Overclock.net › Forums › Industry News › Software News › [AMD/YT] AMD Mantle Technical Deep Dive - APU13 Event - PC Perspective
New Posts  All Forums:Forum Nav:

[AMD/YT] AMD Mantle Technical Deep Dive - APU13 Event - PC Perspective - Page 12

post #111 of 119
Quote:
Originally Posted by skupples View Post

I really don't feel like having to have two gaming rigs in the future, that's all I know.

Amen to that.
post #112 of 119
1. If the graphic manufacturer asked game developers about their requirements then its some sort of groundbreaking approach.
2. Now we see more figures how much faster can mantle be in amount of draw calls, and technical details how it works.

If its true if developers have direct (or at least more direct) control over rendering process if they utilize Mantle, then we can expect change in gaming... sooner or later.

3. Still waiting for real-life application/game engine to see the real performance.

At all I am quite impressed about the approach.
post #113 of 119
For those interested in all things Mantle, I really recommend reading this article from back in March of 2011:

Farewall to DirectX?

No ground breaking information (obviously) but it gives you an idea of how long devs have been asking for this, the main benefits of using a low level API and the disadvantages as well. Excellent read.
post #114 of 119
Quote:
Originally Posted by SlackerITGuy View Post

For those interested in all things Mantle, I really recommend reading this article from back in March of 2011:

Farewall to DirectX?

No ground breaking information (obviously) but it gives you an idea of how long devs have been asking for this, the main benefits of using a low level API and the disadvantages as well. Excellent read.

I have to mention one quite important thing. Most similar studies were focusing on DirectX 9.0c as the most used API. Developers have been criticizing it for years (different reasons). DirectX 10, and 11 are here for some time, yet the game titles were adapting to new versions quite slowly, and the real improvements? Hard to see if you dont know what to search for.

If converting game from DirectX to Mantle took about 2 months I want to know how many people were working on the conversion, how good was the support from AMD side. Usually conversions from consoles took about 1 year (case of Mortal Kombat 9 for example, however I dont know how long took the development itself.)

Even if this brings back the " spirit of bad times" of 3dfx glide and specialized API with poor support, now we have quite different situation.

a) Less GPU manufacturers (AMD, NVidia, Intel?, Matrox? any other?).
b) Higher GPU performance
c) "Standard" APIs which can be applied to any GPU (DirectX and OpenGL).

So as a game developer you can decide if you want to use only standard approach. The game will run, but the performance may be limited. Or you can use "GPU specific" API instead of DX/OpenGL. If you can afford to pay both GPU manufacturers to improve the performance using their proprietary API, you can guarantee that you product will run flawlessly on AMD and Nvidia card.

In this situation I completely support the idea of AMD Mantle and similar API in green camp. Only game developers and gamers themselves can benefit from it. Generally I hate AMD "only" or Nvidia "only" game titles.

Looks nice "on paper" but I am very curious on the real implementation in next 2 years.
post #115 of 119
Quote:
Originally Posted by Offler View Post

I have to mention one quite important thing. Most similar studies were focusing on DirectX 9.0c as the most used API. Developers have been criticizing it for years (different reasons). DirectX 10, and 11 are here for some time, yet the game titles were adapting to new versions quite slowly, and the real improvements? Hard to see if you dont know what to search for.

Similar studies? well that's okey, I'm pretty sure the people involved in this article were fully aware of Direct3D 10 and 11, after all Direct3D 11 debuted all the way back in mid 2008 IIRC, and devs and GPU guys were familiarized with it much sooner.
Quote:
Originally Posted by Offler View Post

If converting game from DirectX to Mantle took about 2 months I want to know how many people were working on the conversion, how good was the support from AMD side. Usually conversions from consoles took about 1 year (case of Mortal Kombat 9 for example, however I dont know how long took the development itself.)

Well apparently Johan Andersson did say that it roughly takes around ~2 months of man hours, just in case if you don't know what man hours means, is the amount of work performed by the average worker in one hour.
Quote:
Originally Posted by Offler View Post

Even if this brings back the " spirit of bad times" of 3dfx glide and specialized API with poor support, now we have quite different situation.

a) Less GPU manufacturers (AMD, NVidia, Intel?, Matrox? any other?).
b) Higher GPU performance
c) "Standard" APIs which can be applied to any GPU (DirectX and OpenGL).

So as a game developer you can decide if you want to use only standard approach. The game will run, but the performance may be limited. Or you can use "GPU specific" API instead of DX/OpenGL. If you can afford to pay both GPU manufacturers to improve the performance using their proprietary API, you can guarantee that you product will run flawlessly on AMD and Nvidia card.

In this situation I completely support the idea of AMD Mantle and similar API in green camp. Only game developers and gamers themselves can benefit from it. Generally I hate AMD "only" or Nvidia "only" game titles.

Looks nice "on paper" but I am very curious on the real implementation in next 2 years.

Agreed.
post #116 of 119

2 months of man hours is the really impressive part, in the grand scheme of things that isn't that expensive - AMD could pay devs to put it in and add value that way.

post #117 of 119
Quote:
Originally Posted by SlackerITGuy View Post

Similar studies? well that's okey, I'm pretty sure the people involved in this article were fully aware of Direct3D 10 and 11, after all Direct3D 11 debuted all the way back in mid 2008 IIRC, and devs and GPU guys were familiarized with it much sooner.

DirectX "End of life" was never officially announced and I believe MS never planned such step. On the other hand, if DXAPI is increasingly becoming an obstacle in further development, it is a huge problem. People started to do similar studies around 2006, and most of them blamed Microsoft for keeping DX9.0 alive for so long. But thats partially related to longevity of Windows XP.

I understand - Microsoft cant do much from position of "middleman", but first figures which were comparing DX and Mantle were still using DX 9.0c data ... Not quite fair in 2013...

2 months of manhours? Impressive.
post #118 of 119
Quote:
Originally Posted by Offler View Post

DirectX "End of life" was never officially announced and I believe MS never planned such step. On the other hand, if DXAPI is increasingly becoming an obstacle in further development, it is a huge problem. People started to do similar studies around 2006, and most of them blamed Microsoft for keeping DX9.0 alive for so long. But thats partially related to longevity of Windows XP.

I understand - Microsoft cant do much from position of "middleman", but first figures which were comparing DX and Mantle were still using DX 9.0c data ... Not quite fair in 2013...

2 months of manhours? Impressive.

I would like to see that actually.
post #119 of 119
I was referring to this:
http://www.anandtech.com/show/7371/understanding-amds-mantle-a-lowlevel-graphics-api-for-gcn

There is a figure by AMD from 2006 when they were comparing DX9 and dX10. Actually I saw that figure sometime back in 2009...

But if you are implying that there is no figure which directly compare DirectX and Mantle by draw calls, I agree - there is still only material by the AMD...
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Software News
Overclock.net › Forums › Industry News › Software News › [AMD/YT] AMD Mantle Technical Deep Dive - APU13 Event - PC Perspective