Slashdot Mirror


Linux Kernel 4.10 Officially Released With Virtual GPU Support (softpedia.com)

"Linus Torvalds announced today the general availability of the Linux 4.10 kernel series, which add a great number of improvements, new security features, and support for the newest hardware components," writes Softpedia. prisoninmate quotes their report: Linux kernel 4.10 has been in development for the past seven weeks, during which it received a total of seven Release Candidate snapshots that implemented all the changes that you'll soon be able to enjoy on your favorite Linux-based operating system... Prominent new features include virtual GPU (Graphics Processing Unit) support, new "perf c2c" tool that can be used for analysis of cacheline contention on NUMA systems, support for the L2/L3 caches of Intel processors (Intel Cache Allocation Technology), eBPF hooks for cgroups, hybrid block polling, and better writeback management. A new "perf sched timehist" feature has been added in Linux kernel 4.10 to provide detailed history of task scheduling, and there's experimental writeback cache and FAILFAST support for MD RAID5... Ubuntu 17.04 (Zesty Zapus) could be the first stable OS to ship with Linux 4.10.
It required 13,000 commits, plus over 1,200 merges, Linus wrote in the announcement, adding "On the whole, 4.10 didn't end up as small as it initially looked."

16 of 90 comments (clear)

  1. Does this mean... by Zontar+The+Mindless · · Score: 4, Interesting

    ...that we'll finally get a Nouveau driver that isn't a crash-prone piece of crap?

    --
    Il n'y a pas de Planet B.
    1. Re:Does this mean... by dbIII · · Score: 3, Informative

      Same with nvidia's own driver. Every time I've seen a machine where it looked like the nvidia driver was doing something wrong it's been a dead or dying fan on the card.
      The nouveau driver is great for what it does, but if someone is going to be using google earth you need the one from nvidia. Both are rock solid for what they do IMHO (and the fans on video cards are frequently shit).

    2. Re:Does this mean... by kuzb · · Score: 2

      It's stable enough, it just doesn't hold a candle to the binary drivers when it comes down to features and performance.

      Personally I prefer the driver which gets the job done, and while Nouveau is a solid effort, it just doesn't.

      --
      BeauHD. Worst editor since kdawson.
    3. Re:Does this mean... by Zontar+The+Mindless · · Score: 3, Interesting

      Your experience is the opposite of mine, then, which is that the *only* safe way to run Nouveau on a machine with an Nvidia card is to add nomodeset or acpi=OFF to GRUB_CMDLINE_LINUX_DEFAULT, or get used to random lockups. I've had this issue on 3 different machines (2 desktops and 1 laptop).

      --
      Il n'y a pas de Planet B.
    4. Re:Does this mean... by Gravis+Zero · · Score: 4, Funny

      Does this mean that we'll finally get a Nouveau driver that isn't a crash-prone piece of crap?

      Sure, it's virtually yours! ;)

      --
      Anons need not reply. Questions end with a question mark.
  2. Re:Backported to 2.6? by Anonymous Coward · · Score: 2, Interesting

    Well, even the newest kernels (I can vouch for 4.9.x at least) have no systemd requirement. I run that and am still happily systemd-free.

    The bigger question I have is re. virtual GPU: would it finally be possible to use Wayland in a QEMU-based VM? It would be nice to get see how that works--but still I wait for Wayland to have some ability to work across networks.

  3. vGPU seems cool by Kjella · · Score: 5, Interesting

    Looks like you can get near-native performance even though you're sharing hardware. With this maybe instead of a dual boot PC you can have a dual VM PC, one runs Linux and the other Windows and both at near native performance and you don't have to dedicate a graphics card. That sounds like a real gateway drug, use Linux for the desktop and the games that run on it but be able to switch to your Wintendo and play that one must-have game your friends want. That said right now it looks like an an Intel tech, did anyone see anything about AMD/nVidia support? Because sharing that Intel iGPU wasn't really what I'm looking for....

    --
    Live today, because you never know what tomorrow brings
    1. Re: vGPU seems cool by dknj · · Score: 4, Informative

      there is definitely support for nvidia

      Vgpu seems very very cool. Now how can we turn this into something commercially viable?

      -dk

    2. Re:vGPU seems cool by fuzzyfuzzyfungus · · Score: 5, Informative

      My understanding is that it is more extensive: PCI(mostly 'e' these days) passthrough allows you to assign a physical device to a VM; but the device can't be shared: if a given piece of hardware is being passed through to one of the guests, none of the other guests or the host OS can use it.

      This 'virtual GPU' stuff is supposed to make allocating GPU resources between VMs closer to how it is with CPU time or memory, where all the guests and the host can't exceed the capabilities of the machine they are running on; but they can all have access, with relatively modest overhead, to the same device.

      I don't know if things work as pleasantly as desired yet; but in principle it should be a lot more convenient than full device passthrough. Especially in cases where you might be interested in the GPU for its computational capabilities, video transcoder, etc.

    3. Re:vGPU seems cool by Poeli · · Score: 2

      AMD has something similar with their FirePro S series. The difference is those cards implement the SR-IOV standard to 'split' the GPU into 16 parts (at most). Unfortunately, the kernel driver is still not part of the kernel. There are patches floating around so it's probably for 4.11. The primary use case for this is virtualization. You can give several vm's a slice of the GPU and deliver near bare-metal performance. But it may be of interested for gamers too: the can share their GPU with a windows vm to play games under Linux. No hassling with an additional GPU or monitor. The intel (and nVidia) solution look interesting as it should work on all recent hardware while AMD only adds it to their S series. The advantage of SR-IOV is that it is a proven hardware solution.

  4. Re: Backported to 2.6? by corychristison · · Score: 4, Informative

    Running 4.9 on 4 physical machines in my home. And also running 4.9 on over a dozen VMs in a datacenter without systemd.

    There are a few distributions that don't push it down your throat. There are even a few others that offer (optional) alternative kernels and init systems.

    Personally I use funtoo.

    Take a look at www.without-systemd.org for more.

  5. Virtual gpu? by Kohath · · Score: 3, Interesting

    Can someone link to a concise discussion of what this does and some use cases? Thanks.

  6. Why did you write that? by dbIII · · Score: 4, Informative

    SystemD is in userspace. There are no dependencies for SystemD in the kernel.
    Now I hate SystemD far more than most due to it causing actual failures for me in production, but spreading disinformation about it is dishonest.

  7. Version 10 by hackwrench · · Score: 2

    Well, now we are on Windows version 10. At this rate we'll be eventually at Windows version . 000...0001

  8. Re:Does virtual GPU support mean that I'll be able by Anonymous Coward · · Score: 3, Informative

    No, but it would help you run Windows CAD packages inside a Windows VM and get near native GPU performance both in the Windows VM and Linux host. vGPU lets you share a single compatible GPU across multiple VMs running on the same system. This way each VM can benefit from full hardware accelerated 3D rendering as well as access to OpenCL and/or CUDA even though only a single physical GPU is present in the system.

    Wine wouldn't benefit because it already has direct access to the GPU as long as it's associated with the host system, it's issues are caused by imperfect compatibility between how Windows does something, and how WINE implemented it. For example, Wine translates Direct3D calls into OpenGL calls which can then be run natively in Linux, but if the translation isn't perfect, or a certain Direct3D call hasn't been implemented in WINE yet, then the program is going to behave differently than it should.

  9. ZFS? by Bodhammer · · Score: 2

    How about native ZFS support?

    --
    "I say we take off, nuke the site from orbit. It's the only way to be sure."