The AMD Mantle Thread

[3DVagabond]

I present 3.5 million draw calls per second in Open GL...with video and detailed information:
http://www.openglsuperbible.com/2013/10/16/the-road-to-one-million-draws/

It looks to me that OpenGL is makeing a joke of Mantle now...

that's 3.5mil/sec. That's fine if you want to render whatever the test was (not a game, i'll bet) at 1fps.

 

[Lonbjerg]

that's 3.5mil/sec. That's fine if you want to render whatever the test was (not a game, i'll bet) at 1fps.

So now you are guessing...nice

There is a lot of text and even code snippets...try reading it

 

[blastingcap]

It's Christmas guys, give it a rest

 

[Noctifer616]

I present 3.5 million draw calls per second in Open GL...with video and detailed information:
http://www.openglsuperbible.com/2013/10/16/the-road-to-one-million-draws/

It looks to me that OpenGL is makeing a joke of Mantle now...

Would we still have the same result if we were to add high resolution textures, tessellation, light, reflections, soft shadows, have each object have it's own AI and move about randomly with added collision physics? You know, the stuff relevant to games.

Unless you can have the same result in an actual game, the video isn't relevant.

We seen Mantle do 70k batches in a actual game demo, have we seen OpenGL do that? No benchmark, actual game demo.

 

[Lonbjerg]

Would we still have the same result if we were to add high resolution textures, tessellation, light, reflections, soft shadows, have each object have it's own AI and move about randomly with added collision physics? You know, the stuff relevant to games.

Unless you can have the same result in an actual game, the video isn't relevant.

We seen Mantle do 70k batches in a actual game demo, have we seen OpenGL do that? No benchmark, actual game demo.

Can I get my hands on this game...and test for me self?
Or would that be a "little no...but, but,BUT, BUT!!!"?

I seen devs talk their own game/engine up (and no demo for me to try)...but I always return to the logical stumbling block:

If all this is so true...why don't consoles perform 2-3 faster than similar PC hardware.

Reality and PR often conflict.

 

[Noctifer616]

Can I get my hands on this game...and test for me self?
Or would that be a "little no...but, but,BUT, BUT!!!"?

If you had watched the video, you would know that they will release it in Q1 next year, so yes, you can get your hands on it in a few months.

I seen devs talk their own game/engine up (and no demo for me to try)...but I always return to the logical stumbling block:

If all this is so true...why don't consoles perform 2-3 faster than similar PC hardware.

Reality and PR often conflict.

They don't? Don't look at the PS4 and XBOX ONE, look at the last generation to compare performance to PC. The new generation looks better than what the average gamer experiences in IQ. Thus, they aren't getting 100% out of the hardware. It will take time for developers to figure out all the tricks with the new hardware, the same thing happened with the last generation.

If consoles can't get a lot more out of their hardware than PC's, then how come GTA 5 runs on the last generation?

 

[Lonbjerg]

If you had watched the video, you would know that they will release it in Q1 next year, so yes, you can get your hands on it in a few months.



They don't? Don't look at the PS4 and XBOX ONE, look at the last generation to compare performance to PC. The new generation looks better than what the average gamer experiences in IQ. Thus, they aren't getting 100% out of the hardware. It will take time for developers to figure out all the tricks with the new hardware, the same thing happened with the last generation.

If consoles can't get a lot more out of their hardware than PC's, then how come GTA 5 runs on the last generation?

Lets try again.
Show me where the PS3 outperforms a 7800GT by 2-3 times.
Show me where a Xbox360 outperforms a X18000 by 2-3 times.

The notion "avarage gamer" is a fallacy here.

We are talking performance of hardware under different API's, stay on track please.

And you should read up on all the tricks used to make game run on consoles:
Lower textures
Lower draw distance.
Lesser shadow and lightmaps.

The list is very LONG, here is a video to highlight that even if a game bear the name...the image quality is NOT the same:
http://www.youtube.com/watch?v=2WJG14uLA3k

 

[Noctifer616]

Lets try again.
Show me where the PS3 outperforms a 7800GT by 2-3 times.
Show me where a Xbox360 outperforms a X18000 by 2-3 times.

GTA 5

The notion "avarage gamer" is a fallacy here.

No it isn't. The PS4 and XBOX ONE have more powerful hardware than what the average gamer is running.

 

[desprado]

GTA 5



No it isn't. The PS4 and XBOX ONE have more powerful hardware than what the average gamer is running.

Do even know Ps4 and Xbone cannot even beat a mid range PC.


Ps3 was 400 flops raw power
Xbox 360 was around 300 flops

these both console cannot beat 9800GTX in res of 720 or less.

 

[Lonbjerg]

GTA 5

Could you link me to the PC version of GTA5? ^^
If not...are you using false arguments..and why?!

No it isn't. The PS4 and XBOX ONE have more powerful hardware than what the average gamer is running.

It is a fallacy here.
Again, this thread is about the Mantle API and how it compares to other API's on identical hardware.

Not about some sort of artifical "metric" you decide is needed in order to make Mantle look good.

I am still waiting for you to present evidence this:
Show me where the PS3 outperforms a 7800GT by 2-3 times.
Show me where a Xbox360 outperforms a X18000 by 2-3 times.

I mean..that is the claim being made in this thread a lot of times..."coding to the metal" shoud give so much mor performance...I am just not seeing the evidence in any of the data available.

 

[psoomah]

If all this is so true...why don't consoles perform 2-3 faster than similar PC hardware.

Reality and PR often conflict.

Because it's all lies. All this 'close to the metal' propaganda ... what does that even mean? It doesn't even make sense, it isn't even metal, it's silicon, that's a give away right there.

Put the guts of any console in a computer case and run Windows on it and right there you've got a killer gaming machine that will run circles around that console running the same game 'close to the metal'.

Nothing but corporate propaganda.

 

[Noctifer616]

Again, this thread is about the Mantle API and how it compares to other API's on identical hardware.

Not about some sort of artifical "metric" you decide is needed in order to make Mantle look good.

And yet you show a irrelevant video of OpenGL doing massive amount of draw calls in a scenario that isn't even close to games.

 

[desprado]

Because it's all lies. All this 'close to the metal' propaganda ... what does that even mean? It doesn't even make sense, it isn't even metal, it's silicon, that's a give away right there.

Put the guts of any console in a computer case and run Windows on it and right there you've got a killer gaming machine that will run circles around that console running the same game 'close to the metal'.

Nothing but corporate propaganda.

A tell u fact u can go and search even a AMD 7850 was running BF4 on 1080p with high settings at 45 to 50 fps while while PS4 cannot even run 60 fps on 900p.

 

[NTMBK]

Could you link me to the PC version of GTA5? ^^
If not...are you using false arguments..and why?!





It is a fallacy here.
Again, this thread is about the Mantle API and how it compares to other API's on identical hardware.

Not about some sort of artifical "metric" you decide is needed in order to make Mantle look good.

I am still waiting for you to present evidence this:
Show me where the PS3 outperforms a 7800GT by 2-3 times.
Show me where a Xbox360 outperforms a X18000 by 2-3 times.

I mean..that is the claim being made in this thread a lot of times..."coding to the metal" shoud give so much mor performance...I am just not seeing the evidence in any of the data available.

The problem is that it is ridiculously difficult to get benchmarks of modern games on such old hardware. I would absolutely love it if someone could compare the performance of BF4 on a 360 against a PC with a Pentium D and an X1800- I suspect that the PC wouldn't come out looking too hot. But of course this isn't possible, as you can't even run Battlefield 4 on such an old graphics card (too little VRAM, no DirectX 10 or 11). And the games which show the most console optimization are the ones which come at the end of the lifecycle when the developers are more used to the architecture, such as e.g. GTAV.

 

[SirPauly]

All this 'close to the metal' propaganda ... what does that even mean? It doesn't even make sense, it isn't even metal, it's silicon, that's a give away right there.

Surely, there is actual metal!

From Sand to Silicon “Making of a Chip” Illustrations

http://download.intel.com/newsroom/kits/chipmaking/pdfs/Sand-to-Silicon_45nm-Version.pdf

 

[NTMBK]

Surely, there is metal!

The trick is in squeezing the most out of the flux capacitors.

 

[SiliconWars]

Xbox 360 IQ difference in 8 years - http://www.forbes.com/sites/insertc...-far-xbox-360-games-have-come-in-eight-years/

PS3 IQ difference in 7 years - http://www.forbes.com/sites/insertc...playstation-3-games-have-come-in-seven-years/

 

[Lonbjerg]

Not sure what you argument is?

 

[SirPauly]

The trick is in squeezing the most out of the flux capacitors.

Talking about flux capacitors reminds me of future time-tables and back on topic:

How will the market react to Mantle in the near term and long term?

Definitely has peaked my curiosity!

 

[Lonbjerg]

And I am still waiting for that link to GTA 5 for PC?

 

[Gloomy]

And I am still waiting for that link to GTA 5 for PC?

itll prolly run as crap as gta 4 did, for the sake of your argument do not bring gta's pc versions into this :thumbsdown:

 

[SiliconWars]

Not sure what you argument is?

It's obviously that console games improve vastly over their lifespan. This generation of consoles has started close to PC quality and if it weren't for Mantle they would overtake PC's in image quality soon.

 

[Lonbjerg]

No, what consoles do, over their lifetime, is to reduce some settings too free up limited resources to add new features....subtract here too add there.

There is no magical free performance.

 

[SiliconWars]

There's nothing magical about it for sure, it's just doing what the hardware was always capable of. Go watch a youtube vid of Gears of War 3 or Last of Us and ask yourself if an x1800 could run at that performance on directx.

Now imagine what games on Mantle could look like. There will be some learning to do but the quality difference is going to be staggering very early on. I think a lot of people just don't realise what is about to happen.

 

[ShintaiDK]

I present 3.5 million draw calls per second in Open GL...with video and detailed information:
http://www.openglsuperbible.com/2013/10/16/the-road-to-one-million-draws/

It looks to me that OpenGL is makeing a joke of Mantle now...

Nice :thumbsup:

I wonder if AMD used the draw call limit against their own driver. And OpenGL is another weak point for AMDs drivers.

It keep looking like Mantle is a hotfix for their own limitations rather than API. And a desire to stagnante progress. Aka no need to evolve beyond GCN.