Mantle launches this month in BF4.

Status
Not open for further replies.
Glad I didn't buy AMD for this.

Of course in a Tri-CF system the CPU will be bottlenecked, but that doesn't occur in single card setups with Intel CPUs which over 90% of gamers have.
 
Last edited:
Shark974 said:
It does help the GPU too. Just, it looks to only be a few percent difference so far when GPU limited.
Click to expand...

I'm not sure if you're trying to argue against my "in essence" or not. The improvement in non-CPU limited scenarios shown so far seem to be in line the up to 10% (against DX) I personally predicted on these forums when Mantle was announced.

The BF4 improvements in multi-GPU are probably related to direct and specific multi-GPU optimizations made by DICE to their renderer when using Mantle, so I wouldn't expect to see that in general. That is to say, to get that kind of improvement the developer will have to dedicate significant resource to it.
 
polonyc2 said:
Mantle was released at 4:00 AM EST...stealth late night release AMD?...anyhow...

the biggest gains will be achieved on systems with mid-to-low end CPUs that are essentially CPU constrained. According to AMD, "Mantle makes the most significant (and we do mean significant) performance benefit for gaming scenarios that are CPU-limited."

For those of you with beefy CPUs where the GPU is the limiting factor, AMD says don't get your hopes up. It writes, "An API change is unlikely to make a drastic change in these scenarios, as GPU resources are being maximally utilized in a fashion that is difficult to improve at the API-level."

this honestly is a major letdown as it makes no sense...to run Battlefield 4 you need a pretty hefty CPU/GPU...so are they saying that an older 'low end' CPU will now run BF4 just as good as a high-end enthusiast CPU with Mantle enabled?
Click to expand...
What's so difficult to understand? In CPU-bound scenarios(usually low-end CPU and high-end GPU, but could also be high-end CPU and multi-GPU), you'll see tremendous gains in performance. In GPU-bound scenarios, i.e. when you've both a high end CPU and GPU, or a high end CPU and low-end GPU, you'll see minor improvements.
 
polonyc2 said:
Mantle was released at 4:00 AM EST...stealth late night release AMD?...anyhow...

the biggest gains will be achieved on systems with mid-to-low end CPUs that are essentially CPU constrained. According to AMD, "Mantle makes the most significant (and we do mean significant) performance benefit for gaming scenarios that are CPU-limited."

For those of you with beefy CPUs where the GPU is the limiting factor, AMD says don't get your hopes up. It writes, "An API change is unlikely to make a drastic change in these scenarios, as GPU resources are being maximally utilized in a fashion that is difficult to improve at the API-level."

this honestly is a major letdown as it makes no sense...to run Battlefield 4 you need a pretty hefty CPU/GPU...so are they saying that an older 'low end' CPU will now run BF4 just as good as a high-end enthusiast CPU with Mantle enabled?
Click to expand...
As I mentioned in the previous page:
With the links above, we now know Mantle is the following:

  • Benefits with highly threaded processors
  • Benefits CPU-limited scenarios
  • Benefits low to mid-range processors
  • GPU-limited in high-end processors
In another thread here on [H], I suggested that with Mantle, we're probably going to see a "blurring the lines" so-to-speak between CPU architecture and CPU models. Previously, it was all about getting the most high-end midrange or enthusiast processor your money can buy to play the latest PC game released. With Mantle and Mantle-supported games, you can quite literally go with a lower-end quad-core or dual core/4-thread processor, and with the money you save put it into a higher-end video card.

When asynchronous Crossfire and the ability to split physics compute and graphics rendering between an AMD APU IGP and an AMD GCN-based GPU is supported, this may further the cause.

Will it make high-end processors especially those from Intel obsolete?

No, absolutely not. However, those on tight budgets and want a gaming PC will no longer have to try to get the highest end processor. The highest end enthusiast processors will continue to be niche products serving the needs of those that need the extra CPU power outside of gaming and games that do not have Mantle support baked in.

I believe, in my honest opinion, Mantle is going to further explode the budget to mainstream market, as well as the mobile laptop market beyond the midrange and enthusiast market of processors and computer systems.

For the enthusiasts that want the best out of nearly everything, they will see gains in performance when moving from one high-end video card to one in the next generation as it now shifts the performance limit from CPU to GPU. Enthusiasts will also have to consider highly multi-threaded processors starting with the Intel 4 core/8 thread variety up to the 6 core/12-thread models. Why? Mantle even benefits from more threads made available to the API.

Lastly, it's going to take developers to further push themselves to support Mantle because I see a lot of benefits to this. Think about it: If a person is using a non-AMD APU high-end Intel processor, there is going to be a lot of LEFTOVER CPU cycles going unused. I can now see scenarios where shifting something like CPU-based PhysX-calculations to feed CPU cores in non-Nvidia systems being more and more plausible, while AMD GCN GPUs does the grunt work of graphics rendering and processing graphical and particle effects.

It's going to be an entirely new way how we're going to be looking at how computer gaming systems are built, and how computer games are going to be programmed in the future.

Sidenote: If AMD adds Linux support for Mantle, lower cost Valve SteamOS-based computers may benefit greatly from this because the lower priced Steam Machines from OEMs are going to have lower end processors.
 
Yeah but then again, if you're going to buy a 290,you're likely to go for the best cpu too. People that buy apu usually don't buy gpu because they mostly only game very light.

I want to see an Apu vs a i7 3770+ with a 290x, directx vs Mantle, that should be cool to see
 
Ryoohki360 said:
Yeah but then again, if you're going to buy a 290,you're likely to go for the best cpu too. People that buy apu usually don't buy gpu because they mostly only game very light.

I want to see an Apu vs a i7 3770+ with a 290x, directx vs Mantle, that should be cool to see
Click to expand...
Yeah, that's another test configuration I didn't consider.

An A10-7850K + 290X vs. a 3770K/4770K + 290X would be interesting to see what results come out of that.

That and compare 4 threads vs. 4 threads (HyperThreading off) to test as well. We know, based on the Oxide demo/presentation that Mantle benefits with more threads available to it. So, seeing a test between a mainstream 4-core processor versus a 4 core/8-thread processor may see different results.
 
Raap said:
What's so difficult to understand? In CPU-bound scenarios(usually low-end CPU and high-end GPU, but could also be high-end CPU and multi-GPU), you'll see tremendous gains in performance. In GPU-bound scenarios, i.e. when you've both a high end CPU and GPU, or a high end CPU and low-end GPU, you'll see minor improvements.
Click to expand...

again having Battlefield 4 as the Mantle showcase title makes no sense as it pretty much requires a mid to high end CPU and GPU...as far as Mantle + multi-GPU setups it is still a small minority in the big picture...why is Mantle so limited as an API?...I thought it was going to work like DirectX...Mantle doesn't really seem like a DX killer at all

octoberasian said:
In another thread here on [H], I suggested that with Mantle, we're probably going to see a "blurring the lines" so-to-speak between CPU architecture and CPU models. Previously, it was all about getting the most high-end midrange or enthusiast processor your money can buy to play the latest PC game released. With Mantle and Mantle-supported games, you can quite literally go with a lower-end quad-core or dual core/4-thread processor, and with the money you save put it into a higher-end video card
Click to expand...

that's a good point
 
Ryoohki360 said:
Yeah but then again, if you're going to buy a 290,you're likely to go for the best cpu too. People that buy apu usually don't buy gpu because they mostly only game very light.

I want to see an Apu vs a i7 3770+ with a 290x, directx vs Mantle, that should be cool to see
Click to expand...

Probably vast majority of people who buy high end gpu's do not have high end processors. I haven't upgraded my cpu in well over 2 years, but I've changed gpu three times ;/.
 
Just a point worth considering - BF4 is an engine that was built and designed for DX and then later ported to Mantle, and it also requires a high-end CPU due to the physics calculations in the game. So BF4, while a good test case to prove that Mantle works, is probably not going to show the same benefits than an engine built around Mantle support from the get-go, or a game that is less CPU bound graphically will show. We'll have to see. In any case, I don't find these results disappointing. Some people need to temper their expectations :)
 
Ryoohki360 said:
Yeah but then again, if you're going to buy a 290,you're likely to go for the best cpu too. People that buy apu usually don't buy gpu because they mostly only game very light.

I want to see an Apu vs a i7 3770+ with a 290x, directx vs Mantle, that should be cool to see
Click to expand...

Eh, what about the large subset of people on sandy bridge or using older 920 style processors with a new gcn GPU hoping to stretch out their CPU lifetime.

It's what I'm doing.

Even if the gains are 10%, ill take it. I received roughly 10% by putting my 290 on water and heavily over volting... For a quick DL, its great.

Launch pricing on 290s made them an easily affordable card available to many mid ranged gamers.
 
polonyc2 said:
again having Battlefield 4 as the Mantle showcase title makes no sense as most enthusiasts will already have a mid to high end CPU and GPU...multi-GPU setups are still a small minority in the big picture...why is Mantle so limited as an API?...I thought it was going to work like DirectX...Mantle doesn't really seem like a DX killer at all
Click to expand...

The majority of users are midrange to low-end. Thats the bulk of gamers. In Dice's test scenario 2, it shows an upper midrange system (7970+8350) getting a 25% increase in BF4 multiplayer just by changing from DX 11.2 to AMD Mantle. Considering that BF4 is highly optimized already for DX 11.2, its quite a lot. And for free.

Test case 2: Standard 64-player multiplayer
CPU: AMD FX-8350, 8 cores @ 4 GHz
GPU: AMD Radeon 7970 3 GB (AMD will add support for the AMD Radeon™ HD 7970 in a later stage of Mantle’s release schedule, learn more)
Settings: 1080p ULTRA 1x MSAA
OS: Windows 8 64-bit
Level: Siege of Shanghai
Level was tested with 64 “pseudo players” that we have for our own internal testing that simulates heavy game workload that we have in multiplayer in order to get more deterministic results compared to full real multiplayer. 64 players on the large Battlefield levels is really demanding of the CPU so this test case is primarily CPU-bound.
Result: 18.87 ms/f -> 15.08 ms/f = 25.1% faster
Click to expand...

Its not like developers aren't capable of making games more CPU bound or that games wouldn't benefit for features like higher draw distance and other things that has been suggested using the increased possibilities of more draw calls under Mantle.

Don't forget, this is only the beginning. Already it looks very promising!

AMD white paper said:
It¡¦s also prudent to note that Mantle is still in the beta phase and may not reflect the full performance we might be able to achieve through the optimization time we¡¦ll be investing in the months ahead. And, as developers are still familiarizing themselves with Mantle and its relationship to Graphics Core Next, they may not have capitalized on all available opportunities for optimizations¡Xbut that will come with time.
Click to expand...
http://www.guru3d.com/news_story/am..._end_cpusnot_mainstream_and_high_end_pcs.html
 
johan anderson said on his twitter they got working 7000 series drivers. (guess amd still gotta vet em first)

its not all bout fps, gotta look at frame times and situation where you'd get stutters or slow downs from
drawcalls maxing out.

http://wccftech.com/dice-officially-launches-mantle-update-battlefield-4-123-gb-size/
ingame pictures with frame graph.

Mantle release is like an alpha not even a beta, more waiting for it to be polished :eek:

at least frame pacing phase 2 is outs for ppl who didn't give a crap bout mantle woooooo

45% on apus is awesome tho ,also if this is the future buying a mid range cpu + top of the line amd card will mean getting the max out of it not so bad.

yet still more waiting, get your shit together amd!
 
gordon151 said:
Probably vast majority of people who buy high end gpu's do not have high end processors. I haven't upgraded my cpu in well over 2 years, but I've changed gpu three times ;/.
Click to expand...

Arcygenical said:
Eh, what about the large subset of people on sandy bridge or using older 920 style processors with a new gcn GPU hoping to stretch out their CPU lifetime.

It's what I'm doing.

Even if the gains are 10%, ill take it. I received roughly 10% by putting my 290 on water and heavily over volting... For a quick DL, its great.

Launch pricing on 290s made them an easily affordable card available to many mid ranged gamers.
Click to expand...
I think a lot of gamers on the PC are in this scenario-- upgraded to the latest GPU, but have a processor as old as the Core 2 Duo/Quad or Phenom II X4 or X6.

Why?

Software hasn't shown a real need for the best and most expensive processor out there outside of gaming, video encoding, and 3D rendering. Time and time again, I have always said that we are going to see less and less benefits using higher end processors when our older processors from X-years ago can still browse the web, watch Youtube, and use Word.

But, for gaming... entirely different story depending on the game.
Rizen said:
Just a point worth considering - BF4 is an engine that was built and designed for DX and then later ported to Mantle, and it also requires a high-end CPU due to the physics calculations in the game. So BF4, while a good test case to prove that Mantle works, is probably not going to show the same benefits than an engine built around Mantle support from the get-go, or a game that is less CPU bound graphically will show. We'll have to see. In any case, I don't find these results disappointing. Some people need to temper their expectations :)
Click to expand...
Yeah, that's the thing and what I suggested this scenario come out of a game with Mantle-engine built from the ground up instead of ported:
There are going to be leftover CPU cycles.

So, it got me thinking: Why not use it for physics-based particle effects and something like PhysX on the CPU? (Or, other non-PhysX based physics engine for that matter). Mantle is freeing up the CPU more and more with more threads, but as we see in the links already posted here in the past few pages, higher end processors show little benefit. Why is that? One, the GPU is still going to be GPU-bound limited. Second, there has to be a lot of leftover processing power that can be used for other things in the game.

In such a scenario, game developers may have to consider using that extra power on higher end processors for physics and AI now.
 
This is only the first beta release and improvements are massive. My system is similar to the third case scenario. I have a 6 core i7 980@4Ghz with a pair of Sapphire Tri-X OC 290Xs.

I'd download and test BF4 myself but I'd rather have the cards working on paying for themselves mining lol. :D
 
octoberasian said:
Yeah, that's the thing and what I suggested this scenario come out of a game with Mantle-engine built from the ground up instead of ported:
There are going to be leftover CPU cycles.

So, it got me thinking: Why not use it for physics-based particle effects and something like PhysX on the CPU? (Or, other non-PhysX based physics engine for that matter). Mantle is freeing up the CPU more and more with more threads, but as we see in the links already posted here in the past few pages, higher end processors show little benefit. Why is that? One, the GPU is still going to be GPU-bound limited. Second, there has to be a lot of leftover processing power that can be used for other things in the game.

In such a scenario, game developers may have to consider using that extra power on higher end processors for physics and AI now.
Click to expand...
Exactly. I think the real benefits will be what game developers can accomplish with extra CPU cycles once the API overhead is removed. Ports of existing games aren't going to be able to take advantage of this, but games built with Mantle support in mind will.

I really hope we start to see a bigger push with physics and AI, especially in RPGs. It makes the world so much more authentic.
 
Edit: Whoa, posted at the same exact time lol ^^

AMD @Facebook said:
AMD Catalyst 14.1 Beta will support ALL desktop GCN products, though we are working with EA to further optimize performance on 280X, 270X, HD 7000 and HD 8000. We have asked EA to update their blog to reflect this.
Click to expand...

News keep getting better :)
 
Rizen said:
Just a point worth considering - BF4 is an engine that was built and designed for DX and then later ported to Mantle... So BF4, while a good test case to prove that Mantle works, is probably not going to show the same benefits than an engine built around Mantle support from the get-go
Click to expand...
Considering Johan Andersson co-developed Mantle, I think BF4's implementation is probably reference-quality.
 
wonderfield said:
Considering Johan Andersson co-developed Mantle, I think BF4's implementation is probably reference-quality.
Click to expand...
That's a fair point. But I do know that Frostbite has been in continuous development since at least 2007. And BF4 was developed based on Frostbite before Mantle was released. So I don't know if there is more that DICE could have done with the engine/game if they had built it with Mantle in mind. Does that make sense? I might be totally wrong. I'm not a software engineer. It just seems based on what I understand of Mantle that the benefits lend themselves to more creative freedom from removing overhead which might not have been part of their design when they were planning BF4.
 
from http://www.guru3d.com/news_story/update_on_amd_catalyst_14_1_mantle_driver_release.html

AMD:

During the QA process, we identified an installation issue in AMD Catalyst 14.1 that renders it unsuitable for distribution for testing in your labs. We are testing a new build overnight and will update you on its status by mid-afternoon Eastern Standard Time on January 31. We apologize for the delay and appreciate your patience and understanding.

***This is the press-driver as we have just learned, AMD promised the media at least 24 hours with the driver prior to release to do some internal testing. We do hope it'll be out fairly soon. But at this stage, nothing is certain whatsoever.
Click to expand...

Takes a bit longer but installation issues are news to me.
 
Venomous said:
This is total bullshit man. Like the press has any business testing drivers... No wonder why AMD has issues with drivers if they are relying on press and news to " test " their drivers. What a joke.
Click to expand...
No, they're giving them time to prepare articles and news coverage, it's not a QA session.
 
Venomous said:
This is total bullshit man. Like the press has any business testing drivers... No wonder why AMD has issues with drivers if they are relying on press and news to " test " their drivers. What a joke.
Click to expand...

Uhm, I think Rizen is merely stating what press outlets have stated. They were told by AMD that they would get a 24 hour advance peek before the public - AMD stated this, it's not something Rizen wants (i'm assuming). Personally I think AMD Should just adhere to giving their consumers the driver as promised in January, even if it means the press doesn't get an advanced peek. That's how I feel about it - the consumer gets precedence. I think that if the driver is done (as an example) at 3PM today, that AMD should stick to their promise - customers get first priority. Even if it means press doesn't get an advanced look at it - they can still do their reviews and what not, but they just won't have an advanced look prior to the public. Which is fine, because if customers were promised January - that should happen IMO even if it's at the expense of the press.

But, the main point being, Rizen is just stating what AMD said. AMD emailed their press to alert them of a 24 hour advance peek - Guru3d posted this on their front page. That isn't necessarily what users WANT to happen. I think everyone here, Rizen included, would like to get the driver ASAP. What happens at this point is anyones guess.
 
xoleras said:
Uhm, I think Rizen is merely stating what press outlets have stated. They were told by AMD that they would get a 24 hour advance peek before the public - AMD stated this, it's not something Rizen wants (i'm assuming). Personally I think AMD Should just adhere to giving their consumers the driver as promised in January, even if it means the press doesn't get an advanced peek. That's how I feel about it - the consumer gets precedence. I think that if the driver is done (as an example) at 3PM today, that AMD should stick to their promise - customers get first priority. Even if it means press doesn't get an advanced look at it - they can still do their reviews and what not, but they just won't have an advanced look prior to the public. Which is fine, because if customers were promised January - that should happen IMO even if it's at the expense of the press.

But, the main point being, Rizen is just stating what AMD said. AMD emailed their press to alert them of a 24 hour advance peek - Guru3d posted this on their front page. That isn't necessarily what users WANT to happen. I think everyone here, Rizen included, would like to get the driver ASAP. What happens at this point is anyones guess.
Click to expand...


I'm aware of that... It's just that I don't agree with it.
 
What a clusterfuck. At least AMD identified the issue prior to release — something they weren't fortunate enough to have done with the infamous Rage driver.
 
eloj said:
The BF4 improvements in multi-GPU are probably related to direct and specific multi-GPU optimizations made by DICE to their renderer when using Mantle, so I wouldn't expect to see that in general. That is to say, to get that kind of improvement the developer will have to dedicate significant resource to it.
Click to expand...

Sweclockers seems to confirm that this is the case. The current Mantle implementation will not have generic CF support, it's up to each developer to implement something like AFR.
 
Venomous said:
I'm aware of that... It's just that I don't agree with it.
Click to expand...
I don't necessarily either, I'm just reporting to what various hardware sites stated. AMD isn't using the press as QA like you suggested, though, I think it's just for media relations.
 
eloj said:
Sweclockers seems to confirm that this is the case. The current Mantle implementation will not have generic CF support, it's up to each developer to implement something like AFR.
Click to expand...

I'm not sure you'll ever see implicit multi GPU in mantle. That kind of defeats the purpose.
 
Venomous said:
This is total bullshit man. Like the press has any business testing drivers... No wonder why AMD has issues with drivers if they are relying on press and news to " test " their drivers. What a joke.
Click to expand...

They are not relying on the press to test their drivers. They are relying on the press putting out news stories to promote AMD products. This is a marketing move.
 
PRIME1 said:
They are not relying on the press to test their drivers. They are relying on the press putting out news stories to promote AMD products. This is a marketing move.
Click to expand...

They should be doing a press release on how bad they screwed this up and how customers keep being promised something, but get bait and switched. I mean now that I know gcn 1.0 cards weren't even considered with this first driver, it more or less says AMD only cares about media exposure and 290/X cards. I wish this would of been said from the get go instead of all of these * asterisks being listed such as:

* no cfx support
* driver not optimized for gcn 1.0 cards
* driver released to press/review sites first

I mean who is the one making these decisions? It looks really really bad.
 
Venomous said:
They should be doing a press release on how bad they screwed this up and how customers keep being promised something, but get bait and switched. I mean now that I know gcn 1.0 cards weren't even considered with this first driver, it more or less says AMD only cares about media exposure and 290/X cards. I wish this would of been said from the get go instead of all of these * asterisks being listed such as:

* no cfx support
* driver not optimized for gcn 1.0 cards
* driver released to press/review sites first

I mean who is the one making these decisions? It looks really really bad.
Click to expand...


Whaaa? DICE used CFX in one of their testing scenarios and gained %58.
 
Status
Not open for further replies.
Back
Top