Happy Holidays: AMD Finally Pushing Out Open-Source Vulkan Driver

[darkbasic]

It does get discussed from time to time, but so far the conclusion has been that for Windows the competitive advantages of closed source outweigh the efficiency benefits of giving app developers zero-effort access to source code.

I say "efficiency" and "zero effort" because the alternative to open source is exposing source code under NDA or having AMD engineers (with access to driver source they can inspect, modify and build) working closely with the app developers.

That is probably one of the most old-minded and less forward-looking choice you ever made. I don't really care about Windows, so you can do whatever you want and I won't care at all, but I talked to several (small) game developers about that and everybody agreed this is one of the most stupid mistakes AMD is doing.

[darkbasic]

Once you open up code for other OSes you also pick up code for other APIs which we can not expose publicly. Trying to make an open source Windows Vulkan-only driver would be another big refactoring effort. Remember the internal code supports more than just Vulkan.

That's because the whole Vulkan driver should have been wrote with open sourcing in mind since day zero.

[bridgman]

That's because the whole Vulkan driver should have been wrote with open sourcing in mind since day zero.

It was, but only for Linux. There was never a plan to open source for Windows.

[bridgman]

That is probably one of the most old-minded and less forward-looking choice you ever made. I don't really care about Windows, so you can do whatever you want and I won't care at all, but I talked to several (small) game developers about that and everybody agreed this is one of the most stupid mistakes AMD is doing.

You probably need to talk to an equivalent number of large game developers to get a balanced view, but I agree that the closed source model does tend to favour larger development houses.

Obvious question - are they saying it is a stupid move from AMD's perspective (ie believe that a competitive advantage on smaller games combined with a loss of competitive advantage on larger games would still result in increased market share for AMD), or just from their perspective ?

[darkbasic]

Obvious question - are they saying it is a stupid move from AMD's perspective (ie believe that a competitive advantage on smaller games combined with a loss of competitive advantage on larger games would still result in increased market share for AMD), or just from their perspective ?

Their perspective and AMD's are much more entangled than how you would like to show.

[darkbasic]

Obvious question - are they saying it is a stupid move from AMD's perspective (ie believe that a competitive advantage on smaller games combined with a loss of competitive advantage on larger games would still result in increased market share for AMD), or just from their perspective ?

Anyway feel free to think whatever you want: history proved you (the "industry") wrong regarding the benefits of closed source linux drivers and will prove you wrong for Windows as well

[bridgman]

Their perspective and AMD's are much more entangled than how you would like to show.

Can you explain please ? Obviously there is entanglement but I don't see how that interferes with answering the question I asked.

[bridgman]

Anyway feel free to think whatever you want: history proved you (the "industry") wrong regarding the benefits of closed source linux drivers and will prove you wrong for Windows as well

Sorry, how did "history prove us wrong" ? We decided some years ago to focus our development efforts for consumer Linux drivers on the open source drivers, and the result (after a lot of work) was good open source drivers for gaming. If anything that proves us right, doesn't it ?

Are you saying that was a mistake and we should have stayed with closed source ?

[darkbasic]

Can you explain please ? Obviously there is entanglement but I don't see how that interferes with answering the question I asked.

I'm not going to lie: they were obviously talking mainly from their point of view. But they also saw it as a missed opportunity for AMD to break the industry schemes and open themselves to new ways to interact with developers. Obviously since the Windows business model is so much closed compared to Linux, no one can know for sure if this is a bet which is going to repay: the market could not be ready as well. But the impression I got is that NDAs are often overrated and there are no valuable enough secrets to protect at the expense of a better and more open business model. Doing business is a continuous bet and without taking some risks it's harder to shake the current state of the things.

Sorry, how did "history prove us wrong" ? We decided some years ago to focus our development efforts for consumer Linux drivers on the open source drivers, and the result (after a lot of work) was good open source drivers for gaming.

Are you saying that was a mistake and we should have stayed with closed source ?

With "you" I was referring to the whole industry, including the old ATI/AMD when it was still still reluctant to actively pursue an open source strategy.

[bridgman]

But they also saw it as a missed opportunity for AMD to break the industry schemes and open themselves to new ways to interact with developers. Obviously since the Windows business model is so much closed compared to Linux, no one can know for sure if this is a bet which is going to repay: the market could not be ready as well.

Yep, that is totally fair. The question is how many "known and significant risk, unknown benefit" risks we can afford to take at any one time.

Failing to take on a significant risk in exchange for an unknown payback doesn't sound like something one would call "a stupid mistake" though,

But the impression I got is that NDAs are often overrated and there are no valuable enough secrets to protect at the expense of a better and more open business model. Doing business is a continuous bet and without taking some risks it's harder to shake the current state of the things.

In fairness, pretty much everyone believes that other peoples secrets are less valuable than their own. Do these developers all publish their game engine source or do they keep it secret and prevent driver developers from optimizing it ?

Doing business is a continuous bet and without taking some risks it's harder to shake the current state of the things.

Yep, but we are involved in a number of "bet the company" risks already - there's a limit to how much risk you can pile on before things start to crumble.

With "you" I was referring to the whole industry, including the old ATI/AMD when it was still still reluctant to actively pursue an open source strategy.

I have to disagree there. ATI was funding and working on open source driver development as early as 1998. We got out of the open source business partly because of rapidly growing business risk associated with a big DRM push in the industry, and partly because we had a new workstation division (FireGL) that came with a pre-existing Linux workstation driver (fglrx) which we needed to focus on.

Once we joined up with AMD and had a second major revenue stream (CPUs) we could afford to take more risks, and we also had a few years of experience with fglrx under our belts. Discussions about restarting open source driver work started fairly quickly after that, and since then we have supported open source gfx driver work for over a decade.

We were in from 1998 through 2002, out from 2003 through 2006, and back in from 2007 onwards.