AMD Contributes 8.5x More Code To The Linux Kernel Than NVIDIA, But Intel Still Leads

[phoronix]

Phoronix: AMD Contributes 8.5x More Code To The Linux Kernel Than NVIDIA, But Intel Still Leads

Given all the new hardware enablement work going into the Linux kernel recently, I was curious how the code contributions were stacking up by some of the leading hardware vendors... Here are those interesting numbers...

http://www.phoronix.com/scan.php?page=news_item&px=AMD-NVIDIA-Intel-Kernel-Contrib

[clavko]

Lines of code is hardly a metric you should be proud of. If anything, with mature codebase, you should be removing lines, rather than adding them. In my last company we wrote a 5k LoC project that replaced 45k LoC legacy solution, heavy templated and consolidated. And still we removed a couple of hundreds of lines after a while through refactoring.

[Brisse]

So besides NVIDIA having contributed at least ~8.5x as much code to the kernel as NVIDIA

*AMD*

[AsuMagic]

*AMD*

Still technically true: 0 * 8.5 = 0

[DanL]

This article is hilarious. From trying to use lines of code as a metric to italicizing "open-source" for Nvidia shaming, to telling us what we already know, this is non-news news.

[Raven3x7]

You can't really compare AMD's and Intel's contributions directly like that, or nVidia really. AMD only really does GPUs and CPU's in-house these days, while Intel also does chipsets, network adapters, FPGAs and a ton of other stuff. What is true though is that nVidia contributes very little aside from Tegra, but we already knew that and it fits in well with their corporate culture having a penchant for building walled gardens.

[perpetually high]

Everyone needs to calm their titties. We know KLOC isn't a metric and Michael is a developer himself. And please don't compare your cutesy work projects to the linux kernel with MLOC and thousands of developers across the world.

AMD has their open stack an is far ahead of NVIDIA in that regard. However, NVIDIA does seldomly contribute to nouveau here and there as far as I know. From a 2014 slashdot article: Torvalds publicly gave a thumbs-up to Nvidia for contributing basic support for the recently released Nvidia K1 processor to Nouveau; something that was totally unexpected but received with open arms. 'Hey, this time I'm raising a thumb for nvidia. Good times,' said Linus."

We should be putting more pressure on all companies (CPU/GPU/hardware) companies to continue providing open-source software and drivers, understanding its greater good is better than the sum of its parts. rather than chiming in here with KLOC isn't a metric. (duh)

And besides, of course Intel is winning, most of those commits are Spectre/Meltdown related. I'll see myself out.

[Guest]

Lines of code is hardly a metric you should be proud of. If anything, with mature codebase, you should be removing lines, rather than adding them. In my last company we wrote a 5k LoC project that replaced 45k LoC legacy solution, heavy templated and consolidated. And still we removed a couple of hundreds of lines after a while through refactoring.

True, I once saw 2k LoC for DAO(s), and reduced it to like 150 LoC. Problem was, that original author created for each category a separate table, with exactly same structure, and separate DAO for each table with insert/update/select functions, instead of parametrized functions... Yeah, LoC is neever good metric. But, amount of features, performance, robustness, and stability is good metric,... Also, less LoC doesn't always lead to better quality, as I saw very complex and hardly readable/understandable code using weird polymorphism approaches, just to reduce LoC,...

[bridgman]

Lines of code is hardly a metric you should be proud of. If anything, with mature codebase, you should be removing lines, rather than adding them. In my last company we wrote a 5k LoC project that replaced 45k LoC legacy solution, heavy templated and consolidated. And still we removed a couple of hundreds of lines after a while through refactoring.

A lot of the lines are from register headers for new GPUs.

Last time I looked the kernel driver was about 250K lines of code and 1,250K lines of headers, minus white space and comments.

[PublicNuisance]

For all those trying to save Nvidia by saying that lines of code isn't a good metric why aren't any of you listing the quality of the contributions Nvidia has given. I'll be waiting.