Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

CUDA 11.0

docs.nvidia.com

Post reply on HN
Print view View original

21–30 of 67 posts

Re: CUDA 11.0

#21

Post by jabl »

jjoonathan wrote: post #12
usmannk wrote: post #3

I noticed CUDA 11.0 was almost ready for release last week when I went to install CUDA and the default download page linked to the 11.0 Release Candidate. The 10.1 and 10.2 links were buried behind a link off to the side labeled "legacy". The thing is, no library you use is going to be supporting the CUDA 11.0 RC, that's ridiculous. For example, Pytorch stable is on 10.2 and Tensorflow only goes up to 10.1. This is g…

Yeah, and the CUDA 10.0 official Visual Studio demo project build was broken for... looks like a year, at least, because they didn't want to populate the toolkit path. NVidia, you're better than this. https://forums.developer.nvidia.com/t/the-cuda-toolkit-v10-0... > The Conda dependency manager has made this a lot easier Yeah but conda is "Let's do dependency management with a SAT solver, it'll be great!" On a good d…

My biggest gripe with the conda depencency manager is that it doesn't keep track of which packages own which files, and if multiple packages own the same file the last one to be installed will happily scribble over whatever was there before. With hilarious results, of course.

This means that keeping a conda installation up to date is often very tricky, when upgrading you frequently have to uninstall and reinstall some packages.

It works better if you start from scratch with a requirements.yml file.

Re: CUDA 11.0

#22

Post by dijksterhuis »

bbatsell wrote: post #13
usmannk wrote: post #3

I noticed CUDA 11.0 was almost ready for release last week when I went to install CUDA and the default download page linked to the 11.0 Release Candidate. The 10.1 and 10.2 links were buried behind a link off to the side labeled "legacy". The thing is, no library you use is going to be supporting the CUDA 11.0 RC, that's ridiculous. For example, Pytorch stable is on 10.2 and Tensorflow only goes up to 10.1. This is g…

I have to use containers with nvidia-docker because NVIDIA so consistently and relentlessly breaks things without so much as a glance at backward compatibility.

I moved our Deep Learning servers over to Docker images + JupyterHub DockerSpawners recently because maintaining all the various version dependencies between frameworks was an absolute PITA.

Images are publicly available here in case anyone else needs something similar: https://hub.docker.com/u/uodcvip

Re: CUDA 11.0

#23

Post by etaioinshrdlu »

bbatsell wrote: post #13
usmannk wrote: post #3

I noticed CUDA 11.0 was almost ready for release last week when I went to install CUDA and the default download page linked to the 11.0 Release Candidate. The 10.1 and 10.2 links were buried behind a link off to the side labeled "legacy". The thing is, no library you use is going to be supporting the CUDA 11.0 RC, that's ridiculous. For example, Pytorch stable is on 10.2 and Tensorflow only goes up to 10.1. This is g…

I have to use containers with nvidia-docker because NVIDIA so consistently and relentlessly breaks things without so much as a glance at backward compatibility.

The annoying thing is that nvidia-docker is still not great. You still have to deal with the driver installed outside the container, and it makes a big difference.

Furthermore it seems like even the CUDA runtime is typically not installed in the container, but rather injected in by the nvidia-docker container runtime.

It is not fun to deal with.

Re: CUDA 11.0

#24

Post by cjhanks »

Does anyone understand why such minor upgrades resulted in a major version bump? Is this some sort of stability check point? Or some other versioning convention?

Re: CUDA 11.0

#25

Post by flx42_ »

ziddoap wrote: post #8

>cuFFT now accepts __nv_bfloat16 input and output data type for power-of-two sizes with single precision computations within the kernels. This exact sentence is listed both under "New Feature" and "Known Issues". I'm not super familiar with CUDA stuff, but, it can't be both right?

Thanks, I have reported it internally and it is now fixed.

Re: CUDA 11.0

#26

Post by ibn-python »

cjhanks wrote: post #24

Does anyone understand why such minor upgrades resulted in a major version bump? Is this some sort of stability check point? Or some other versioning convention?

Usually for an API it indicates a breaking change. In this case the removal of some functions which might require refactoring on the consumers end.

Re: CUDA 11.0

#27

Post by gowld »

ibn-python wrote: post #26
cjhanks wrote: post #24

Does anyone understand why such minor upgrades resulted in a major version bump? Is this some sort of stability check point? Or some other versioning convention?

Usually for an API it indicates a breaking change. In this case the removal of some functions which might require refactoring on the consumers end.

And to keep users on a hardware upgrade treadmill.

Re: CUDA 11.0

#28

Post by TheGuyWhoCodes »

bbatsell wrote: post #13
usmannk wrote: post #3

I noticed CUDA 11.0 was almost ready for release last week when I went to install CUDA and the default download page linked to the 11.0 Release Candidate. The 10.1 and 10.2 links were buried behind a link off to the side labeled "legacy". The thing is, no library you use is going to be supporting the CUDA 11.0 RC, that's ridiculous. For example, Pytorch stable is on 10.2 and Tensorflow only goes up to 10.1. This is g…

I have to use containers with nvidia-docker because NVIDIA so consistently and relentlessly breaks things without so much as a glance at backward compatibility.

I'm never sure of the relation between the driver, nvidia-docker and the container with a specific cuda version.

Last time I tried it the cuda inside the container tough it was using some old driver version while a much newer version was installed on the host. So I had to manual install the older version, not sure where the issue was but maybe it was because I was using the deprecated nvidia-docker version 2 which is still needed to pass gpu resources to containers run inside kubernetes.

Re: CUDA 11.0

#29

Post by bigdict »

cjhanks wrote: post #24

Does anyone understand why such minor upgrades resulted in a major version bump? Is this some sort of stability check point? Or some other versioning convention?

I think the page lists the changes since 11.0 RC, not the previous major version.
Post reply on HN