Comments on: A JWRD desktop aiming for open graphics

By: Robinson Dorion

Robinson Dorion — Sat, 09 Mar 2024 16:39:28 +0000

Myeah, poking around myself, looks like pretty much all Corsair's stuff is glass now.

By: Jacob Welsh

Jacob Welsh — Mon, 05 Feb 2024 04:41:00 +0000

Apparently steel is just as fickle as silicon these days: even the case listed here is "Currently unavailable. We don't know when or if this item will be back in stock." on amazon (and that's after bypassing its own defective search that spams six "results" not matching the given terms, four not even matching the brand).

By: Jacob Welsh

Jacob Welsh — Fri, 01 Oct 2021 18:34:58 +0000

@Luc Verhaegen: sorry we had this languishing in the spam queue and then the otherwise neglected items bucket for quite so long.

I actually went and added full native VGA display init for the unichrome on the VIA K8M890 after RadeonHD got cancelled

I hadn't heard of unichrome (not to be confused with monochrome, apparently!) but found it was from S3 Graphics which was not related to ATI except perhaps as a competitor. So I infer the RadeonHD connection is only incidental, in the sense of unichrome being what you switched to working on once the RadeonHD revenue was cut; have I got that right?

Otherwise what I get from the article is that one family of VIA boards and one particular Asus board gained fully-free text mode support, with more conceivably to follow. That might make a good option for low-end or server boxes but not for 3D graphics or even basic x11.

And the amount of code needed to get a discrete card with ATOMBIOS going is also pretty limited, it should be contained in ASICInit, which we used to initialize non-primary cards just fine.

I found an rhdAtomASICInit in xf86-video-radeonhd which serves as wrapper to an ASIC_Init which I take to be a bytecode routine provided by the GPU's ROM. So what's the implication - that the amount of such bytecode required is less than previously believed? That this makes radeonhd interesting despite supporting only hardware that's now out of production?

By: Luc Verhaegen

Luc Verhaegen — Thu, 10 Jun 2021 21:28:52 +0000

I actually went and added full native VGA display init for the unichrome on the VIA K8M890 after RadeonHD got cancelled (and the world experienced the result of the 2008 financial crash): https://libv.livejournal.com/19432.html
And the amount of code needed to get a discrete card with ATOMBIOS going is also pretty limited, it should be contained in ASICInit, which we used to initialize non-primary cards just fine.

By: A Gentoo install on JWRD desktop « Dorion Mode

A Gentoo install on JWRD desktop « Dorion Mode — Thu, 24 Dec 2020 18:14:15 +0000

[...] managed to install Gentoo on my JWRD desktop with the software in the table below. Below the table I've documented the steps I took and the [...]

By: #jwrd Logs for Nov 2020 « Fixpoint

#jwrd Logs for Nov 2020 « Fixpoint — Tue, 01 Dec 2020 19:14:22 +0000

[...] [...]

By: Jacob Welsh

Jacob Welsh — Sun, 01 Nov 2020 21:37:01 +0000

For completeness or for other graphics noobs (though isn't it funny how none of this actually has anything to do with graphics? welcome to hardware-accelerated bureaucracy I guess) the alternatives for 3D graphics hardware - excluding the embedded stuff found on phones and such - are Intel and nVidia. AMD got into the game by acquisition of ATI. Intel is the low end with integrated chips only (on the motherboard, sharing system memory with CPU) but supposedly has strong open source driver support. Not sure of the firmware situation - I know they play the "'open driver' with closed firmware" game on their WiFi adapters at least - and of course it won't help if one has set the evil bit on Intel. nVidia traditionally had the best support for Linux through their fully-closed drivers but there's a reverse-engineered "nouveau" driver that's been in the works for ages and might be good enough in some cases. So in short, from what we know so far AMD is the least-bad option especially if one wants decent performance.