Hot answers tagged

59

Short answer: This should be safe on well-designed hardware. Long answer: The GPU (and its software environment: drivers, OS, daemons) are designed to protect from overheating - the GPU should first turn the fans to a higher RPM, if that can't keep a safe temperature then the GPU throttles the workload (usually by reducing the clock frequency). This will ...


53

The NVIDIA headers were moved out of the FFmpeg codebase to a standalone repository in commit 27cbbbb. From the commit message: External headers are no longer welcome in the ffmpeg codebase because they increase the maintenance burden. However, in the NVidia case the vanilla headers need some modifications to be usable in ffmpeg therefore we still ...


15

I had the same problem for weeks and just managed to fix it! The solution was to set num_workers to 0 during creation of the DataLoader object.


14

Upon much digging this is currently possible but only with limited configurations, specific host OSes, and the use of enterprise software. Windows Server 2016 and above but with Hyper-V This is only possible with Windows Server as the host and Hypervisor. It is not available with windows 10 pro as the host as stated in the question above. Ubuntu 18.04 and ...


11

In layman terms, CUDA Cores and Stream processors are exactly the same. The question is similar to asking whether Intel and AMD CPUs are the same or not. The difference in names is mostly commercial branding. Both NVIDIA and ATI/AMD cards are multi-core units excelling in executing parallel programs. The difference is that AMD stream processors are smaller,...


11

A house fire is extremely unlikely, but the lifespan of the card may be reduced. Long-term overheating of the GPU chip probably won't start a fire. The chip may deteriorate and start misbehaving or die completely, but silicon chips aren't too flammable. Bad things usually happen when electrolytic capacitors fail and blow up, but these won't be subject to ...


10

This seems possible under Linux and there are detailed instructions for doing that. Your CPU supports the required VT-d capability, and hopefully your (unspecified) motherboard as well. You also have two graphical adapters, so can afford to give up one to the VM (remembering that your other GPU is rather limited). There are various articles on the subject ...


7

Firstly, you are getting your terms mixed up. CUDA is an NVIDIA technology for programming their GPU (and other things, but that's the simplest description). Microsoft's RDP uses a it's own graphics driver which converts the rendered screen into network packets to send to the client. This is the core of how RDP works and you cannot change it. On the ...


7

Yes, the card is likely to wear out sooner if it is under constant load. At small geometries, Electromigration is a significant source of device failures, and devices will typically be designed with a specific target lifetime in mind. This might be generous for typical operation (e.g. 5 years continuous operation), but might not assume 100% maximum operating ...


6

You should just use your compute capability from the page you linked to. For example, if your compute capability is 6.1 us sm_61 and compute_61. SM stands for "streaming multiprocessor". The arguments are set in this confusing looking way because they are used as arguments for nvcc where the compute_XX sets the architecture for a virtual (intermediate) ...


5

Modern graphics engines are based on shaders. Shaders are programs that run on graphics hardware to produce geometry (the scene), images (the rendered scene), and then pixel based post-effects. From the Wikipedia article on shaders: Shaders are simple programs that describe the traits of either a vertex or a pixel. Vertex shaders describe the traits (...


5

You might be trying to install the "patch" instead of the main installer. Is your file about 70 MB size instead of the more expected 1.7 GB size? If yes, then you are probably installing the "patch" first before the main installer. I also tried installing the "patch" first and got an error and then I was confused. Then I realized I had the wrong filename ...


5

I encountered a similar error but it turns out, it was caused by missing tool package config. On Ubuntu 16.04 you can run apt-get install pkgconf and check whether the missing package is really missed or not: pkgconf --list-all | grep package-name


4

I started a bounty for this question, but after a while I figured out how to solve it, so I might as well post it here as an answer. It looks like nvcc is asking for /usr/lib/R/include, which does not exist on ubuntu 12.04. Here is the solution (tested on ubuntu 12.04) sudo aptitude install r-base-dev cd /usr/lib/R sudo ln -s /usr/share/R/include . Now ...


4

As OP stated Tegra K1 indeed does have CUDA (CUDA 6.5) support as stated in CodeWorks for Android 1R4 release notes. However K1 isn't the only CUDA capable android device. According to release notes; TK1 Reference Device (Ardbeg) Google Project Tango Tablet MiPad SHIELD Tablet 8 SHIELD Android TV support CUDA as well. There seems to be versional ...


4

Not possible. It's only via NVENC. See https://trac.ffmpeg.org/wiki/HWAccelIntro#CUDACUVIDNVDEC


4

I've seen this error message for three different reasons, with different solutions: 1. You have cache issues I regularly work around this error by shutting down my python process, removing the ~/.nv directory (on linux, rm -rf ~/.nv), and restarting the Python process. I don't exactly know why this works. It's probably at least partly related to the second ...


4

If your cooling system works OK, and your hardware is of any kind of even vaguely modern design that includes on-chip temperature monitoring and thermal throttling/suspend/shutdown, then it's entirely safe. It can't overheat so long as the cooler keeps running, and if that fails, the chips will throttle back until they're no longer producing more heat than ...


4

Now there is one. You can install a WSL 2, that has CUDA and with it TensorFlow support: https://docs.microsoft.com/de-de/windows/win32/direct3d12/gpu-cuda-in-wsl Then you can install PyCharm and configure it to to work with WSL: https://www.jetbrains.com/help/pycharm/using-wsl-as-a-remote-interpreter.html#


3

Both cards have their pros and cons: The GTX is much better at performance: http://gpu.userbenchmark.com/Compare/Nvidia-Quadro-K4200-vs-Nvidia-GTX-980/2838vs2576 The GTX 980 is based on the newer second-generation Maxwell architecture, which means that it will more likely support newer technologies than the Quadro K4200 (which is still based on the ...


3

Everything in the above answer is correct except for "This is the core of how RDP works and you cannot change it". Never say never. There are two ways to utilize a better graphics driver over RDP without 3rd party slow laggy software and without modifying any windows DLLs. (super hard) Install windows server 2012 r2 on a physical host. Then use Hyper V ...


3

The reason we are still using CPUs is that both CPUs and GPUs have their unique advantages. See my following paper, accepted in ACM Computing Surveys 2015, which provides conclusive and comprehensive discussion on moving away from 'CPU vs GPU debate' to 'CPU-GPU collaborative computing'. A Survey of CPU-GPU Heterogeneous Computing Techniques


3

No, there is no complete implementation of the JVM that runs on the GPU (at least I've never heard of one). This would not make a lot of sense: Usually only certain tasks that are suitable for the GPU are moved there, while the CPU does the rest of the work, so it does not make sense to have the whole JVM running on the GPU. That said, there are multiple ...


3

FINALLY, I found this on the official documentation, The Runfile installation asks where you wish to install the Toolkit and the Samples during an interactive install. If installing using a non-interactive install, you can use the --toolkitpath and --samplespath parameters to change the install location: ./runfile.run --silent \ --toolkit --toolkitpath=/...


3

Before you do anything else, write these two commands for escaping from a login loop on a piece of paper, so if you get stuck in a login loop when booting you will be able to do something about it. sudo chown $(whoami):$(whoami) .Xauthority sudo dpkg-reconfigure lightdm Also write down the link to this answer which has more detailed information about ...


3

Press Ctrl+Alt+F3 to switch to a different terminal Login Uninstall nvidia* packages sudo apt-get purge nvidia-* After that, install nvidia-346 sudo apt-get install nvidia-346 reboot


3

If the graphics card is not the principal one in your system, you can think about using PCI passthrough to the VM, so the guest OS can access it directly. Unfortunately it seems the VBox does not support this feature (read here), and I don't know if it is supported on other desktop virtualization software. Anyway if your graphics card is also used by the ...


3

EDIT: Before you try the long guide and install everything again, you might solve the error "(DLL) initialization routine failed. Error loading caffe2_detectron_ops_gpu.dll" by downgrading from torch = 1.7.1 to torch=1.6.0, according to this (without having tested it). This is a selection of guides that I used. https://github.com/pytorch/pytorch#...


2

gpus are good stream processors. you can think of stream processing as multiplying a long array of numbers sequentially. cpus also have stream processing capabilities (it's called SIMD extensions) but you can't implement all programming logic as stream processing, and compilers have the option to create btyecode which meakes use of simd instructions whenever ...


Only top voted, non community-wiki answers of a minimum length are eligible