AMD IS FINALLY doing what we have all been waiting for, promoting its developer relations program to the masses. With that in mind, it introduced three new logos to go along with the program.
Any semiconductor company that wants to be taken seriously has to have a developer relations (devrel) program. On the positive side, it promotes the upcoming products to the people making software and hardware that use them, gets early samples out, and generally answers questions. Things can extend to co-marketing campaigns, bundled software, and other high profile mutually beneficial things.
There is a dark side though, some companies use many of those same programs to lock the competition out. “If our logo is on the box, you can’t have features X, Y and Z work with our competitors cards” is a common one, but the less blatant ones are more evil and harmful. Since the ‘support’ provided in these ‘relationships’ can include checks for $1M and up, developers tend to heed these ‘hints’.
Think about AA in the latest Batman game or implementing features that detect rival cards and turn off if one is so much present in the system. These ‘relationships’ only hurt users and the community, but some companies are insistent on putting money into them instead of hardware development. Others are much more egalitarian.
AMD, specifically ATI drew a line in the sand a few days ago and categorically said that it has never used its devrel or related marketing programs to lock out a rival, and never will. Chats with game developers over the years have, at least anecdotally, backed this claim up. Other companies don’t agree, and were spending $20 to $30 million a year to hurt the industry, but that has almost totally dried up of late.
With that in mind, AMD’s Nigel Dessau kicked off what AMD calls the Gamer’s Manifesto, basically that AMD will put gamers first. It has a few major points, drive innovation, industry standards, committing to PC gaming, continuing education, and developer support. Although these things may seem blindingly obvious to anyone looking in from the outside, doing them is a bit tough.
Not quite biohazardous, but still symbolic
Innovation is a key here. If you don’t have a product that does something cool to stand out from the crowd, people probably won’t buy very many. ATI has been a bit behind here. It can’t do three monitors in stereo using more than one card, and really can’t do it with a DX10.0 card. With the announcement of the Open Stereo 3D initiative, ATI caught up somewhat.
The red team can now do Stereo 3D across three 120Hz monitors, but it has to do it with one card, and that card is DX11, not DX10.0. Luckily, it can use both active and passive glasses, and has six companies making compatible eyewear. If one size fits all and exorbitant price doesn’t meet your ideal for competition, help is on the way, ATI will have a much greater range of options and technologies in very short order.
Sadly however, there doesn’t appear to be any hope for ATI rebranding a DX10 card. The 2900 line has long since sold out, and with TSMC on allocation, more is not an option. Everything since those parts are DX10.1 or greater, so Nvidia wins hands down here. ATI will do quad buffering, Blu-Ray 3D, bundle 3D hardware with partners, and have notebooks ready for the tech.
On the physics side, the big announcement was the Open Physics Initiative. There are two components, the Bullet physics library, and Pixelux’s Digital Molecular Matter 2 (DMM). Bullet is an open source physics engine, and DMM is a kind of finite element analysis ‘lite’ solution to make things move and interact realistically. Both are ‘tightly’ integrated, but can also be used with other components as well. Think of this as a free and open solution for physics that anyone can implement.
These programs will live and die on support, and support is exactly what AMD is promising will be there for developers when needed. Hopefully it will, but as always, time will tell. AMD has as serious advantage here in that it can parse the processing out to whatever resource, CPU or GPU that is available. OpenCL is the big driver here, while its rivals tend to push one or the other exclusively even if OpenCL is used.
Complexity tends to kill initiatives like this, and that is where the education and developer support comes in. ATI was asked about the number of developer support people that it had, and Richard Huddy dodged the question but came up with a very good retort. It isn’t how many people you have, it is how effective they are. Before you sneer at that remark, next time you go to the dentist, do you want good dentist with lots of experience, or do you want to save $10 and have the drilling done by a team of 23 people in a third world country that has had over a week in training in tele-dentistry?
Huddy went on to say that when you are not making proprietary solutions, and not trying to bend over backwards to justify breaking your rival’s code, it takes a lot fewer people. AMD trotted out a bunch of top tier game developers to point this out in painful detail, including Chris Taylor from Gas Powered Games.
Taylor was introduced by saying that Gas Powered Games added Eyefinity support to Supreme Commander II in less than a week. Chris shot back that it was far less than that, probably a few hours getting it up, and a day or three tweaking things like where menus went, and other minutia. He went on to show a video of a single 5870 running Supreme Commander II across three monitors.
You will never see this in a real game
They then put the most complex unit from the game, a King Kryptor, on screen, multiplied it by over 500 times, and let them run around. It ran smoothly, the units didn’t stutter in pathfinding, and you could pan around fluidly. Not bad for a single card in a scenario that is much harder than anything real gameplay will throw at it.
These sentiments of ‘easy, quick and useful’ were backed up by developers from Rebellion, EA/Dice and Activision. Multiple monitor gaming is worthwhile, DX11, unlike DX10, is useful and productive, and more GPU power is still a good thing. It is hard to argue those points.
In addition to helping devs out with what they are doing now, AMD sends higher level architects and planners out to developers to ask about their wishes for the far future. Those ideas get fed back into the chips that are not even on paper yet, three to five years out or more. The end result is hopefully a card that has all the features that devs want running at the speed that gamers need. If you don’t do this, you end up with a waffle iron that is really good at double-precision floating-point, something that gamers have absolutely no need for. As Nigel Dessau said, gamers first.
The last chunk is marketing and co-marketing. This is where AMD and a developer, an AIB and a developer, or all three work together to do things like bundle games. You get an XFX AvP bundle with a 5870 for example, or ads by Sapphire featuring a new card and a new game. These all take a lot of coordination, time, effort, and money. AMD is putting a lot into this as well.
To sum it all up in three little boxes, AMD has three new logos that devs can get. If they validate their software to work right under Eyefinity, there are two that apply, and one overarching logo for everything. AMD was adamant that it would not lock out the competition, should it ever get a high end card out in volume that can do three monitors. The ATI multi-monitor APIs are open as well, and can be easily implemented by anyone, so the logos could mean that a game works with any multi-display tech, but I won’t hold my breath.
The logos look like normal ATI logos with “Eyefinity Multi-Display Validated” and “Eyefinity Multi-Display Ready” underneath. Not earth shattering, but worth looking for if you want things to work with all your LCDs.
The last one is called Gaming Evolved, and it is an AMD green and black logo with “Gaming Evolved” under it. This one supposedly incorporates all the above principles in one tidy logo. Think about it as Nvidia’s TWIMTBP program without the antisocial bits, pay-offs and things that hurt gamers while lining specific pockets.
Normally, I would laugh at things like this program, since many are no more than pictures lacking any serious technical backing. However, in talks with several game devs over the past few years, it has become very clear that AMD has been putting the pieces for this program in place for a long time.
From cleaning up drivers to answering questions in the background, there is a lot going on here, the work is being done. Up until now, it was being done in the background, silently, without much fanfare. The pieces seem to be in place now, and here comes the fanfare. About time.S|A
Latest posts by Charlie Demerjian (see all)
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018
- Intel can’t supply 14nm Xeons, HPE directly recommends AMD Epyc - Sep 7, 2018
- AMD reintroduces the Athlon name with two CPUs - Sep 6, 2018