The AMD Mantle Thread

[monstercameron]

why are people going crazy over this, mantle games will probable use both mantle and direct x at the same time[maybe not so much for cross platform devs]...the comparison should be direct 3d and mantle not directX...this might be a nitpick, the back and forth was interesting at first but now this thread has devolved into noise...

 

[AnandThenMan]

He never said break DX. He said break the stagnation of DX. Not the same thing.

Yes exactly. And yes DirectX has stagnated, or perhaps bloated is a better term. Doesn't matter because either way, it leaves way, way too much performance on the table. The whole point of Mantle to me is to bring console efficiency to the PC world, and it's about time. Just imagine console efficiency with a high end (or two) GPUs.

On the subject of Linux, I think Mantle here has even more potential. A leaned down Linux system is really, really fast.

 

[3DVagabond]

Yes exactly. And yes DirectX has stagnated, or perhaps bloated is a better term. Doesn't matter because either way, it leaves way, way too much performance on the table. The whole point of Mantle to me is to bring console efficiency to the PC world, and it's about time. Just imagine console efficiency with a high end (or two) GPUs.

On the subject of Linux, I think Mantle here has even more potential. A leaned down Linux system is really, really fast.

Come on now. Everyone knows that the fastest computers in the world all run on Windows... Oh, wait... no they don't. :\

 

[SiliconWars]

I think it's pretty likely we're going to see a Mantle or DX option as games load, just like in the old days with Glide.

In GE titles, all they need to do is get the devs to find an extra 20% performance while increasing visuals at the "same settings" as the DX version. After that both versions will be reviewed at the "same settings", obviously the game looking better on Mantle yet still having the performance gains (lets say global illumination like in Dirt Showdown). Now the reviewers have to review both or it's just plain bias.

This is different from PhysX in a couple of ways ie PhysX has a performance penalty and can be used to hurt AMD cards more. It is also questionable whether or not the IQ gains are even worth it - plenty of people say an over-abundance of PhysX, or PhysX done wrong just looks bad. This won't be the case with Mantle, it'll most be increased draw distances which everybody can agree is extremely desirable, or better quality lighting.

I suppose a poorer quality AA like MLAA could be used and therefore AMD could gain a huge performance advantage at an IQ penalty, but that would be lazy and stupid of them to do so. AMD will go for the double advantage of better IQ and better performance.

Just look at Showdown actually. Seems logical that AMD discovered a way to get global illumination to thread properly on GCN cards but not on Nvidia cards. If you recall the part in Mantle about the serial nature of graphics and compute in DX, all that would take was them finding a way to thread the compute shaders properly on GCN in Showdown. While not exactly "hobbling" Nvidia's cards, they found a way to make the GCN cards get the extra compute performance without harming normal GPU performance.

 

[sniffin]

The problem with PhysX is it runs like crap even on the nicest Nvidia hardware available at the games release. It's also had very few impressive implementations. Devs seem to only want to do enough to appease Nvidia, which is fair enough

 

[3DVagabond]

I think it's pretty likely we're going to see a Mantle or DX option as games load, just like in the old days with Glide.

In GE titles, all they need to do is get the devs to find an extra 20% performance while increasing visuals at the "same settings" as the DX version. After that both versions will be reviewed at the "same settings", obviously the game looking better on Mantle yet still having the performance gains (lets say global illumination like in Dirt Showdown). Now the reviewers have to review both or it's just plain bias.

This is different from PhysX in a couple of ways ie PhysX has a performance penalty and can be used to hurt AMD cards more. It is also questionable whether or not the IQ gains are even worth it - plenty of people say an over-abundance of PhysX, or PhysX done wrong just looks bad. This won't be the case with Mantle, it'll most be increased draw distances which everybody can agree is extremely desirable, or better quality lighting.

I suppose a poorer quality AA like MLAA could be used and therefore AMD could gain a huge performance advantage at an IQ penalty, but that would be lazy and stupid of them to do so. AMD will go for the double advantage of better IQ and better performance.

Just look at Showdown actually. Seems logical that AMD discovered a way to get global illumination to thread properly on GCN cards but not on Nvidia cards. If you recall the part in Mantle about the serial nature of graphics and compute in DX, all that would take was them finding a way to thread the compute shaders properly on GCN in Showdown. While not exactly "hobbling" Nvidia's cards, they found a way to make the GCN cards get the extra compute performance without harming normal GPU performance.

They need to make it to where Mantle can't be bypassed on GCN cards. Then they have no choice but to run it.

 

[SiliconWars]

They need to make it to where Mantle can't be bypassed on GCN cards. Then they have no choice but to run it.

Well that would be a simple vendor id check on startup I suppose. I do believe that they will go with a selection option however - maybe Mantle breaks something that isn't shown in development and then at least they have the option of falling back to DX...

Another potential issue is DLC, namely having to code for DX and Mantle again at this stage (this is actually one of my biggest Mantle fears). It's probably not that much though is it...

 

[SiliconWars]

With all this doom and gloom for everyone but AMD, i just want to know one thing. If mantle is so easy to code, so easy to convert console code to run on PC..... if this is remotely close to true, then why is there still no BF4 mantle patch? The game has launched on the consoles. The game has launched already and running perfect on PC in DX. You see, it doesnt add up. Mantle obviously is more work and not simply taking console code and running it on a PC. Or it would be here already. If you cannot see the issue with this then I dont know what to tell you.

They always said it would be 2 months. BF4 was released on the 29th October, not even 1 month ago.

Also I wouldn't exactly say the DX version was running perfectly. In fact it's been pretty awful from what I can gather. Any issues with the game engine in DX would be prioritised for now, possibly pushing the Mantle port back.

In reality we don't know when they started but given the alpha nature of Mantle it seems logical to assume that they started after getting the console and DX versions finished.

Mantle is a very big deal to AMD. They are trying to push it and push it hard. Yet BF4 launch came and went, without mantle. Lucky this is a game that has a huge MP aspect that people play for many months but not every game is like that. As a matter of fact, most games are not like this. And if AMD could not implement their huge game changing Mantle to spotlight their huge architecture with a grade A title at launch, what makes anyone think that any other game would be different. Why is so many in this thread so sure of so many things? Wishful thinking? Praying and hoping that AMD will rise above and crush Nvidia and Intel based on nothing more than blind faith?

The point you're missing is Mantle wasn't ready. It's still not ready, it's not even in beta.

Look, BF4 is their Mantle debut. Their big Mantle showcase and its completely MIA. This is a great way to start off your huge game changing API, wouldnt you think. Well, i dont think so. I think this says a lot in itself. It says that Mantle is not easier for devs, that it is more work. If it wasnt, it would be here before a complete DX 11 version of a game comes out.
You can try but this is a very hard reality to squirm out of. It says a lot in itself. Now if you months late on the most important Mantle title ever........ I guess you can fill in the rest. Because, yes! BF4 is the most important Mantle title because it is the very one they chose to showcase their new API on. And of course, pay no attention to the fact that its nowhere to be seen. This makes everything about Mantle questionable. If your launching Mantle versions several months after everyone else beats the game, would it really have that huge of a significance? Lucky for AMD, BF4 is a shooter whos multiplayer mode will be popular for awhile. But this is the exception not the rule. And if your this late on a game as important and as big as BF4, the Mantle highlight? Games of less importance will get much less money thrown at them. And pretty much all games out there are less important.

And what about the games that are Mantle first or Mantle exlusive? What about those console games that weren't going to get PC releases because of the time to port to DX? I can easily see a case where games like say, Last of Us get ported to Mantle simply because it's much easier and is harming microsoft/DX, which is pretty much exactly what Sony would want.

I just think way to many people have driven off to fantasy land here. Obvious what your wishes are but there are many many many cold hard facts right in front of you that for some reason no one seems to be bringing up. So quick to swear doom and gloom when the very the weapon at hand is misfiring right out of the gate. There is no way anyone should be sure of anything right now. Except the obvious: Mantle is extra work for developers, games, and AMD. If it was not we wouldnt be waiting on the BF4 matle patch that single handedly puts intel and nvidia out of business. You know, the one that is so much like console code that porting games over is so simple yet the DX11 PC version will be out ages before Mantle ever shows face.

Now, i am sure mantle is capable of better performance. As any close to the metal code could be. But everything else is marketing hype plus buckets of wild speculation

/rant end

Jesus that was a rant. If you'd just considered the possibility that Mantle wasn't ready you could have saved yourself a whole lot of trouble. This is early software, very much in development - of course a title like BF4 is going to cater to the majority at this stage, it's not like they were ever going to hold off on general release because of Mantle. But now that Mantle is hard coded into Frostbite there is little reason to believe that most games won't support it out of the box.

 

[NTMBK]

Yes exactly. And yes DirectX has stagnated, or perhaps bloated is a better term. Doesn't matter because either way, it leaves way, way too much performance on the table. The whole point of Mantle to me is to bring console efficiency to the PC world, and it's about time. Just imagine console efficiency with a high end (or two) GPUs.

Odd that people think DirectX will stand still.

Last time DirectX had serious competition, back in the 90s, it iterated rapidly and changed significantly over time. It's only since DirectX 9 that stagnation has really kicked in. If Mantle poses a genuine threat to DirectX then you bet that Microsoft will up their game- they don't want everyone developing for a multiplatform API.

Microsoft know exactly what would be needed for a Mantle style API- who do you think wrote the XBox One API? If they need to bring that back to the PC to kill off Linux games, you bet they will.

 

[frozentundra123456]

Snip

/rant end

I wont take up space with the entire quote, but you do have a point. Obviously mantle will be easier to implement once the code is in the game engine. But I do question if it will ever be as trivial to implement as some developers and AMD fans want it to be. It is surprising to me that it is taking so long to implement in their showcase game which has had years of work closely between AMD and Dice. As AMD supporters are saying, yes it is not ready yet, etc. so it takes a long time to implement. But they seem to want to have it both ways: claiming to know what impact mantle will have, but then saying it is not ready yet when asked why it is taking so long. To me you cant really have it both ways.

Mantle is a big unknown at this point, it has a lot of potential apparently, but as I said earlier, only time will tell the final impact.

 

[Paul98]

Well that would be a simple vendor id check on startup I suppose. I do believe that they will go with a selection option however - maybe Mantle breaks something that isn't shown in development and then at least they have the option of falling back to DX...

Another potential issue is DLC, namely having to code for DX and Mantle again at this stage (this is actually one of my biggest Mantle fears). It's probably not that much though is it...

DLC will take almost no or no mantle coding.

 

[psoomah]

The writing is on the wall. AMD is fully aware that discrete GPU's are going to basically cease to exist for regular consumers at some point (I'd say within 10 years). They are making a push to get in on the ground floor right now.

I'm thinking AMD doesn't need to make a push to get in on the iGPU ground floor considering they built the iGPU ground floor.

 

[psoomah]

I wont take up space with the entire quote, but you do have a point. Obviously mantle will be easier to implement once the code is in the game engine. But I do question if it will ever be as trivial to implement as some developers and AMD fans want it to be. It is surprising to me that it is taking so long to implement in their showcase game which has had years of work closely between AMD and Dice. As AMD supporters are saying, yes it is not ready yet, etc. so it takes a long time to implement. But they seem to want to have it both ways: claiming to know what impact mantle will have, but then saying it is not ready yet when asked why it is taking so long. To me you cant really have it both ways.

Mantle is a big unknown at this point, it has a lot of potential apparently, but as I said earlier, only time will tell the final impact.

It's not one size fits all.

For example this (slide 37) bullet point from Dice stating Plants vs. Zombies: Garden Warfare will have "Very different rendering compared to BF4". On the same Frostbite engine. The ease of implementing Mantle on a game will be variable on each engine and even within each engine. That said there's no getting around the fact that the head developers already on board with Mantle are very experienced heavy hitters and if they consider it worthwhile to put Mantle in their games, it's probably worthwhile.

We can have it both ways applying a little logic.

Logic dictates Mantle, with the sheer number of highly experienced developer's listed improvements that Mantle is providing, over 60 listed here alone, with some bringing substantial performance increases in themselves such as an order of magnitude increase in draw calls, is going to bring very substantial performance increases. It defies logic and common sense to think otherwise.

Logic also dictates it's an amazing feat just to bring a large complex AAA game like Battlefield 4 to the PS3, Xbox 360, PC, PS4 and Xbox One all within two weeks and that to question the validity of Mantle because EA/DICE was not also able to supply a completed PILOT build of Mantle, an entirely new API, for that extremely complex new game within that same time frame is flat out nonsensical. Seriously dude, it doesn't make sense.

 

[NTMBK]

It's not monotone. For example this (slide 37) bullet point from Dice stating Plants vs. Zombies: Garden Warfare will have "Very different rendering compared to BF4". On the same Frostbite engine. The ease of implementing Mantle on a game will be variable on each engine and even within each engine. That said there's no getting around the fact that the head developers already on board with Mantle are very experienced heavy hitters and if they consider it worthwhile to put Mantle in their games, it's probably worthwhile.

We can have it both ways applying a little logic. Logic dictates Mantle, with the sheer number of highly experienced developer's listed improvements that Mantle is providing, over 60 listed here alone, with some bringing substantial performance increases in themselves such as an order of magnitude increase in draw calls, is going to bring very substantial performance increases. It defies logic and common sense to think otherwise.

I thought my eyes were going funny reading that post, but it was just your crazy use of colors. 😵

 

[OCGuy]

An AMD employee is quoted as saying Kaveri may see a double-digit "performance" increase, in one game initially, if his optimistic view of Mantle is correct. Already quoted in this thread.

You can say that APUs are currently too weak to benefit, or you can say that Mantle is for more powerful CPU/GPU combos, but at this point you can't be talking about 20%+ FPS increases for APUs, unless you have some new data to post.

 

[psoomah]

I thought my eyes were going funny reading that post, but it was just your crazy use of colors. 😵

Lol, found myself somewhat discombobulated reading that too, so I changed it up to what I think is a workable use of the colors. :\

 

[SiliconWars]

An AMD employee is quoted as saying Kaveri may see a double-digit "performance" increase, in one game initially, if his optimistic view of Mantle is correct. Already quoted in this thread.

You can say that APUs are currently too weak to benefit, or you can say that Mantle is for more powerful CPU/GPU combos, but at this point you can't be talking about 20%+ FPS increases for APUs, unless you have some new data to post.

His comments were...http://www.techradar.com/news/compu...after-nvidia-graphics-with-kaveri-apu-1199636

"What I have heard from DICE is that what [Mantle] does with the discrete card is it equalizes the CPUs," Kozak explained. "It was only a couple of frames faster before because the CPU doesn't really play into things like that, but [Mantle] eliminates any gap. And essentially it does that by allowing the graphics card to do more, so it becomes the bottle neck."

"Personally my expectations are low," he said of a Mantle-plus-Kaveri combo. "But there is an Oxide demo here and they are seeing substantial speed-ups, beyond what anyone internally has guessed at. I'm optimistic it's going to be more than the 5% I'm hoping for and more towards the double digits."

Then at the end he said...

"I expect this one to get even better for us," he said, referring to the Battlefield 4 desktop demo. "We still have our engineers working with DICE on Kaveri optimizations. That's going on as we sit here, and we still have DICE working with Mantle. There's a two-step prong that's going to make this even better."

Seems clear to me that initially, Dice used Mantle mostly for equalizing the CPU's but there are ongoing optimizations. I mean they've already optimized the CPU's by this stage (2 weeks in) and have left 6 weeks out of the 8 to do the GPU optimizations.

 

[psoomah]

His comments were...http://www.techradar.com/news/compu...after-nvidia-graphics-with-kaveri-apu-1199636


Then at the end he said...

Seems clear to me that initially, Dice used Mantle mostly for equalizing the CPU's but there are ongoing optimizations. I mean they've already optimized the CPU's by this stage (2 weeks in) and have left 6 weeks out of the 8 to do the GPU optimizations.

HA!

I recognize that Indian looking dude in the first picture. He was the one fiddling with his phone alongside the desk in that leaked video from the hotel room where some people from the AMD team were playing with that Kaveri demo set-up and seeing how far they could push the settings and still have smooth motion. Turns out they could push it to the extreme setting plus some serious eye candy and still have smooth motion. Granted, it was one person walking/running through a scene, but still darned impressive and that was WITHOUT Mantle.

"Personally my expectations are low," he said of a Mantle-plus-Kaveri combo. "But there is an Oxide demo here and they are seeing substantial speed-ups, beyond what anyone internally has guessed at. I'm optimistic it's going to be more than the 5% I'm hoping for and more towards the double digits."

"more than the 5% I'm hoping for" - my sweet patootie!

[redacted]

Profanity isn't allowed in the technical forums.
-- stahlhart

 

[Atreidin]

Maybe this has been said already, but if Mantle lowers CPU usage in games, that also lowers CPU power and heat. If the rest of the system load stays the same, that could potentially be a significant effect, depending on the CPU and other variables. It would be cool to see power usage comparisons between DirectX and Mantle when games start arriving.

 

[krumme]

On the subject of Linux, I think Mantle here has even more potential. A leaned down Linux system is really, really fast.

And is called ps4 🙂

 

[OCGuy]

And is called ps4 🙂

??

 

[psoomah]

Maybe this has been said already, but if Mantle lowers CPU usage in games, that also lowers CPU power and heat. If the rest of the system load stays the same, that could potentially be a significant effect, depending on the CPU and other variables. It would be cool to see power usage comparisons between DirectX and Mantle when games start arriving.

With BF4 Mantle and then Kaveri reviewers are going to be patootie deep in unicorns sorting out all the newly significant permutations of CPU, APU and GPU that apply.

Best of all will be the return of AMD based rigs on the review sites. 😵

 

[psoomah]

??

here - "The PS4, which is scheduled to be released in November at the delightful price of $400, appears to run an operating system called Orbis OS, which is a modified version of FreeBSD 9.0. FreeBSD is a free version of BSD Unix that is generally fairly compatible with most Linux applications, and to the untrained eye a BSD-based system looks a lot like Linux. In theory, with a bit of work, this means you could almost take a PS4 game and run it on a Linux PC — but don’t get your hopes up for some kind of Linux gaming renaissance."

That can be considered modified by these (slides 31 and 32):


"Mantle and PlayStation 4 will drive our future Frostbite designs & optimizations"

and

"Mantle + SteamOS = powerful combination!"

Apparently Sony not officially supporting Mantle doesn't prevent DICE/EA from taking full advantage of the cross synergies that exist in actuality.

Doesn't sound very unicorny for Nvidia though.

 

[OCGuy]

Apparently Sony not officially supporting Mantle doesn't prevent DICE/EA from taking full advantage of the cross synergies that do in actuality exist.

Ah ok...you weren't clear, and I know the whole "Mantle isn't on consoles" thing has been beaten to death. Thanks.

 

[ShintaiDK]

And is called ps4 🙂

PS4 runs FreeBSD. Using FreeBSD, Linux etc doesnt magically make it faster. And no Mantle for PS4.