AMD Mantle update...

[vario]

I think you missed his whole point.

Kromosto said there was no gains in BF4 using mantle with GCN 1.0 cards ,280X is a tahiti based card, thats GCN 1.0.The tests i provided say otherwise...
So what did i miss ?

[kromosto]

yes i admit for lower cpu's it is better then i know (most of the community doesn't use that lower cpus so never got feed back from that kind) but still what i am trying to ask is didn't answered. this problem generally occurs because i can't express myself clear have to learn to tidy up my mind before asking questions i guess. but the site you gave results will help me to ask my question better i think.

from the same site you gave test results for 290x:
4670k stock with dx11: 79.8 mantle 108.3
4670k oc with dx11: 96.3 mantle 110.1

stock dx11 to mantle %35
oc dx11 to mantle %14
stock dx11 to oc dx11 % 20
stock mantle to oc mantle %1

it is obvious that fps gains comes from lesser overheads. if 290x gets this amount of performance from lesser overheads for stock and oced 4670k why 280x gets only %1 or %2 ? cpus are same driver is same mantle api is same. what optimizations can make this difference? what i want to know is this.

EDIT: Also what should i think about stock oc mantle difference is only %1 but dx11 %20. Mantle is that good to lessen cpu overheads and still supports and emphasizes my main question.

[vario]

yes i admit for lower cpu's it is better then i know (most of the community doesn't use that lower cpus so never got feed back from that kind) but still what i am trying to ask is didn't answered. this problem generally occurs because i can't express myself clear have to learn to tidy up my mind before asking questions i guess. but the site you gave results will help me to ask my question better i think.

from the same site you gave test results for 290x:
4670k stock with dx11: 79.8 mantle 108.3
4670k oc with dx11: 96.3 mantle 110.1

stock dx11 to mantle %35
oc dx11 to mantle %14
stock dx11 to oc dx11 % 20
stock mantle to oc mantle %1

it is obvious that fps gains comes from lesser overheads. if 290x gets this amount of performance from lesser overheads for stock and oced 4670k why 280x gets only %1 or %2 ? cpus are same driver is same mantle api is same. what optimizations can make this difference? what i want to know is this.

EDIT: Also what should i think about stock oc mantle difference is only %1 but dx11 %20. Mantle is that good to lessen cpu overheads and still supports and emphasizes my main question.

Obviously this kind of question can only be answered in depth by an amd engineer.However, simple part of this would be the fact that tahiti based cards are siginificantly slower to start with so they have less room to "work with" using mantle.So it can mean that it is in big part just from the fact that tahiti cards get saturated sooner and "the blame" shifts from CPU to GPU part.Have a look at this, its from the same site, just in this instance they set high MSAA to bring 290X down:


So when the GPU is being hammered down, mantle doesnt bring high gains even on GCN 1.1.But you have to keep in mind that hawaii has more computing power, it has faster memory access and was built in mind with mantle.
As i said earlier , this thing is gonna be hard to quantify, it will all depend on the hardware used, on the settings used, and even with same hardware and same gfx settings it can be almost no difference in scene where there arent LOADS of objects on the screen to a MASSIVE difference when there are thousands of them.Thats why i think amd works with starswarm, they want to have the capability to show what it CAN do.Lets remember that mantle isnt only for same games and same scenes like on dx just with less overhead, its also for things that DX ultimately wouldnt be able to handle.

[kromosto]

yes msaa results indicate what you are saying. so based on all of these if we make another assumption in a system with stock 4670k and 280x, 280x is the bottleneck.

[vario]

yes msaa results indicate what you are saying. so based on all of these if we make another assumption in a system with stock 4670k and 280x, 280x is the bottleneck.

That is a wide generalization, as i said earlier, its gonna be helluva tricky because that statement CAN be true, but only on specific settings ,with specific things going on the screen at the moment taking the tests.Also, its the first iteration, beta in fact, with first title supporting it (beta in fact) so theres probably a long way to go in terms of mantle and what it can bring on the table.
However its pretty logical that the BIGGEST improvements RIGHT OF THE BAT , will see people with weakish CPUs, and highest performing GPUs ,AND crossfire on all CPUs be it weak or fast ones, .Thats with titles made with DX in mind doing just a port with no added content, things could change dramatically with mantle only settings or titles made with mantle in mind.

[kromosto]

I am not interested if it is beta or not my onlu concern is a good explanation of 290x and 280x performance gain difference from amd. Actually i dont believe i can get one but i stuck at that point.

[kromosto]

...

[STEvil]

Obviously this kind of question can only be answered in depth by an amd engineer.However, simple part of this would be the fact that tahiti based cards are siginificantly slower to start with so they have less room to "work with" using mantle.So it can mean that it is in big part just from the fact that tahiti cards get saturated sooner and "the blame" shifts from CPU to GPU part.Have a look at this, its from the same site, just in this instance they set high MSAA to bring 290X down:


So when the GPU is being hammered down, mantle doesnt bring high gains even on GCN 1.1.But you have to keep in mind that hawaii has more computing power, it has faster memory access and was built in mind with mantle.
As i said earlier , this thing is gonna be hard to quantify, it will all depend on the hardware used, on the settings used, and even with same hardware and same gfx settings it can be almost no difference in scene where there arent LOADS of objects on the screen to a MASSIVE difference when there are thousands of them.Thats why i think amd works with starswarm, they want to have the capability to show what it CAN do.Lets remember that mantle isnt only for same games and same scenes like on dx just with less overhead, its also for things that DX ultimately wouldnt be able to handle.

Now add extra CPU time into the equation .

In these benchmarks you are seeing the card get maxed out, but you arent seeing (very well) the spare free cycles the CPU has with which to do things like physics or anything else its can now do.

[kromosto]

Now add extra CPU time into the equation .

In these benchmarks you are seeing the card get maxed out, but you arent seeing (very well) the spare free cycles the CPU has with which to do things like physics or anything else its can now do.

yes that will be a good input too.

[tajoh111]

One thing that hasn't been answered yet is how much work or extra development time does it take to implement mantle?

Depending on how much work it takes, its going to really going to affect developer adoption of it and affect if AMD has to pay developers to use it. If it's easy on the other hand, then developers are more likely to use it. But regardless, I think at this point, I think we can say it takes a significant amount of effort.

[Ace123]

As I understand it, developers had requested this. I cannot imagine they would ask for something intentionally to create more work for themselves for little gain.

[kromosto]

I don't think mantle brings a very new way of 3d programming beside some low level access to hardware for optimizations so i believe adoption is not a big problem.

[Iconyu]

As I understand it, developers had requested this. I cannot imagine they would ask for something intentionally to create more work for themselves for little gain.

Exactly, DICE approached both Intel and Nvidia to create Mantle but they said no, only AMD said yes. I've not seen any developers that have actually used Mantle say anything against it, so I imagine the gains outweigh the costs by a large margin.

[tajoh111]

As I understand it, developers had requested this. I cannot imagine they would ask for something intentionally to create more work for themselves for little gain.

I think the ideal for developers is Directx 11.1 to be dead and developers have to only deal with a low level API and have to program only for it. As it is now, developers have to code for both directx 11.1(this they have to do) and a mantle version. Considering that Dice has been working on mantle before BF4 release till now and even now it is still very beta, there seems to be considerable work done for each game that uses it. No one wants to work for free, particularly when the only hardware that supports it makes up the minority share compared to Nvidia based cards + pre GCN hardware.

Some developers were asking for it, but not all. And even if they wanted it, it doesn't mean they will work for nothing.

[kromosto]

if unreal make a mantle engine then it is over.

[tajoh111]

if unreal make a mantle engine then it is over.

If Unreal was going to make something, i think they would go out on to their own and make their own low level API. Unreal is a market leader when it comes to these things and not a follower. And if unreal made one, I am not sure if it would be open, but with the Mantle threat, I bet nvidia would be willing to open up its close to 5 billion dollar cash reserves and try to hammer out a deal.

[kromosto]

i don't think mantle didn't take unreal's attention. if they be quick to respond to problems and make a good mantle 2.0 in a year i am sure we can see a mantle unreal engine.

also of course nvidia unreal pact for a new api is a high probability but mantle has a very huge being early advantage.

[tajoh111]

i don't think mantle didn't take unreal's attention. if they be quick to respond to problems and make a good mantle 2.0 in a year i am sure we can see a mantle unreal engine.

also of course nvidia unreal pact for a new api is a high probability but mantle has a very huge being early advantage.

I think unreal could develop something quicker and better than mantle(atleast when it comes to bugs) for two reasons. They have far more experience and programmings with this type of thing. Secondly, they don't have EA breathing down their backs so they could divert more resources and attention on developing it. EA crazy deadlines and pushing DICE for their own agenda wrecked BF4 and delayed Mantle.

[kromosto]

developing api and engine have some similarities but actually they are very different and no way unreal can make it by itself only. they must work with a gpu manufacturer.

[MaddMuppet]

Finally a fix for the hitching

PC Game Update Feb 13

Mantle
-Fix for a crash that would occur when activating full screen in portrait mode
-Fix for stuttering that could appear during video sequences on multi-GPU PCs
-Fix for a memory system leak that could cause stalls, which would result in frames taking longer to process
-Reduced the amount of stalls that occurred when running with high graphics setting that require more GPU memory than is currently available
-Fixed screenshots on multi-GPU PCs

-General stability improvements
-Fix for an issue where spawning into, or switching to, a gunner seat in an IFV/MBT sometimes could cause the game to crash
-Fix for missing sound in Team/Squad Deathmatch
-Fix for an issue in the Defuse game mode, where a bomb carrier would be permanently spotted
-Decreased the rate at which the kill card would incorrectly display 0 health, while the enemy was clearly alive
-Fixed an invisible wall that was incorrectly positioned in one of the fallen concrete pipes on Zavod 311
-Fix for an issue where bullet impact sounds weren’t properly matching the actual number of impacts
-Fix for an issue where the “Draw” message would not display on-screen once a Conquest round ended with both teams having the same amount of tickets
-Fix for an issue where long IDs wouldn’t scroll on dog tags
-Fix for missing grass physics in terrain

[Heinz68]

Nice to see new Batlefield 4 game patch fix including some for Mantle ^^^

Having new system update including CPU, motherboard and graphic cards (in my signature) I decided not to use the Beta 14.11 drivers for now, looking up for more stability while checking the overclocking and benches.
The 13.12 drivers work great for me, guess will try out the next Beta drivers.

[Lightman]

Confirmed to fix stutter on HD 7970!
My brother just did 45 min stint in BF4 MP Mantle edition without any hicups!
Everything Ultra except MSAA 1x and it is fluid on his A10-5800K oc to 3.9/4.3GHz. On that CPU Mantle rocks

[kromosto]

Confirmed to fix stutter on HD 7970!
My brother just did 45 min stint in BF4 MP Mantle edition without any hicups!
Everything Ultra except MSAA 1x and it is fluid on his A10-5800K oc to 3.9/4.3GHz. On that CPU Mantle rocks

can't wait to go home and try this new patch

[kromosto]

And now with my recently bought 280x cfed with good old 7970 lightning

[vario]

And now with my recently bought 280x cfed with good old 7970 lightning

Unreal engine is made by Epic Games, also why in the hell they would want to make an api ? They have nothing to gain here, also im pretty sure they cant do anything serious without them engineering their own gpus.
but yea, in order to really get off the ground with mantle AMD needs epic to be on board.