Slashdot Mirror
Linux 4.3 Bringing Stable Intel Skylake Support, Reworked NVIDIA Driver
An anonymous reader writes: Mr. Torvalds has released Linux 4.3-rc1 this weekend. He characterized the release as "not particularly small — pretty average in size, in fact. Everything looks fairly normal, in fact, with about 70% of the changes being drivers, 10% architecture updates, and the remaining 20% are spread out." There are a number of new user-facing features including stabilized Intel "Skylake" processor support, initial AMD R9 Fury graphics support, SMP scheduler optimizations, file-system fixes, a reworked open-source NVIDIA driver, and many Linux hardware driver updates.
If this were commercial there would be a lot more hype. It's nice to get just the facts.
> Now only if it had a decent text editor for ...
You could always use Vim ...
Or if you are really evil ... Vim mode for Emacs
If you really want to go straight to hell ... Emacs mode for Vim
Pick you religion / devil :-)
how will that affect older grub booting systems?
some very old systems know only about ext2. then there are some that only know about ext3.
I remember that if you diable journalling (or have closed the disk cleanly) that ext2 can read ext3.
is there any risk of an older system that can only read ext2 (or maybe 3) not being able to boot with even a cleanly shutdown ext4 fs?
and, would grub have to be updated?
somehow, removing ext3 seems wrong to me. ext4 has been out a while, but a fs is so important, its hard to believe that it was wise to remove a good, working collection of code like that.
(I do use ext4 on my current desktops but some embedded audio boxes still are ext2 and 3 based).
--
"It is now safe to switch off your computer."
Have you filed bug reports? People can't address problems they don't know exist. Many laptops are fully functional without problem, so you can't expect someone to know you have a problem.
I'm starting to think GNU is the problem with "GNU/Linux" these days.
This kind of stuff is why I use Windows as my primary OS now. I have a laptop and I have to be mobile right now. With Linux on my laptop I lose about 25% of the battery life and it also runs slower compared to Windows. Even worse it does not hibernate correctly and even sleep sometimes screws up. Sometimes when it wakes up from sleep under Linux the USB ports don't work.
I have just gotten tired of dealing with these issues and after all this time it is pretty clear that it is not a priority for developers of Linux. It is just easier to have virtualbox with Linux installed under Windows and use that.
Computer modeling for biotech drug manufacturing is HARD!
This kind of stuff is why I use Windows as my primary OS now. I have a laptop and I have to be mobile right now. With Linux on my laptop I lose about 25% of the battery life and it also runs slower compared to Windows. Even worse it does not hibernate correctly and even sleep sometimes screws up. Sometimes when it wakes up from sleep under Linux the USB ports don't work. I have just gotten tired of dealing with these issues and after all this time it is pretty clear that it is not a priority for developers of Linux. It is just easier to have virtualbox with Linux installed under Windows and use that.
I'm sure they'd like to do more, but I guess volunteers lack the hardware and none of the major laptop vendors have seen much profit in selling Linux preinstalled so there's no funding for paid employees to support it. Unfortunately there's a lot of hardware quirks that can't easily be worked out without being able to diagnose and test it on that particular hardware. Particularly if it just fails every once in a while due to a particular state/timing/condition. Maybe it would be possible to create some kind of remote hardware lab where you get remote management + button operation + webcam + microphone + hard power reset switch to do testing, scripted suspend/resume runs and self-tests to give developers broader access but you still need people with time and interest to fix your particular model. That might be hard in itself.
Live today, because you never know what tomorrow brings
Some companies test their software being releasing it, instead of expecting their customers to be their beta testers.
If you're going to use EMACS mode for VIM you might as well set your BASH prompt to this:
(I promise it's safe, it only calls echo, tr and sed none of which can read or write arbitrary files, create sockets or any of that)
SJW n. One who posts facts.
It only works if it's all running in the browser. Bonus points for running asm.js.
"First they came for the slanderers and i said nothing."
Some companies test their software being releasing it, instead of expecting their customers to be their beta testers.
Eat some of your own dog food and proofread your comment before releasing it. You know, that's kind of what preview is for.
CLI paste? paste.pr0.tips!
"Sounds like somebody's got a case of the Mondays."
Get thee glass eyes, and, like a scurvy politician, seem to see things thou dost not.--King Lear
It also doesn't work, you have too many \ characters in the ${PWD//\//\\\} section. I can't figure out what the sed command is doing though, that part is tough.
"First they came for the slanderers and i said nothing."
This kind of stuff is why I use Windows as my primary OS now. I have a laptop and I have to be mobile right now. With Linux on my laptop I lose about 25% of the battery life and it also runs slower compared to Windows. Even worse it does not hibernate correctly and even sleep sometimes screws up. Sometimes when it wakes up from sleep under Linux the USB ports don't work.
I have just gotten tired of dealing with these issues and after all this time it is pretty clear that it is not a priority for developers of Linux. It is just easier to have virtualbox with Linux installed under Windows and use that.
I'll readily admit that it's more work and tinkering with Linux, but --- given the willingness to take the time to get things done right up front, which is a one-time effort --- I have to disagree with your points. The following is based on my experience with the three laptops I've owned over the past few years.
1. I did some power optimization and battery life on Linux is about the same as on Windows. Not better, but not worse either. (It was definitely worse before optimization, I'll admit ... but that's what optimizing is all about.)
2. Runs slower? I've never seen that in any noticeable way. Neither can I say it's faster ... except that I can control the amount of strange stuff running in the background and I'm not phoning home to Microsoft.
3. Hibernation / sleep were indeed an issue on 2 of the 3 laptops. But they work fine after somewhat substantial effort. The key thing seemed to have been finding out which hardware drivers to restart on resume. In one case I put in alternative hibernation software. It was all a pain but the point is that it can be done.
A day or two of work up front and I run Linux problem free for months and years. Not everyone is willing to do the work, and/or feel that such work shouldn't be necessary. Fine --- then run Windows if it does everything you need. For me, Linux is about productivity and being able to get things done without the major annoyances of Windows (I'm being polite and understating things here). I find Linux is worth the extra effort.
My family have preferred Linux on their laptops since about 1999, with three exceptions, including my 90 year old mother - all three exceptions are Mac users.
Sent from my ASR33 using ASCII
Some companies charge for the OS so they can pay people to test it.
I do not fail; I succeed at finding out what does not work.
By runs slower I mean when I am doing OpenMP applications they are actually running slower under linux. I am not sure how it works but for some reason under Linux I get higher reported temperatures and the CPU does not stay in the higher turbo ranges like it does under windows.
I installed the Linux pstate driver for a Haswell i7 chip and I have it using the performance governnor.
My other issue is when I try to use the dedicated GPU for GPGPU work it is a pain in the ass under Linux. I know it is not Linux's fault that it does not support technology like optimus very well but that doesn't change that it still makes it a pain in the ass.
I used to use Linux as pretty much my only OS. But since going back to school and working on biotech drug process development I have found my goals have changed. I will use free software if it works better but overall for doing very high performance c++ (OpenMP and MPI) it is easier to develop and debug with Visual Studio and the Intel developer tools under windows.
At this point I just want software that works so I can do my primary work of writing computers simulations to help us manufacture drugs that we can't currently figure out how to manufacture and make cures and treatments available to people that need them. Fiddling with my OS to get it to work is just not worth it anymore. I have not had Windows 8.1 crash on this laptop and the upgrade to Windows 10 went without any problems.
Computer modeling for biotech drug manufacturing is HARD!
I know they would like to do more and I understand it is a very hard problem to solve.
I just don't want to deal with the problems anymore. If they figure it out great I will try it and see how it works but if not I will just use Windows 10, Visual Studio, Intel developer tools and MATLAB.
Linux runs on the laptop and all hardware is detected and supported but power management does not work as well as it does under windows by a long shot and GPGPU is a pain in the neck because of optimus support.
Computer modeling for biotech drug manufacturing is HARD!
For me Linux runs slightly better then Windows 10 on my laptop (Precision M4500). Windows 10 keeps breaking things (like my touchpad), so for smooth operation Linux actually wins in my case (battery life is roughly the same, CUDA works great on my Quadro card, I don't have switchable graphics, but I'm told they're a pain). Matlab is one of those wonderful cross platform pieces of software - works great for me in Solaris, OS X, Windows, or Linux.
I'm starting to think GNU is the problem with "GNU/Linux" these days.
this is what I hate about regex and compressed grammars.
it quickly becomes a write-only language.
blech!
I avoid them if I can. readable and supportable code is often more important. you may be more efficient with a regex but its also much easier to have subtle bugs creep thru. and fixing them by the non-author is always 'fun' ... ;(
--
"It is now safe to switch off your computer."
Often doing something without a regex takes more lines, and is less readable.....
"First they came for the slanderers and i said nothing."
Microsoft is not responsible for that. Laptop vendors test microsoft's products on their own equipment. Why don't you hold ms responsible when your laptop fails to sleep properly? Right, it's the hardware vendor's fault for lack of support and/or documentation.
I agree that there is something wrong with Linux on some Haswell parts. I am using Fedora 22 on a Thinkpad T440p with a quad core Haswell and it runs hot all the time and has terrible battery life, while it runs great on my wife's Thinkpad T440s with a dual-core Haswell. I've never had a Thinkpad with such terrible thermal performance in Linux, and that's over a history that includes X20, X22, X40, T40, T41, X200, T410s, T440s, and T440p models.
I think it actually has something to do with the HD4600 GPU and system P-states, as I've only managed to get idle power lower when I put the X server into unusable graphics modes like 16-bit color and panel self-refresh that induced horrible flicker. I've had to go back to normal settings and live with a palm-heater.
My T440p has a GeForce GT 730M with Optimus which turns out to be useless for me, as it isn't enough faster than the HD4600 graphics to matter for OpenGL and it isn't enough faster than the CPU SIMD to matter for my OpenCL codes, but that's partly because my problems are more bandwidth-limited than raw GFLOPS limited.
Using the 'optirun -b none clpeak' on the 730M gives me about 575 GFLOPS while just doing 'clpeak' with Intel's SIMD drivers on the quad-core CPU does 275 GFLOPS. At the office, I have an older Xeon and it only reaches 22 GFLOPS on the CPU while a GeForce GTX 760 gives me 2300 GFLOPS. In real OpenCL application performance, the quad-core Haswell and the GTX 760 often tie for first, followed by the older Xeon and the GT 730M which swap places by a larger margin depending on the code. I also run into total data size problems with the 1GB GT 730M and the 2GB GTX 760, so they have to use smaller sub-block decompositions than the SIMD code that runs out of main RAM.
Actually, if you just ran emacs over systemd, you'd not need either Linux or *BSD.
Just get a Thinkpad.
I run Ubuntu 14.04 on a Dell Precision M4400, and I'll admit the hibernation doesn't work 100% but ya know what? I have an SSD on the system and it coldboots in less time than it would take to come out of hibernation, so I really don't care.. For the VERY few Windows programs I need, I have a Windows 7 Virtualbox VM on the system.. After trying out Window 10 preview and seeing what a privacy nightmare it is, there ain't NO way thats ever gonna be on ANY machine *I* control....
THANK YOU, Edward Snowden!! Americans owe you a debt of gratitude (whether they know it or not..)
i knew the guy who wrote vim
I have never understood the obsession with hibernation, or why so many people regard it as a must-have to the point they're willing to use a crap excuse for an OS on account of that one thing.
Il n'y a pas de Planet B.
MATLAB is cross platform but if you have a lot of toolboxes you can start to run into errors under Linux with it.
http://stackoverflow.com/quest...
It has to do with static thread local storage and dynamically loading libraries. I don't know why the error does not occur with Windows ever but it seems to be within the design of glibc. There are ways to work around it but making sure some of the libraries you need the most are loaded first but then other stuff can just fail later.
MATLAB is easy to install and easy to use under Linux but with this bug that it seems there is no realistic way to fix it can be a pain in the ass sometimes.
Computer modeling for biotech drug manufacturing is HARD!
It's not too many (maybe too few---but it works with BASH's parser).
I can't figure out what the sed command is doing though, that part is tough.
Have you tried running it?
SJW n. One who posts facts.
I have 2 1TB 850 EVOs in my laptop and it does not boot as fast as it comes out of hibernation. From cold boot to full working desktop is probably 20 seconds and hibernate is 5 seconds.
Computer modeling for biotech drug manufacturing is HARD!
i just want SLI support for nvidia gpu on linux. you cant really game without either using SLI or a ridiculously expensive gpu. linux gaming wont overtake windows until the nvidia driver support is equivalent.
Hibernation means that I can have my laptop go to zero power usage but also restore the exact state it was at (with all programs running) as it was when I powered it off and it will return in about 5 seconds to full running again.
I can have virtual machines, IDE, profiler all setup and running and hibernate the computer for several hours and then use it again at another location.
Computer modeling for biotech drug manufacturing is HARD!
Of course. I figured out what it would do before running it (at least enough to make an educated guess).
Weirdly when I run it with the 'PS1=' at the front, it works fine. When I run it just as a normal command, it doesn't parse correctly.
"First they came for the slanderers and i said nothing."
Which, judging by the responses above, might also have terrible support. Yay?
Thank you for your willingness to present actual information and have a civilized discussion, something not so common on /.
Given your focus, obviously you've made an OS choice that best meets your needs. I am definitely a Linux bigot, but I recognize that other people have other requirements that may best be met in other ways.
As to the poster who used this as an opportunity to criticize your research efforts, I think we can both agree to just ignore ignorance when we see it. There may be issues with the "big pharma" business model, but there's no doubt that critical life-saving research takes place.
It is nice to talk with someone having a civil conversation also.
I find that if I stay to standard c++, MPI and OpenMP my code is just a recompile away from running on a Linux cluster. Most of my code ends up running on Linux clusters it is just the development on a Laptop that is easier under Windows.
In the end I suspect that nothing is ever 100% ideal. It is easier to do development under Linux for python by quite a bit compared to Windows You just end up using what works best given the resources you have at the time.
I pretty much only work on biotech development so protein, RNA and DNA based medications. The problem is that they are so hard to make and in many cases companies do know of drugs to cure some diseases and they are not brought to market because they can't figure out how to make them. When you are talking about making 10^23 molecules of something and each one has 10K atoms and each one has to have every single one of those connected correctly and an error rate of less than .001% it is basically at the limits of our technology. There are things that teams of scientists work on for years in a lab and they come up with only a few doses and attempts to build the same drug on a larger scale just turn out to be failures. Some drugs companies try to make for 5 years or more before finally giving up on them until the technology improves.
My goal is to use computer simulations to figure out how to manufacture the drugs. I like Linux a lot but in the end I care more about the drugs being available and will use whatever tool is the most appropriate at the time. That means Linux, Windows, free and proprietary software.
Computer modeling for biotech drug manufacturing is HARD!
I only work with biotech drugs. I do not do ANYTHING with small molecule drugs and those are the kinds that have all the nasty side effects.
Biotech drugs are things like Filgrastim https://en.wikipedia.org/wiki/...
This one drug has raised cancer survival rates for many types of cancers from 5% to 95% or so by allowing your immune system to be kept strong while undergoing chemotherapy. This drug is temporary only and once you are off chemotherapy as your bone marrow recovers the dosage is lowered until your body fully recovers and it is no longer needed.
There are also DNA and RNA based drugs and even gene therapies using editing techniques like CRISPR-CAS9. Editas just receiver a little over a hundred million to work on gene editing to correct the gene that causes Huntington's disease and also to work on upgrading tumor suppressor genes.
Most of these are all permanent cures and not treatments. Some are used as treatments short term until we are more confident of their safety since once you change someones DNA is becomes a much bigger problem if you screw up.
I agree that small molecule drugs have a massive number of side effects since the molecules are so non-specific. That is why in the USA at least most research is in biotech drugs. They are highly specific and they usually have almost no side effects since we are not making anything new. Pretty much all of biotech is just artificially creating a natural molecule that for some medical reason your body is not making at the time. It is not some new molecule that is foreign to your body.
Computer modeling for biotech drug manufacturing is HARD!
When you come out of hibernation, the computer has to read the entire contents of the ram back from the disk. So the more ram you have, the slower it is actually to come out of hibernation.
Sod off skids, you narcoleptic red-light running gerontophile.
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
I have 24G of ram and resuming from hibernation and having my apps all working again is MUCH faster than a cold boot to apps working.
Computer modeling for biotech drug manufacturing is HARD!
JWZ would tend to agree with you:
* http://regex.info/blog/2006-09...