• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

AMD Mantle Technical Deep Dive - APU13 Event

Status
Not open for further replies.
really interesting video.. especially about how memory is handled now in GPUs (forever linked). and explaines why GPU memory size is really growing so fast..

and WOW at the multiGPU usage in Mantle.. probably the biggest info out of the video.

giving asyncronize workload gives the ability to bring in an APU or even smaller GPU and allow them all to contribute to the workload instead of relying on AFR.

they are giving the developers more control over the mulitgpu environments.
 
Until we see actual in-game framerates with Mantle, I'm going to remain skeptical.
 
chiablo said:
Until we see actual in-game framerates with Mantle, I'm going to remain skeptical.
Click to expand...

True, but its been too long since we've seen a real game-changer. usually lots of hype, no payoff.

I'd love to see this thing deliver and take off.
 
I'm not going to hold my breath. I still remembed when there was Glide on 3dfx and together they were kings of 3D. Glide was faster/better than D3D and OpenGL, but when was the last time you bought a glide game or a 3dfx card? I can't see another proprietary api taking off.
 
its quasi properitary.

There is no reason why Nvidia cant implement the instruction set with their own binaries.
 
I don't think API is a good solution as Graphics are moving to cloud streaming meaning there will be little requirements on the PC / Tablet / Phone side other than decoding. Look at the market its pushing that direction. Adding a proprietary API isn't going to fly these days I think they are looking for means to keep up with competition and secure something. Next gen cards are around the corner and they are very fast compared to what we have today. I wish them all the luck, but I think this will bite them.
 
The Mac said:
its quasi properitary.

There is no reason why Nvidia cant implement the instruction set with their own binaries.
Click to expand...

1. they dont need to
2. even if they wanted to it wouldnt be for 2-3 gens out at lest

Maxwell will do all the same things with out devs having to change any thing and should work with games already out


so far this looks like AMD saying we dont have the money to compete with NV in R&D so will just stick with GCN and make the devs do all the hardwork
Mantle doesnt do any thing OpenGL 4 and DX11.2 dont already do and do just as good or better with better tools
 
The reason you dont see a Glide game is because 3dfx is dead (duh), but they died by not having a proper 2D/3D card, not to mention, the incredibly stupid decision of buying STB and stopping selling GPU's to others OEMs.

But Glide was faster and better than D3D and OpenGL on those days. And given that Nvidia seems to love to overcharge for their hardware, a strong AMD will keep them honest, not to mention, after the dick maneuver they did with Physx, its only fair to have something like that done to them.
 
The Mac said:
its quasi properitary.

There is no reason why Nvidia cant implement the instruction set with their own binaries.
Click to expand...
The instruction set isn't the biggest issue. Mantle code will be tuned for underlying GCN architectures, because that's what it runs on currently. As you can see in OpenCL code, having a common language doesn't mean much as performance is all over the place, depending on what hardware was used for development and tuning.

And there's no quasi about Mantle's proprietary nature. ;)
 
Darcschnider said:
I don't think API is a good solution as Graphics are moving to cloud streaming meaning there will be little requirements on the PC / Tablet / Phone side other than decoding. Look at the market its pushing that direction. Adding a proprietary API isn't going to fly these days I think they are looking for means to keep up with competition and secure something. Next gen cards are around the corner and they are very fast compared to what we have today. I wish them all the luck, but I think this will bite them.
Click to expand...

I just love playing streamed games with super varying lag. I guess if gamers these days can't tell the difference between a local game (with no lag) and a streamed game with tons of lag, then that might take off.

Until we all have at least 1Gb dedicated connections, the lag is going to be horrible unless they super compress the sent images on the fly, or if they limit it to some stupidly low fps.

Even then, you have to take into account input delay.

No thanks to having even 1ms of delay between a key press and it registering in game.
 
Elios said:
1. they dont need to
2. even if they wanted to it wouldnt be for 2-3 gens out at lest

Maxwell will do all the same things with out devs having to change any thing and should work with games already out


so far this looks like AMD saying we dont have the money to compete with NV in R&D so will just stick with GCN and make the devs do all the hardwork
Mantle doesnt do any thing OpenGL 4 and DX11.2 dont already do and do just as good or better with better tools
Click to expand...

Lol, I love your blind faith,

How exactly will Maxwell do all of the same things as mantle WITHOUT developer involvement?
 
Digital Viper-X- said:
Lol, I love your blind faith,

How exactly will Maxwell do all of the same things as mantle WITHOUT developer involvement?
Click to expand...

some thing about using an ARM cpu to do all the things AMD is doing with an API but at the driver level
noticed they avoided talking about DX11.2 and OpenGL BUT LOTS of comparing to DX9...
Hmmmm maybe that DX11.2 and OpenGL can already do all of that
 
bildad said:
I'm not going to hold my breath. I still remembed when there was Glide on 3dfx and together they were kings of 3D. Glide was faster/better than D3D and OpenGL, but when was the last time you bought a glide game or a 3dfx card? I can't see another proprietary api taking off.
Click to expand...

Its not completely proprietary, they mention that it will work to an extent with their competitors but im sure at a sacrificed level of performance for the application. Also, when's the last time a proprietary API had control of both next gen consoles and a decent share of the PC market as well? I think AMD has quite a bit of leverage to make Mantle successful if developers like it.

I'm not well versed in this side of technology, the inner workings of devices but, being able to use multi-GPU configurations for different tasks seems like it could create some pretty compelling experiences for those that have the hardware. Mantle seems to give a ton of control to the developers which may allow for greater performance and/or fidelity in games but, it will definitely require that the developers invest time in learning the API, some of the lazier or lower paid developers may not show much interest in doing that.
 
Elios said:
some thing about using an ARM cpu to do all the things AMD is doing with an API but at the driver level
noticed they avoided talking about DX11.2 and OpenGL BUT LOTS of comparing to DX9...
Hmmmm maybe that DX11.2 and OpenGL can already do all of that
Click to expand...

And how is an ARM cpu used to interpret code BETTER than code natively written for something?
 
cyclone3d said:
No thanks to having even 1ms of delay between a key press and it registering in game.
Click to expand...
If you're pulling 60 Hz on your local machine, you're waiting at least 15ms between a keydown/keyup and something coming out the back of your graphics card, and that's only if your keyboard poll rate wasn't too low for that event to register for that tick. That isn't even considering the time it takes for your display to start throwing that frame's photons at you.

What you want is not only practically infeasible, but pointless as well.
 
Elios said:
1. they dont need to
2. even if they wanted to it wouldnt be for 2-3 gens out at lest

Maxwell will do all the same things with out devs having to change any thing and should work with games already out


so far this looks like AMD saying we dont have the money to compete with NV in R&D so will just stick with GCN and make the devs do all the hardwork
Mantle doesnt do any thing OpenGL 4 and DX11.2 dont already do and do just as good or better with better tools
Click to expand...

lol.. watch the video.. mantle has less to do with the hardware, it has more to do with how developers access the hardware
 
cyclone3d said:
I just love playing streamed games with super varying lag. I guess if gamers these days can't tell the difference between a local game (with no lag) and a streamed game with tons of lag, then that might take off.

Until we all have at least 1Gb dedicated connections, the lag is going to be horrible unless they super compress the sent images on the fly, or if they limit it to some stupidly low fps.

Even then, you have to take into account input delay.

No thanks to having even 1ms of delay between a key press and it registering in game.
Click to expand...

Agreed 100%.
 
Elios said:
1. they dont need to
2. even if they wanted to it wouldnt be for 2-3 gens out at lest

Maxwell will do all the same things with out devs having to change any thing and should work with games already out


so far this looks like AMD saying we dont have the money to compete with NV in R&D so will just stick with GCN and make the devs do all the hardwork
Mantle doesnt do any thing OpenGL 4 and DX11.2 dont already do and do just as good or better with better tools
Click to expand...

Troll/Fanboi post award of the day goes to Elios
 
You people are comparing this to companies which withered and died so long ago? Whats wrong with you...
 
cyclone3d said:
I just love playing streamed games with super varying lag. I guess if gamers these days can't tell the difference between a local game (with no lag) and a streamed game with tons of lag, then that might take off.

Until we all have at least 1Gb dedicated connections, the lag is going to be horrible unless they super compress the sent images on the fly, or if they limit it to some stupidly low fps.

Even then, you have to take into account input delay.

No thanks to having even 1ms of delay between a key press and it registering in game.
Click to expand...

lag is more a matter of latency than bandwidth, 1gb may not make it any better better
 
release some independent benchmarks and then we'll see...pre-release hype only sounds good on paper...preferably benchmarks for more then just BF4 as well
 
polonyc2 said:
release some independent benchmarks and then we'll see...pre-release hype only sounds good on paper...preferably benchmarks for more then just BF4 as well
Click to expand...

this wasn't even amd hype. This was an information session for developers.
Even you don;t know how to code , the graphs and explanation were simple enough
to understand where the extra performance and benefits will comes from.

I do agree with you, having something more tangible would've been nice.
 
So is there confirmation that Mantle is coming to BF4 in december? If so, we're weeks away from seeing some frame rates.. right?
 
maboblivion said:
this wasn't even amd hype. This was an information session for developers.
Even you don;t know how to code , the graphs and explanation were simple enough
to understand where the extra performance and benefits will comes from.

I do agree with you, having something more tangible would've been nice.
Click to expand...

technically it was for developers but we all know that AMD is using this as their official kickoff of Mantle...they want gamers and enthusiasts to start getting excited about Mantle's potential
 
NickJames said:
He's held that award for months now.
Click to expand...

I concur. Every time I read anything about AMD lately Elios is always in the thread to play down anything positive about AMD. No one really knows how well Mantle will do yet, but we know one thing for sure; AMD is going to dominate console gaming for the next 5-6 years. I want competition and Nvidia's answer to AMD lately is really weird to me, a card that is slightly faster for $150 more. The only real difference is heat and noise. I still might get the 780Tis but that depends on how well they do Sli vs Xfire. But for the price of 2 780Tis I can get 3 290X for $150 more.
 
polonyc2 said:
technically it was for developers but we all know that AMD is using this as their official kickoff of Mantle...they want gamers and enthusiasts to start getting excited about Mantle's potential
Click to expand...

Pretty sure 10% of gamers watched the summit and the 90% that did couldn't understand a word that was spoken.
 
cyclone3d said:
I just love playing streamed games with super varying lag. I guess if gamers these days can't tell the difference between a local game (with no lag) and a streamed game with tons of lag, then that might take off.

Until we all have at least 1Gb dedicated connections, the lag is going to be horrible unless they super compress the sent images on the fly, or if they limit it to some stupidly low fps.

Even then, you have to take into account input delay.

No thanks to having even 1ms of delay between a key press and it registering in game.
Click to expand...

You do realize even with everything being rendered right there infront of you, you still can't achieve that low of latency. Keyboards, Mice, Monitors all have between 5-15ms of latency right of the bat. Wireless? Double it.

Also, seeing as my Wireless G router is capable of doing HD streams, I'd say you could do 1080p gaming on current connections.

Your expectations are foolish, as are your misconceptions about game streaming.
 
Zion Halcyon said:
True, but its been too long since we've seen a real game-changer. usually lots of hype, no payoff.

I'd love to see this thing deliver and take off.
Click to expand...

Personally I hope it fails. I don't like features that are limited to one brand. I assume it will also increase the workload on developers who have to program for both Mantle and Direct X.

I wish Physx had either failed in favor of something that works equally on AMD or Nvidia opened it up to AMD (maybe AMD pays them?) but I don't see that happening.
 
Flogger23m said:
Personally I hope it fails. I don't like features that are limited to one brand. I assume it will also increase the workload on developers who have to program for both Mantle and Direct X.

I wish Physx had either failed in favor of something that works equally on AMD or Nvidia opened it up to AMD (maybe AMD pays them?) but I don't see that happening.
Click to expand...

Here's the thing, AMD isn't stopping them. Nvidia could use this if they wanted to but they won't because they are too proud.
 
NickJames said:
Here's the thing, AMD isn't stopping them. Nvidia could use this if they wanted to but they won't because they are too proud.
Click to expand...

Nothing to do with proud it is their way of marketing. That is also the reason why there is piss poor openCL driver.

If they haven't invented it they won't do anything with it. Look at SLI compatibility in the end they had to sell it to still get some money or be forced out.
 
Flogger23m said:
Personally I hope it fails. I don't like features that are limited to one brand. I assume it will also increase the workload on developers who have to program for both Mantle and Direct X.
Click to expand...
The best case scenario is that it sees limited success. Just enough to drive certain industry bodies to swing the pendulum toward more low-level access and some level of standardization on hardware front-ends. If Mantle doesn't dissipate within the next two to three years, things are going to start getting really problematic for consumers the same way the clusterfuck of proprietary graphics APIs became problematic during the early 3D accelerator era, but limited, short-term success isn't going to be cataclysmic.

Expect some turmoil, but don't expect an apocalypse.
 
pxc said:
The instruction set isn't the biggest issue. Mantle code will be tuned for underlying GCN architectures, because that's what it runs on currently. As you can see in OpenCL code, having a common language doesn't mean much as performance is all over the place, depending on what hardware was used for development and tuning.

And there's no quasi about Mantle's proprietary nature. ;)
Click to expand...
I don't think that's the case. They've stated it's a thin hardware abstraction, it's not coding directly to the GCN metal. That wouldn't even make any sense, as it would make Mantle useless when AMD changes their architecture.
 
Elios said:
1. they dont need to
2. even if they wanted to it wouldnt be for 2-3 gens out at lest

Maxwell will do all the same things with out devs having to change any thing and should work with games already out


so far this looks like AMD saying we dont have the money to compete with NV in R&D so will just stick with GCN and make the devs do all the hardwork
Mantle doesnt do any thing OpenGL 4 and DX11.2 dont already do and do just as good or better with better tools
Click to expand...
Just stop posting, you have no idea what you are talking about.

1) Mantle does a lot of things that DX11.2 and OGL don't
2) It's not hardware specific so it wouldn't take NVIDIA multiple generations to implement, they could likely implement it on current hardware
3) Devs have been ASKING for this kind of control over the hardware for quite some time
4) Maxwell is hardware and won't address any of the software environment concerns Mantle is designed for.
 
How so? TressFX is not comparable since it works on Intel and Nvidia too. Ironically Tomb Raider runs better on NVIDIA hardware now.
 
polonyc2 said:
that's more about making themselves look good...they know Nvidia will not use an AMD driven feature...same thing with TressFX
Click to expand...

I bet if Nvidia was the one coming up with this the community wouldn't be as vocal. AMD finally wants to push the limits of hardware and now everyone is complaining. Nvidia has the choice to embrace it or ignore it, if they want to ignore it then that's their problem.
 
Status
Not open for further replies.
Back
Top