AMD Mantle Details from APU13

[erunion]

...or they've changed their plans?

 

[ShintaiDK]

You know very well that even if he posts 5 apps they are close to nothing compared to hundreds of OpenCl apps available today.
Everyone was touting about AVX and 256bit AVX before Haswell release, where are the apps today ??? Nobody even mentioned AVX again since then

No amount of apps listed would satisfy you anyway.

But unlike the niche of OpenCL, you dont flag the news with AVX, AVX2, FMA or SSEx usage. Simply because they are universal and broadly used and supported by compilers.

I will not use the F word, the Intel supporters

Is it that personal for you?

 

[sm625]

I would agree with the statement that AVX is dead. There may be a few applications that use it but its still dead... stillborn...

The subject is Mantle, guys. Stick to that - and specifically the CPU implications of it - or I'll close this thread.
-ViRGE

 

[AtenRa]

Is it that personal for you?

The F was for Fanboys, simply i am not using that word again.

 

[AtenRa]

AMD Mantle Q&A with Developers at APU13

Chris Roberts of Star Citizen, Daniel Baker and Tim Kip of Oxide, Jurjen Katsman of Nixxes, Guennadi Riguer of AMD and Johan Andersson of DICE answer questions about Mantle.

 

[el etro]

Then load up on AMD stock...

BTW, nobody will take over AMD. Apple poached all of the CPU and GPU talent that it needed from the AMD school, and Samsung got a bunch too. Nobody wants a debt laden firm that can hardly compete in its core markets.

And the company still impressing/surprising us and innovating on both CPU(this not so much these years) and GPU sides.


BTW, my FX-4100 is waiting for mantle...

 

[Meekers]

I will say all these talks have been enlightening about why PS4 and Xbone went with what most consider an anemic cpu. It seems these developers are not worried about it.

 

[inf64]

You know what was really anemic? The poor powerpc/cell thing in xbox360 and ps3. That thing was dog slow and hard to program for. IPC wise, Jaguar is miles ahead even at low clock of ~2Ghz, has 8 relatively strong threads(thread= x86 core, unlike SMT stuff in say xbox360) and has dev. software stack that is just another ballpark Vs development tools that were available for previous gen. of consoles.

Mantle will milk the best possible performance from GCN units in nextgen console hardware while at the same time it won't require anything special from the x86 side, Jaguar will do just fine.

 

[NTMBK]

I will say all these talks have been enlightening about why PS4 and Xbone went with what most consider an anemic cpu. It seems these developers are not worried about it.

Plus there seem to be quite a lot of effort to offload compute work in the new consoles. Sony has TrueAudio to offload audio processing, and lots of ACEs to improve GPGPU performance; Microsoft has its custom audio DSP, special DMAs, and DSPs for the Kinect. They seem to be able to move a lot of the load off the CPU. There's this quote from the Eurogamer article:

Digital Foundry: So often you're CPU bound. That explains why so many of the Data Move Engine functions seem to be about offloading CPU?

Andrew Goossen: Yeah, again I think we under-balanced and we had that great opportunity to change that balance late in the game. The DMA Move Engines also help the GPU significantly as well. For some scenarios there, imagine you've rendered to a depth buffer there in ESRAM. And now you're switching to another depth buffer. You may want to go and pull what is now a texture into DDR so that you can texture out of it later and you're not doing tons of reads from that texture so it actually makes more sense for it to be in DDR. You can use the Move Engines to move these things asynchronously in concert with the GPU so the GPU isn't spending any time on the move. You've got the DMA engine doing it. Now the GPU can go on and immediately work on the next render target rather than simply move bits around.

Nick Baker: From a power/efficiency standpoint as well, fixed functions are more power-friendly on fixed function units. We put data compression on there as well, so we have LZ compression/decompression and also motion JPEG decode which helps with Kinect. So there's a lot more than to the Data Move Engines than moving from one block of memory to another.

Lots of fixed function hardware for typical game tasks.

 

[Lepton87]

The good information is that they are not tied to GCN and can abandon it for some new architecture and still retain mantle compatibility. At first a lot of people were concerned that it would slow their GPU innovation due to inability to move from GCN. IMHO NV is too boneheaded to adopt mantle which is a shame I'd like to have some speed-up just from new API but I'm 99.9% sure it's not going to happen. They even limit their own gpu sales by blocking PsyX on systems that have AMD card as main rendering card and NV card as PsyX card, not to mention allow AMD cards to do their own PsyX computations. JHH should go. Imagine NV with no wooden-screws or dissapearing dragons. Yeah, I can't too

 

[frozentundra123456]

The good information is that they are not tied to GCN and can abandon it for some new architecture and still retain mantle compatibility. At first a lot of people were concerned that it would slow their GPU innovation due to inability to move from GCN. IMHO NV is too boneheaded to adopt mantle which is a shame I'd like to have some speed-up just from new API but I'm 99.9% sure it's not going to happen. They even limit their own gpu sales by blocking PsyX on systems that have AMD card as main rendering card and NV card as PsyX card, not to mention allow AMD cards to do their own PsyX computations. JHH should go. Imagine NV with no wooden-screws or dissapearing dragons. Yeah, I can't too

I dont think it will be that easy to move mantle to a different architecture. In any case, gcn will be still in the consoles for the foreseable future, so if they move to a different architecture, they lose the supposed advantage of easy transfer from the consoles.