AMD Launches GPUOpen Today

[pal666]

Very good for the community but I doubt if that's profitable.

it is more profitable than same thing but closed source

[pal666]

And, this is why I love AMD.

it was pretty clear from four previous posts

[pal666]

AMD users, hope that at no point in releasing Vulkan as open source AMD's legal department will be involved, you may wait years for that.

so what? it will use upstream kernel and it will not conflict with mesa. you could wait for opening or you could use it immediately if there will be games on launch

[Adarion]

Some nice slides here; also Linux-related

http://wccftech.com/amd-launches-gpu-open-cgi-gaming-professional-compute/

("You are not authorized to create or remove attachments." Okay, then no pictures. )


Congrats to the team. And don't take things seriously here, I mean, some people obviously need to talk good things bad (maybe narcissistic problems?). I wonder how somebody can still yell bloody murder at all this. "no better than nvidia..." some people must be living under a rock or be wearing blinders not to see all the progress in the last 7 - 8 years.

[Sethox]

Some nice slides here; also Linux-related

http://wccftech.com/amd-launches-gpu-open-cgi-gaming-professional-compute/

("You are not authorized to create or remove attachments." Okay, then no pictures. )


Congrats to the team. And don't take things seriously here, I mean, some people obviously need to talk good things bad (maybe narcissistic problems?). I wonder how somebody can still yell bloody murder at all this. "no better than nvidia..." some people must be living under a rock or be wearing blinders not to see all the progress in the last 7 - 8 years.

Yeah, that attitude always struck me a little bit odd. I started to show up in this place (phoronix.com) right before Mantle got announced for the public and I only read negative things, just some positive arguments around. It's like just because AMD is a big business it's still evil no matter what, you can blame them for what happened in the past but at least there are some step forward that has improved.. I mean no one ever said that AMD is the "BEST company in the world".

[M@yeulC]

M@yeulC, I'm seeing wierd things when I reply to your posts as well. Have replied twice and both posts have disappeared (not "unapproved", just vanished completely).

Maybe the forum doesn't like your username ?

EDIT - hey, my previous posts appeared. Guess they went into moderation queue but didn't get the usual yellow background "unapproved" copy of my submitted posts.

<skipped the XKCD>

Thanks a lot for your replies, it really seems interesting. Well, then, I might take a look around about microcode. I must say that to me, microcode (regardless of the targeted system) is often a "blob", with it sources rarely released, and frequently buggy. But I guess it's very close to embedded systems, with a structure completely tied to the hardware, and designed to take as little resources at possible. I have been tempted multiple times to jump on the mesa train, but was always constrained by time.
Anyway, thanks a lot for sharing your experience About other people's experience; I've heard that the reverse engineering of the DPM for nouveau was done with an oscilloscope, but I guess it's just different with documentation, and it might vary from one task to another.

I've had a lot of issues with the forum too (I just deleted 4 post in a row, the forum just wouldn't display my question when I submitted it). I wish the messages about the moderation queue would be more clear (or reliable. Or just deterministically behaving).
I am not sure about the username. I guess atomsymbol's one was worse since it must have required some utf-8 handling (and even then, a well-designed forum should be fine). The most painful for him must have been typing his username (U269B, was that it?). Mine is generally OK even on ISO encoding. I think it's more like a problem with the forums themselves, or Michael setting very strict anti-spam policies.

[bridgman]

I must say that to me, microcode (regardless of the targeted system) is often a "blob", with it sources rarely released, and frequently buggy.

Yeah, fair point. It is normally considered part of hardware design, basically a way of making complex logic manageable and maintainable.

The bugginess comes from being written by hardware devs, at least that's what the software devs always say

[theriddick]

I'll be happy with AMD once they finally release competitive Linux drivers that are on par with Nvidia's; oh and resolve the heat problems many of their high end cards suffer from (maybe 14nm will do that for them?)

[bridgman]

I hadn't heard about heat problems with the high end cards, can you point me to a reference ?

Or are you just saying "this generation the AMD cards use more power than the NVidia cards but temperatures are OK" ?

[theriddick]

LMAO, well there is the 390x, mine blackscreens if I don't ramp up the fan. Read online, very very common. Then there is the fact the Fury X card requires a watercooler as stock. You make me laugh not knowing about the problems!