AMD Mantle Deep Dive Video from AMD APU13 Event

Subject: Graphics Cards | November 13, 2013 - 09:54 PM |
Tagged: video, Mantle, apu13, amd

While attending the AMD APU13 event, an annual developer conference the company uses to promote heterogeneous computing, I got to sit in during a deep dive on the AMD Mantle, a new hardware level API first announced in September.  Rather than attempt to re-explain what was explained quite well, I decided to record the session on video and then intermix the slides presented in a produced video for our readers.

The result is likely the best (and seemingly first) explanation of how Mantle actually works and what it does differently than existing APIs like DirectX and OpenGL.

Also, because we had some requests, I am embedding the live blog we ran during Johan Andersson's keynote from APU13.  Enjoy!

 

November 14, 2013 | 12:09 PM - Posted by nobody special (not verified)

Is it just me, or did the mantle speeches say nothing? We want to help you innovate, blah blah...We'll let you know performance after you write some stuff using it....

I got nothing from AMD's APU speeches...I expected many benchmarks. I don't want to hear "between this and that" I want to hear it's X percent faster in Y game. I don't want to hear what someone EXPECTS to get either. I want to hear what they GOT for REAL. Did I miss the parts with benchmarks?

November 14, 2013 | 09:55 PM - Posted by Anonymous (not verified)

Apparently you didn't read any of that. They're not releasing performance numbers yet because after six more weeks of work the numbers now wont matter anymore. We already know that Mantle draw calls are 9x faster, and thats the big number they keep hitting on. So the only immediate benefit we will see is CPU-bound scenes rendering much faster. Ultimately this could lead to new render techniques that would previously be impossible due to CPU overhead - how they achieve that is the focus of this presentation.

This talk was for the software engineers, and they actually revealed a lot of details about what they've improved. I am particularly interested in their improvement to occlusion culling by removing a CPU-turnaround. Thats a big deal because culling is used extensively to optimize what geometry ends up getting rendered, this could very well allow much more complex scenes in the near future.

November 14, 2013 | 11:19 PM - Posted by Anonymous (not verified)

This was meant for software developers, it was a 40 minute talk and it went deeply into detail about the improvements mantle brings. Seems to me like you didn't understand a quarter of what they talked about, stared with a puzzled look, and the complained that AMD gave you nothing from the speech.

Sounds like you aren't a software developer or at least a competent one therefore this developer summit was not meant for you.

And here are some juicy details of the kinds you are more likely to appreciate

Oxide was demoing their StarSwarm game using Mantle at APU13 here are some tweets from the event:

"Oxide claims that it only took 2 "man months" to support Mantle. "Not much work or expensive" to integrate it."

"3000 units on screen right now, an order of magnitude more effects - 25000 discrete objects simulating and moving" #APU13 #Mantle #Oxide

"API overhead reduced by at least 10x. In some cases StarSwarm is 3x faster with Mantle."

"Completely changes the landscape for multi-core. Claims an FX-8350 is now just as fast as a Core i7-4770K."

"Oxide's presentation & Mantle demo at #APU13 is awesome, 100k draw calls in their space test game. Great work guys! "

"Oxide wants to see 300k batches by 2015, 1 million by 2018."

So yea... A ridiculous improvement over DirectX and OpenGL

November 14, 2013 | 11:46 PM - Posted by Principle (not verified)

That is awesome to say the least. Even Nvidia attended to learn something.

November 14, 2013 | 12:21 PM - Posted by Anonymous (not verified)

^ These talks seem to be targeted at people with attention span longer than 5 seconds.

November 14, 2013 | 11:46 PM - Posted by Principle (not verified)

Good call

November 15, 2013 | 01:40 AM - Posted by Anonymous (not verified)

Heh. I see what you did there...

November 15, 2013 | 06:17 AM - Posted by praack

the real key here is if when released it will be able to scale across to their competitors GPU's. if so then you have a platform agnostic coding system that can be built out with branches per GPU. - that is a good thing

if in the end all you have is an AMD coding system that will work when emulated on the competitors GPU then you have a failure and Mantle will die an ignominious death

November 16, 2013 | 08:20 AM - Posted by Klimax (not verified)

Frankly, I don't want useless slides, I want full documentation and API, otherwise it is just vaporware for all I can care. And I do know a thing or two about programming for DirectX 11 and I suspect that they wrote horrible code to reach "draw calls" bottleneck. DX 11 is already thin as it is, so I doubt their claims. (Also they just manufactured whole new set of libraries and programs to maintain which won't be kind to their financials)

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.