FedoraForum.org

FedoraForum.org (http://forums.fedoraforum.org/index.php)
-   Reviews, Rants & Things That Make You Scream (http://forums.fedoraforum.org/forumdisplay.php?f=25)
-   -   Power regression is not fixed in 3.2.7 (http://forums.fedoraforum.org/showthread.php?t=277043)

morlock 29th February 2012 03:56 PM

Power regression is not fixed in 3.2.7
 
I was considering the title “I want to use Linux but I can’t.” It pretty much sums up my one and a half decade old relationship with this OS.

I have an IBM T60p laptop. I am running dual boot XP and Linux (various distributions). When kernel 2.6 came out, I was running Fedora (can’t remember which release anymore) and was hit with the power regression problem. My laptop grossly overheated. It stayed up but when I rebooted, it refused to boot until the hardware cooled down a bit. The laptop has no issues on Windows. Windows runs perfectly on it and the hardware stays cool. I tried Ubuntu and Mint as well but they all had the same power regression problem, so I gave up on Linux for a while.

I kept reading about the power regression issue and when it was reported fixed (in kernel 3.2.5) I immediately installed Fedora 16 and upgraded the kernel to the latest of greatest 3.2.7 PAE.

Result: the power regression is alive and well and worse than ever. I am beginning to hate Linux (don’t take this personal).

I discovered other issues too:
- I couldn’t use the Software Update tool on KDE, it just wouldn’t work. I had to do all my updates with yum.
- I can’t connect to Windows shares at all. Previously it wasn’t a problem, now it is. Yes, I opened up the necessary ports through the firewall.

I am slowly coming to a conclusion that I am witnessing the end of Linux as a desktop/laptop operating system. With every release the OS is progressively worse and development is not focused on addresses the issues. The problems users face on Linux are absolutely unacceptable. It’s even more unacceptable that the developers, instead of fixing the frigging power management problem, are busy trying to manufacture excuses for it. They blame the hardware, ignoring the fact that the users reporting the problem usually state that their laptops are dual boot and they have no problems on Windows. An even more outrageous excuse is that Linux is “optimized for performance”. I laughed for minutes when I read it.

Fact: when I run Windows on the same hardware, the laptop stays cool and the fans are quiet. When I run Linux on the laptop, the fans make a horrendous racket. It sounds frightening After an hour or so the laptop heats up to the point where I can barely touch it. It’s obvious that Linux can’t manage the hardware. It’s not just laptops but desktops as well, except users don’t notice when desktops overheat. Linux has a fundamental problem which makes it impossible to use and developers do squat to address it. Well… I guess you get what you paid for…

Question! Have you ever wondered what that extra heat does to your PC? Do you think it will shorten its life? How often can you afford to buy a new PC?

If it dies while still under warranty, what are the odds that the manufacturer honors the warranty if they find out that you ran an unauthorized OS on it which is known to overheat the hardware?

Here is an even scarier thought: if you leave your Linux laptop on the table and it bursts in flames and burns down your house, will your home insurance cover the damage?

I go further: if your laptop burns down the house and your children die in the fire, will you still stand by your beloved “free” operating system and insist that it’s “10 times better than Windows will ever be” (this one is a riot)?

Here is something Linux developers often overlook: 99% of people own computers because they want to do some sort of work of communication on it. The last thing the 99% want to do is spend 99% of their time troubleshooting something that works fine under Windows.

Well, the 15 or so years I have put in so far proves that I am not a blind Linux hater. I would love to use Linux… only if it worked properly! Slackware, Mandrake, RedHat, Suse, Mandriva, Fedora, Ubuntu, Mint, just to mention the ones I tried. Linux is undeniably cool and I so would love to use it but I can’t put up with these idiotic problems!

When the 3.3 kernel comes out I’ll give it another whirl. If it still overheats, I’ll put Linux down until the fall then try the new release again… If it still doesn’t work I’ll put it down again, and so on and so for.

This OS is dying on the PC market. When will you people wake up to this reality? Here is a suggestion: instead of racing to release 2 versions every year, why don’t you people learn from Microsoft and just release a single version every 2-3-4 years and spend your valuable time fixing the issues instead? Just a thought…

hadrons123 29th February 2012 04:17 PM

Re: Power regression is not fixed in 3.2.7
 
instead of ranting here you can install 3.3-rc5 kernel and find whether the ASPM fix is working!

vein 29th February 2012 04:23 PM

Re: Power regression is not fixed in 3.2.7
 
What specifically is overheating? The reason I ask is that when installing linux on my laptop, I found it did just fine until installing the necessary module for my wireless card. The wireless hardware sits right below the track pad and would boil in the unit, with the heat radiating to other parts of the laptop. The "power regression" and heat aren't necessarily synonymous, and in fact the cause is likely a single component whose drivers aren't managing it correctly. I definitely sympathize with the frustration of having to avoid using linux for a given laptop, but if you can identify what is overheating, you may be able to find a better/alternate driver or module that can allow you to avoid the overheating. If it really is power management with the processor, there is already something in KDE that should allow you to set the processor to idle at a specific speed, etc.

Holdolin 29th February 2012 05:01 PM

Re: Power regression is not fixed in 3.2.7
 
Quote:

Originally Posted by morlock (Post 1558812)
Here is an even scarier thought: if you leave your Linux laptop on the table and it bursts in flames and burns down your house, will your home insurance cover the damage?

I was reading right up to that point. If your system allows itself to overheat to that point, you have a serious hardware issue. The system hardware, specifically BIOS settings will shut the system down LONG before it even comes close to flames. Thank you for the sensationalism though, it made my morning a bit brighter :)

marko 29th February 2012 05:58 PM

Re: Power regression is not fixed in 3.2.7
 
Quote:

Originally Posted by hadrons123 (Post 1558819)
instead of ranting here you can install 3.3-rc5 kernel and find whether the ASPM fix is working!

Is that equivalent to this one ? (I'm not sure about the git tags)

http://kojipkgs.fedoraproject.org/pa...t0.3.fc17/src/

hadrons123 29th February 2012 06:06 PM

Re: Power regression is not fixed in 3.2.7
 
not from koji.
i meant from kernel.org
edit:
link here

Fenrin 29th February 2012 07:48 PM

Re: Power regression is not fixed in 3.2.7
 
Quote:

Originally Posted by morlock (Post 1558812)
[...]

IMHO heat is on Linux systems mosty a issue if mobile computers or other thin computers are used.

If your overheated computer uses a ATI or AMD card, you should either use the proprietary driver for proper proper power saving features or change the KMS powermanagement options via this guide if you want to use the default driver.

Intel IGP + Nvidia card is on Notebooks more troublesome than other GPU or APU solutions from AMD/ATI or Intel. At least as long as Nvidia Optimus isn't supported under Linux.

The power saving features of Intels Sandybridge drivers works also not good under Linux yet. The SandyBridge power saving feature RC6 is deactivated by default under Linux. In certain kernels it's possible to turn it on via i915.i915_enable_rc6=1 in the boot line.

pcie_aspm=force can also be useful as option in the boot line (this should work on kernel 3.3 and 3.2.5). If your notebook has a USB 3 port, it probably doesn't shut down properly without some configuring work. Well maybe it works on the newest kernel versions, I didn't read this article here.

I read this test here from a german IT news site, according to their test it was possible to almost double the battery life of an Asus UX31 ultrabook under Linux with much configuring work and tweaks. When they used Unity the power consumption was considerable higher than with KDE.

I'm glad that I mostly don't need mobile computers and my desktop computer has ussually temperatures below 40° C.

bob 29th February 2012 08:26 PM

Re: Power regression is not fixed in 3.2.7
 
Moved to Reviews, Rants & Things That Make You Scream. There's no questions being asked here, just comments and ones that should be addressed to the devs, who, for the millionth time, don't frequent this Forum. In short, get it off your chest, friend but we don't make the distro, we just use it.

BTW, my desktop's running at 18.5C or 65F. right now with just the cheapo stock case fan and cpu fan. Kinda chilly in here, I guess! :)

RupertPupkin 1st March 2012 04:21 PM

Re: Power regression is not fixed in 3.2.7
 
Quote:

Originally Posted by morlock (Post 1558812)
When will you people wake up ... why don’t you people learn from Microsoft...

I think the "you people" part is my favorite! :)

Uh oh, gotta go. Looks like my children are dying in a fire caused by Microsoft Windows. I'm sure you people will understand.

bob 1st March 2012 04:36 PM

Re: Power regression is not fixed in 3.2.7
 
Interesting thread here: https://bbs.archlinux.org/viewtopic.php?id=49980 These guys are claiming that the heat issues were from the manufacturer's problems, although that doesn't explain why Windows runs cooler. There are some "fixes" that seem to work but again it's retrofitting your machine, not related to distro packages.

hadrons123 1st March 2012 04:40 PM

Re: Power regression is not fixed in 3.2.7
 
@BOB

that is an ancient thread. It possibly might not be relevant to new processors.

Holdolin 1st March 2012 04:46 PM

Re: Power regression is not fixed in 3.2.7
 
Quote:

Originally Posted by RupertPupkin (Post 1559102)
I think the "you people" part is my favorite! :)

Uh oh, gotta go. Looks like my children are dying in a fire caused by Microsoft Windows. I'm sure you people will understand.

I like that whole burning down thing cuase, while I doubt my Linux knowledge is equal of many of you foks here, I have been an avid hardware nut for years. Performance builds, overclocking and all that fun stuff. I have inded learned some of the "don'ts" the hard way :doh:

bob 1st March 2012 08:34 PM

Re: Power regression is not fixed in 3.2.7
 
Hadrons123, he's running a IBM T60p laptop which came out in 2006. That's also the lappy that the thread's about.

Dan 1st March 2012 09:34 PM

Re: Power regression is not fixed in 3.2.7
 
Now, wait a minute, guys. Cut the man some slack. A chronically overheating laptop is enough to make anyone a little cranky.

http://forums.fedoraforum.org/showthread.php?t=262259

Although ... in the case of the ol' Fukushima Special ... it runs about 10~20F hotter in Windows Vista than it ever does in Linux.

Go figure?! <..:confused:..>

Holdolin 1st March 2012 10:14 PM

Re: Power regression is not fixed in 3.2.7
 
But Dan (and that sounds weird me saying that, as it's my name also) i was just funning, well aside from my first post. Yea, it's frustrating having any type of chronic computer problem.

Heh, speaking of overheating I remember the day I learned the hard way i NEED an audible alarm for either the CPU fan speed going to low or a CPU temp going to high. For the entertainment of all, I'll explain and once again get laughed at (my friends over at extremeoverclocking had no end of fun with me for quite a while :doh: Long story short, I had this shiney new i7 gaming rig, OC'd and water cooled. I was in the middle of a game one day and the game started to get really choppy, to the point I exited the game to see take a look at my thermal sensors. Imagine the surprise when i noticed my CPU temp was at 99 degrees Celsius, and had topped out at 101. At that point i just hit the power button and opened the case. Would have been neat to see the bubbles in my liquid lines had i not known that that was a sign that my thermal sensors were right, my rig tried to melt itself.

Now, unlike Morlock, i had absolutely nobody to blame but myself. I should have known much better than to have no audible way of knowing my system was overheating. Nonetheless, i was furious. Worst of all I nowhere I could channel it. It wasn't a misbehaving OS, it wasn't a poorly designed application. Yes, it was failed hardware, but stuff happens, and had i taken the reasonable precaution of an audible alarm, or even lowered the temp at which the CPU would throttle itself back, the failure would have simply meant that I would have shut the system down long before I cooked my CPU. Yes, it was cooked. Within a month I started having stability issues, and when i (ouch my wallet) replaced the CPU, things were fine.

Anyway, I do feel yer pain Morlock. Yea, I was a bit quick with the whit in my first post, but I do feel ya. I even gave you a good reason to laugh at me :)


All times are GMT +1. The time now is 12:57 AM.

Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2014, vBulletin Solutions, Inc.