Is it possible to set the device property cudaKernelExecTimeoutEnabled to 0 at runtime? Currently it is 1 because I'm also running an X-server on it. The problem is, that my threads time out.
On Linux, you can disable the timeout (about 6 to 7 seconds for me) by adding an option to your X11 configuration. On modern Linux systems, there is no X11 configuration any more, everything is autodetected, but you can create it manually:
mkdir -p /etc/X11/xorg.conf.d
touch /etc/X11/xorg.conf.d/20-nvidia.conf
Then add the following into this file:
Section "Device"
Identifier "My GPU"
Driver "nvidia"
Option "Interactive" "0"
EndSection
The key here is the "Interactive" option. This works for me even without X11 restart. It was obvious that this configuration is read and applied as soon as my OpenCL application is started.
"Identifier" can be an arbitrary value. "Driver" is "nvidia" in my case, since I'm using the proprietary Nvidia driver for Linux (part of the official Debian package repositories).
I am assuming you are on windows:
You need to disable the windows driver timeout:
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Watchdog\Display\DisableBugCheck
create a REG_DWORD and set it to 1.
On Linux, you need to use the nvidia-smi command to change properties of the device. But I think if you're running a display on the GPU, you might not be able to set that property.
Related
I'm Ronaldo, Newbie of Stackoverflow.
I'm having a problem, On my PC it's running Windows 10 Build 1803, Then I installed QEMU, but not only that, I installed the graphical version which is QtEmu.
Then I wanted to virtualize CentOS 3.1, with Disk 1, Disk 2 and Disk 3. But I don't know how to eject the disk in QtEmu, I tried it in PowerShell and I couldn't, not even in CMD, just the serial0, and the keyboard is not works, only on the PC that works. the Mouse is ok. But QEMU Monitor is out of QtEmu! Is there any way to install QEMU Monitor on Windows, with QtEmu?
Please help me!
QEMU can be started with option "-monitor" so you can send commands to it there. I never used QtEmu so not sure how it starts QEMU. If you figure out how, check out https://qemu.weilnetz.de/w64/2012/2012-08-24/qemu-doc.html#pcsys_005fmonitor.
So basically, you can use "change" command to mount and unmount a CD drive or other storage devices. For example, "change ide1-cd0 /tmp/file.iso" to mount an image file. The device names are not fixed across different VM configurations. "info block" can tell you what name you need to use.
sorry i just discovered QEMU, it's far now, but it can help people in the future.
I dislike QTEmu and others IHM, too much bugs currently (2022)..
With QEMU
default mode: you can directly have an access with the window, go to view and you can switch by clicking on "compat monitor 0" . You can also detach the monitor.
sdl: ctrl+alt 2.
Here:
Identify your devices by "info block"
As you can see here, i ran without anything (it's just to show).
To change/add content of your cd-rom, "Change "
here my device is "ide1-cd0"
Path can be your physical or virtual drive, a disk image, you can always swap according your needs.
As you can see now, i use "R:", a virtual device from a software like daemon tools (an alt).
I think it's better, before to swap, to eject from the guest... But perhaps
it depends of your OS. For the while i used it, the most of time, with win98x.
My laptop support hardware virtualization & enabled in bios & i am trying to compile qemu from source. So Is it possible to compile qemu in a way that i do not enter "-enable-kvm" flage from command line every time?
You could try passing configure --disable-tcg, which will disable the CPU emulation entirely (this only works for some target architectures like x86 at the moment). Then I think QEMU will default to KVM as the only available accelerator.
Alternatively if you rename your executable to something whose name ends with "kvm" then QEMU will default to the KVM accelerator rather than TCG. (I think a hardlink or maybe even a symlink to the binary will also work as well as simply renaming it, but haven't tested this.)
We're using xcodebuild command line to build our nightly jobs, specifying project, target, and configuration.
It seems like the build is done with "Parallelize build" set to "YES" by default, although we're not using schemes for the nightly builds.
This causes build errors (in detail: lipo tool tries to create a universal binary before both architectures complete building).
How can I disable the parallelization in the command line without having to create a scheme?
Ok, I've created a shared scheme (so I could later save it in the source control) and used it to define the parallelization state. That worked for me.
I'm learning how to build a basic OS kernel with https://intermezzos.github.io
I've create my .iso file and I'm at the point where I am runnning qemu-system-x86_64 -cdrom os.iso
When I press enter, QEMU runs a window with the following output:
Booting from Floppy...
Boot failed: could not read the boot disk
Booting from DVD/CD...
Boot failed: Could not read from CDROM (code 0004)
Booting from ROM...
iPXE (PCI 00:03.0) starting execution...ok
iPXE initializing devices...ok
iPXE 1.0.0+git-20131111.c3d1e78-2ubuntu1.1 -- Open Source Network Boot Firmware
-- http://ipxe.org
Features: HTTP HTTPS iSCSI DNS TFTP AoE bzImage ELF MBOOT PXE Menu
net0: 52:54:00:12:34:56 using 82549em on PCI00:03.0 (open)
[Link:up, TX:0 TXE:0 RX:0 RXE:01]
Configuring (net0 52:54:00:12:34:56)...ok
net0: 10.0.2.15/255.255.255.0 gw 10.0.2.2
Nothing to boot: No such file for directory (http://ipxe.org/2d03e13b)
No more network devices
No bootable device.
I went to the website listed in the output (http://ipxe.org/2d03e13b) and one of the tips is that I might Use the iPXE command line to perform DHCP manually, however when I press CTRL + B to access cli, I'm not able to do so.
Where do I look next to troubleshoot this problem of not being able to boot my .iso?
How do I make QEMU have access to keyboard input?
UPDATE
I don't know how, but I'm am now able to use CTRL + B to access iPXE command line.
This seems like a good place to start diagnosing my problem of not being able to boot my .iso.
What am I looking for?
UPDATE 2
Thanks to Peter Maydell's suggestion below, I've tested a known-good iso image (https://alpinelinux.org/), running qemu-system-x86_64 -cdrom alpine-3.4.3-x86_64.iso and it booted perfectly just as I expected.
I've rewritten my files from https://intermezzos.github.io to create a new iso image, this time copy and pasting the code from the repository, just in case I was previously inputting typos.
Still not booting. On to the next clue...
The first thing to do is to check whether this command line and ISO image work on a normal Linux host system. That will tell you whether the problem is (a) the Windows Subsystem for Linux not correctly implementing something QEMU relies on or (b) your ISO image actually not being a bootable CDROM.
You might also try booting a known-good ISO image such as one for a Linux distribution.
(The general principle here is to try to do diagnostic tests to split the space of "what might be the problem" into smaller sections and determine which side your problem is.)
Simple solution to the problem. Has to do with systems that use EFI to boot.
source
I needed to apt-get install grub-pc-bin and then rebuild image.
I have a Lenovo Y510P with integrated IntelHD 4th gen. VGA and a dedicated Nvidia GT755 VGA.
I run Fedora 21 and until recently had no problems. I never installed any proprietary drivers or special tools to deal with graphics, it just worked under fedora.
(I use a Windows partition to play games, everything there works fine too.)
Quite recently I acquired a second GT755 for my Ultrabay-Slot. Booting to Windows went without a problem, SLI activated like a charm and everything seemed fine. However, booting to fedora caused the CPU to rapidly heat up. I found out later that having both VGAs installed causes the bios to disable IntelHD, which apparently was the only VGA Fedora can run out of the box.
So my first try was to install VGA drivers and IntelHD-drivers in parallel.
(following this guide: https://ask.fedoraproject.org/en/question/36506/changing-default-driver-for-displaying-x/ )
After I found out about the "no IntelHD with SLI"-problem I tried out to get the SLI disabled and rely only on my Intel. There I made a huge mistake:
In the bios, I disabled the GT755 while booting with only one VGA. Forgetting about that, I later tried to install the Nvidia driver again, which could not find the implemented GT755. Believing this was due to some xorg.conf-code within /usr/share/X11/xorg.conf.d/20-intel.conf:
Section "Device"
Identifier "Intel"
Driver "intel"
EndSection
I simply plugged in the second GT755, which caused the bios to disable IntelHD.
Now I installed the Nvidia Driver (bumblebee-nvidia) and everything worked fine, xorg.conf was rewritten autmatically, I disabled nouveau in /etc/default/grub and recompild grub (with grub2-mkconfig -o grub.cfg), and rebooted.
Since then, I could not get to the login-screen. When I enter my LUKS-passwd, I see a yellow Message:
[ 0.093961] Ignoring BGRT: invalid status 0 (expected 1)
But if I remember correctly, this was always displayed, just in white.
I have tried reverting to my backups of xorg.conf, grub.cfg and /etc/default/grub and recompiling grub, nothing worked.
Everything was tested with both enabled and disabled internal GT755.
I tried to reinstall bumblebee-nvidia with enabled GT755, no difference.
IntelHD drivers are loaded at startup and appear to be correctly configured in all xorg related confs.
Calling startx causes the login-screen to flash for a moment, just to be instantly replased by an error-message:
(==) Using system config directory "/usr/share/X11/xorg.conf.d"
waiting for X server to shut down "sni-qt/[NUMBER, 4 digits]" WARN [TIMESTAMP] void StatusNotifierItemFactory::connectToSnw() Invalid interface to SNW_SERVICE bluedevil-monolithic: Fatal IO error: client killed(EE) Server terminated successfully (0). Closing log file.
At this point I would be glad to get my display back. Ideally however, I want to be able to keep the second VGA in it's slot and still use IntelHD, since I will never require high VGA performance under fedora.
What I need now is a possibility to revert my system back to it's old state.
With any luck I can actually install all required drivers afterwards and would write a little VGA-guide for the Y510P.
Apparently, I broke the displaymanager (sddm) somewhere somehow.
A friend helped me analysing the problem, but we were unable to find the exact cause.
We fixed the problem by replacing sddm with kdm.
the steps are:
install kdm
sudo yum install kdm
disable sddm:
sudo systemct disable sddm
reboot