Linux kernel 4.12 is Out!

[suicidalfranco]

This weekend saw the release of Linux kernel 4.12 which, by his creators own word, is the biggest kernel ever released so far taking the title away from kernel 4.9.

Quote

As mentioned over the various rc announcements, 4.12 is one of the bigger releases historically

   
What's new in this new release you might want to ask?
Kernel 4.12 brings support for new hardware, new I/O schedulers and over all improvements, here are some worthy notable bullet points:

•   Improvements to F2FS
•   Initial support for Nvidia GTX1000 series cards with Nouveau
•   Initial Radeon RX Vega support
•   Support for Intel RealSense SR300 Camera
•   Support for Razer's sabertooth
•   Support for MadCatz BrawlStick
•   USB Type-C port manager
•   Improved keyboard backlight support in Dell's laptop driver
•   Support for Intel's volume button driver
•   New Thinkpad T470 hotkeys mapping
•   Sound driver fixes for Skylake and Atoms SoCs
•   New sound driver for various sound cards/chipsets
•   Support for ARM devices including: Orange Pi PC2, Rockchip RK3399 and RK3288, Nvidia Tegra X2

Now to deal with the elephant in the room: Dat new I/O scheduler though!
SO BFQ, Budget Fair Queue, as been in the works for the last few years, and only been included now on the mainline kernel. Some distro arleady tried to included by manually patching it onto their OSs and even had a PPA for Ubuntu,
but now it's finally available for all to admire and contemplate! To make it simple for yall: BFQ is an I/O scheduler that balances out how the systems access data located in drives (HDD SSD, you name it), BFQ targets desktop users
by making that at no point a single applications eats up all of the bandwith to said drives. For any Linux users here who have encoutered this issue, yours truly here included, this will now avoid that the system freezes or hangs
when copying large files for example. But, there's always a but, BFQ isn't enabled by default in 4.12 and is not the default scheduler yet unless your distro provider decides it to be, if you want all you have to do is modify
/etc/default/grub and add "scsi_mod.use_blk_mq=1 elevator=bfq" to "GRUB_CMDLINE_LINUX_DEFAULT" like so:

GRUB_CMDLINE_LINUX_DEFAULT="quiet splash scsi_mod.use_blk_mq=1 elevator=bfq"

and run the command update-grub2, finally reboot and you're good to go.
For more info check the source on the BFQ or 4.12 in general check the source or the full linux kernel commit log hereJust to brag how big 4.12 is here are some stats that have been shared to the community:
It took 63 days to create it
in that timelapse 14570 commits were made accross 59806 files
with a total of 24170860 lines of code the kernel files were raised by 3.18% percent and 4.47% in lines of code and 795.58 lines were added each hour

 

And some final words: This year also see Linux's highest grab in marketshare in history! A whoping 2.36%!!!!!1!!!!

MIND = BLOWN!

 

souce: https://kernelnewbies.org/Linux_4.12

 

[fjdhdhcyfhf]

Phoronix has already reported pretty good performance improvements. I want to know if they fixed the USB sleep bug with Nvidia drivers, but somehow I doubt it.

20 minutes ago, suicidalfranco said:

And some final words: This year also see Linux's highest grab in marketshare in history! A whoping 2.36%!!!!!1!!!!

That is a crap place to get stats. Steam hardware surveys are more interesting when you consider they are promoting the platform:

 

 

[suicidalfranco]

6 minutes ago, SCHISCHKA said:

Phoronix has already reported pretty good performance improvements. I want to know if they fixed the USB sleep bug with Nvidia drivers, but somehow I doubt it.

That is a crap place to get stats. Steam hardware surveys are more interesting when you consider they are promoting the platform:

 

 

how is steam survey better when it's poll is smaller than the used by NetMarketShare? I'm not saying they're the be all end all of marketshare analysis tools but to say that a poll who only gets it's data from people who installed steam on their system and accepted to take the survey is better than one who simply sees what has accessed the 40000 websites they keep an eye on, is a bit sily imo. Of curse NMS stats should be taken with a scoup of salt and will always be open to interpretation, debate and potential errors

[suicidalfranco]

1 minute ago, Tedny said:

Use Unix-based systems for games 0_0 

It's not as bad as it used to be

[tlink]

10 minutes ago, Tedny said:

Use Unix-based systems for games 0_0 (lol. wat ) 

>75% of my steam games support linux.

[suicidalfranco]

1 minute ago, Tedny said:

you know.... that made not for....entertain, you know...... Windows for that 

meh, still can play Xcom2 and Civ6 without opening my WinVM, and have no issues when doing soooo... won't complain.

[fjdhdhcyfhf]

22 minutes ago, suicidalfranco said:

how is steam survey better when it's poll is smaller than the used by NetMarketShare? I'm not saying they're the be all end all of marketshare analysis tools but to say that a poll who only gets it's data from people who installed steam on their system and accepted to take the survey is better than one who simply sees what has accessed the 40000 websites they keep an eye on, is a bit sily imo. Of curse NMS stats should be taken with a scoup of salt and will always be open to interpretation, debate and potential errors

The important aspect of statistics to apply here is that with web browsing stats there is no way to measure margin of error, whereas steam surveys are from unique accounts. If you look at w3schools stats they are vastly different. When I look at stats from my clients web sites all linux views really are just me.

 

Web stats have too many variables and double counting. Every time I surf the net my cookies get cleared and I use a different IP so I am counted more than once. Does your grandma on Windows 10 do the same? I set up web analytics occasionally for clients, I have seen my own browsing influence over them and they cannot be taken too seriously. If you remove the double counting you can get interesting stats, for example you can track the sequence of pages viewed and where they stop viewing your site. Another one I can say with certainty 100% of IE6 and windows XP come from china - this stat is actually real and observable across many sites.

 

 

 

 

[suicidalfranco]

1 hour ago, SCHISCHKA said:

The important aspect of statistics to apply here is that with web browsing stats there is no way to measure margin of error, whereas steam surveys are from unique accounts. If you look at w3schools stats they are vastly different. When I look at stats from my clients web sites all linux views really are just me.

 

Web stats have too many variables and double counting. Every time I surf the net my cookies get cleared and I use a different IP so I am counted more than once. Does your grandma on Windows 10 do the same? I set up web analytics occasionally for clients, I have seen my own browsing influence over them and they cannot be taken too seriously. If you remove the double counting you can get interesting stats, for example you can track the sequence of pages viewed and where they stop viewing your site. Another one I can say with certainty 100% of IE6 and windows XP come from china - this stat is actually real and observable across many sites.

 

 

 

 

and the issue with Steam surveys is that not everyone accept to take the survey and Steam doesn't even querry all Steam users to take it, only a portion gets asked, and then Steam makes an estimate around that. How's this better than NMS? At least NMS as a larger population to work with

[leadeater]

1 hour ago, suicidalfranco said:

meh, still can play Xcom2 and Civ6 without opening my WinVM, and have no issues when doing soooo... won't complain.

How does Civ6 compare to Windows performance wise?

[fjdhdhcyfhf]

8 minutes ago, suicidalfranco said:

and the issue with Steam surveys is that not everyone accept to take the survey and Steam doesn't even querry all Steam users to take it, only a portion gets asked, and then Steam makes an estimate around that. How's this better than NMS? At least NMS as a larger population to work with

again unique users. oh and they can count the number who do not participate but i dont know if they report that

[suicidalfranco]

38 minutes ago, leadeater said:

How does Civ6 compare to Windows performance wise?

don't know, haven't played it on Windows. I do get Stable 60 on my 390 while playing on my TV (1080p with vsync and highest preset)

EDIT: With AMDGPU-PRO drivers

[leadeater]

3 minutes ago, suicidalfranco said:

don't know, haven't played it on Windows. I do get Stable 60 on my 390 while playing on my TV (1080p with vsync and highest preset)

Seems there still a bit of stuff to work on, soon as RTS and TBS games start running better on Linux I'll likely consider switching.

 

Quote

Multithreading is supported now, but not on the rendering path. Our implementation of OGL has issues with multithreading unfortunately, and we are actively looking at alternatives.

https://www.gamingonlinux.com/articles/aspyr-media-state-they-are-looking-at-ways-to-improve-civilization-vi-performance-on-linux.9130

[dfsdfgfkjsefoiqzemnd]

2 hours ago, SCHISCHKA said:

whereas steam surveys are from unique accounts.

Sure sure, but gamers tend to use Windows more which already skews the results because a lot of Linux users either don't game or have Steam only on their Windows PC/install. 

Also Linux users tend to be less likely to participate in surveys etc. 

 

There's just no way you're ever going to get an accurate number.

[fjdhdhcyfhf]

18 minutes ago, Captain Chaos said:

Sure sure, but gamers tend to use Windows more which already skews the results because a lot of Linux users either don't game or have Steam only on their Windows PC/install. 

Also Linux users tend to be less likely to participate in surveys etc. 

 

There's just no way you're ever going to get an accurate number.

a lot of assumptions there.

Ill give you a real assumption: People are using linux because it doesnt have anti-cheat.

What do you have to base your assumption linux users tend not participate in surveys?

[samcool55]

That's nice, let's hope that Vulkan helps Linux adoption, honestly M$ for what i care can eat shit and die.

 

[dfsdfgfkjsefoiqzemnd]

Just now, SCHISCHKA said:

What do you have to base your assumption linux users tend not participate in surveys?

Years of frequenting Linux forums.  Linux users tend to be more privacy-conscious than Windows users, so when we see that Steam wants to know what other software we have on our machines, that's usually enough to put us off. 

[GoodBytes]

Looks like it was filled with rewrites.. which can be both good (poorly made the first time) and bad (bugs and issues).

Now the question I am having is looking at posted feature list... since when a kernel becoming an OS? Why should graphics card support be part of the kernel? How moronic is this.. imagine buying a new version of Windows because you got a new graphics card....

[GoodBytes]

5 hours ago, SCHISCHKA said:

Phoronix has already reported pretty good performance improvements. I want to know if they fixed the USB sleep bug with Nvidia drivers, but somehow I doubt it.

That is a crap place to get stats. Steam hardware surveys are more interesting when you consider they are promoting the platform:

 

 

No its not. Linux is used like crazy on the web back-end. In fact, you can be 99.99% sure that each phone app you have or see in the Store (that uses the internet in some fashion) connects to a Linux server. Linux is not growing on the PC space, but is on the server space. This is another "missed the boat" issue for Microsoft.

[tlink]

4 hours ago, Tedny said:

sure, by they made for windows 

no, most of them are not made for windows. otherwise they wouldn't be able to run on linux.

[BuckGup]

5 hours ago, Tedny said:

Use Unix-based systems for games 0_0 (lol. wat ) 

All games that run on source run on UNIX. So there's a lot of games that run of Linux and MaxOS now days

[fjdhdhcyfhf]

16 minutes ago, GoodBytes said:

No its not. Linux is used like crazy on the web back-end. In fact, you can be 99.99% sure that each phone app you have or see in the Store (that uses the internet in some fashion) connects to a Linux server. Linux is not growing on the PC space, but is on the server space. This is another "missed the boat" issue for Microsoft.

Read it again

[fjdhdhcyfhf]

1 hour ago, Captain Chaos said:

Years of frequenting Linux forums.  Linux users tend to be more privacy-conscious than Windows users, so when we see that Steam wants to know what other software we have on our machines, that's usually enough to put us off. 

Try talk to people in the real world. The privacy point confirms what I said earlier about getting double counted in web stats

[suicidalfranco]

2 hours ago, GoodBytes said:

Looks like it was filled with rewrites.. which can be both good (poorly made the first time) and bad (bugs and issues).

Now the question I am having is looking at posted feature list... since when a kernel becoming an OS? Why should graphics card support be part of the kernel? How moronic is this.. imagine buying a new version of Windows because you got a new graphics card....

so that you don't need to install drivers cause whatever device you use is natively supported at the kernel level, the reason why most of the time you install a distro on some random machine you have it up and running without worrying wireless cards not working, resolution stuck at 800x600, no sound, no trackpad, etc. (<- my early linux days back in 2008). Also the reason why LiveCD/USB manage to recognise everything on whatever machine you plug them and actually give you a good testing environment limited only by the speed of your USB drive.

Also Linux is modular af, just as you can have it full featured, you can disable and remove everything you don't need, a desktop distro doesn't need the ARM/MIPS support, so they just remove it, and nothing breaks. Don't need the parts for i386 cause you only want your user base to use more recent hardware to get the best experience possible? done, it's gone.  

[leadeater]

4 hours ago, suicidalfranco said:

(<- my early linux days back in 2008).

Yep, early linux was rather brutal for getting stuff to actually work properly, not just work. Getting the old ATI drivers going for an X800 was a real pain in the ass, the actually good high quality 5.1 audio was another issue, wireless good luck. After you finally finished the pentathlon it was actually quite good and WineHQ worked very well for most games.

 

There is a reason why Ubuntu got so popular, they actually tried damn hard to fix all those issues first and made stuff just work. 

[suicidalfranco]

3 minutes ago, leadeater said:

Yep, early linux was rather brutal for getting stuff to actually work properly, not just work. Getting the old ATI drivers going for an X800 was a real pain in the ass, the actually good high quality 5.1 audio was another issue, wireless good luck. After you finally finished the pentathlon it was actually quite good and WineHQ worked very well for most games.

 

There is a reason why Ubuntu got so popular, they actually tried damn hard to fix all those issues first and made stuff just work. 

even trying to use a 3G dongle was a pain in the ass back then, the amount of shit i downloaded only to see them do nothing and the amount files, lines i had to add to some system file or system directory. I was pioneering, and i'm glad i did, because all that shit i went is what gave me the knowledge in computing i have now, the da before i tried Linux i didn't even know what an OS was until they use Ubuntu as an example at school to explain what OSs are, do and distinguish it from Windows