Pay For Faster Linux Kernel Performance? There's Patches For That

[alpha_one_x86]

Are you saying that the people working on closed source code don't deserve to pay them?

No, I say it's normal to pay open source work (as same it's normal to pay no source source work).

[dext]

I'm a native Polish speaker.

I've spent almost 10 minutes on the site, and didn't find anything interesting. Looks like a scam to me. Just prebuild kernels, with flags for certain CPUs. Also Noop is used by default for SSD drives. Can't find anything else. It's either a very bad website, or a strange scam.

[V10lator]

The hell are you smoking?

GPL: If you distribute binary, you must give source code on request to the ones that also have access to the binary (so your customers and nobody else).

Fixed for you.

ext73: Do you have kernels optimized for AMD Athlon II (X3 if that matters) and if so: Is it only a optimized config or also patched?

[przemoli]

1) Its set of families of kernels, each specializing in given cpu arch (x86 only though!), with extra one for server for docker host
2) Its also modified version of kernel component of Nvidia and AMD (duno if that is code or compilation flags only)
3) And some BIOS settings, and some grub settings, etc.
4) Schedulers and other important things are also used for maximizing responsiveness, etc.

We see it all the time in floss world.

Like Fedora disabling Intel work over some IP worries. Some distros enabling gallium-nine already for performance, etc. Just config patches and or compiler flags.

And test resulsts for budget HPC (200$) are clearly favourable.

(Also some Thinkpad test will follow)

[przemoli]

@Michael
Author of those tests asked for requests for follow up (on Thinkpad), so if You want something I can forward it
But he also asked about any tests that could stress stability of system, or data integrity. Is there anything like that for PTS ootb?

[Delgarde]

GPL: If you distribute binary, you must give source code on request.

Close. If you distribute binaries, you must make the source code available on request to the people you provided the binaries to - you don't have to supply it to any random individual who happens to ask for it.

So it's perfectly legit to create private branches for your own use, and it's perfectly legit to create your own patched binaries and charge for access to them. The catch is that because you're required to provide the source for those patches to the users under the GPL, you can't then control what they do with the source - in particular, you can't prevent them from re-distributing it more widely, or submitting it upstream.

[Guest]

I'm a native Polish speaker.

I've spent almost 10 minutes on the site, and didn't find anything interesting. Looks like a scam to me. Just prebuild kernels, with flags for certain CPUs. Also Noop is used by default for SSD drives. Can't find anything else. It's either a very bad website, or a strange scam.

It's not a scam to me. They just build some kernels and want to make money on it - I don't see anything bad here.

Phoronix gave them great advertisement for free.

[Tares]

It's not a scam to me. They just build some kernels and want to make money on it - I don't see anything bad here.

Phoronix gave them great advertisement for free.

I have read the article ealier today (before it was posted here) and my first thought was that this is perfect news for phoronix :-)

[alpha_one_x86]

When the official kernel will support -march=native as CONFIG?

[Guest]

When the official kernel will support -march=native as CONFIG?

When someone will submit a patch, push it through review and prove that it works fine.

This option will make some bugs harder to repeat, because the same config can be compiled a different way.

BTW. gcc 3.2 doesn't support this option.