because of Flash( it gets soo hot as soon as I run something in Flash), but I can't be sure. In Idle or with no Flash it's about the same as OS X. Right now I have 1 terminal and Opera with plug-ins off running with lm-sensors reporting 37 degrees.

Regarding the screen going black problem, up until today I did not have any problems with it anymore. Updated the kernel and bam, issues again. It might be some weird intel-dri/kernel issue. I'll wait for the Intel update to see if this is true.

Last edited by regulament (2014-05-13 09:38:53)

vanja_z · 2014-05-13 14:09:35

oh, thanks for that info regulament, I'm also sorry I over reacted a bit.

I'm making a bit of progress and will report back tomorrow.

DrKillPatient · 2014-05-13 15:24:11

With regards to the screen-blanking problem: do you have "acpi_backlight=vendor" in your kernel boot line? I had a similar issue a while ago that was fixed by adding that, though for me it only occurred after changing the resolution (plus it would get stuck like that across multiple boots). Still, maybe it's related? The other symptoms seem similar.

Last edited by DrKillPatient (2014-05-13 15:28:03)

Arch Linux

#201 2014-04-28 20:45:55

Re: Macbook Pro Late 2013 Retina

#202 2014-04-30 01:48:29

Re: Macbook Pro Late 2013 Retina

#203 2014-05-01 09:50:23

Re: Macbook Pro Late 2013 Retina

#204 2014-05-01 20:53:03

Re: Macbook Pro Late 2013 Retina

#205 2014-05-01 22:33:40

Re: Macbook Pro Late 2013 Retina

#206 2014-05-02 11:36:39

Re: Macbook Pro Late 2013 Retina

#207 2014-05-02 16:46:16

Re: Macbook Pro Late 2013 Retina

#208 2014-05-02 18:36:33

Re: Macbook Pro Late 2013 Retina

#209 2014-05-03 08:08:27

Re: Macbook Pro Late 2013 Retina

#210 2014-05-05 07:33:39

Re: Macbook Pro Late 2013 Retina

#211 2014-05-05 20:27:05

Re: Macbook Pro Late 2013 Retina

#212 2014-05-06 08:58:53

Re: Macbook Pro Late 2013 Retina

#213 2014-05-06 14:34:43

Re: Macbook Pro Late 2013 Retina

#214 2014-05-07 09:39:49

Re: Macbook Pro Late 2013 Retina

#215 2014-05-12 11:37:36

Re: Macbook Pro Late 2013 Retina

#216 2014-05-12 13:40:53

Re: Macbook Pro Late 2013 Retina

#217 2014-05-13 03:00:45

Re: Macbook Pro Late 2013 Retina

#218 2014-05-13 03:23:28

Re: Macbook Pro Late 2013 Retina

#219 2014-05-13 06:09:18

Re: Macbook Pro Late 2013 Retina

#220 2014-05-13 06:15:14

Re: Macbook Pro Late 2013 Retina

#221 2014-05-13 06:32:49

Re: Macbook Pro Late 2013 Retina

#222 2014-05-13 07:28:39

Re: Macbook Pro Late 2013 Retina

#223 2014-05-13 09:32:01

Re: Macbook Pro Late 2013 Retina

#224 2014-05-13 14:09:35

Re: Macbook Pro Late 2013 Retina

#225 2014-05-13 15:24:11

Re: Macbook Pro Late 2013 Retina

Board footer