TR Forums

clone wrote:

While doing a complete port to PC wouldn't be a mirror of the console code it'd be stunningly close with mirror images of large portions.

Mantle ports will be the "lazy" ports... The ones devs do just because they can.

"Real" ports will still need to support non-GCN cards which includes a large number of Radeon 6xxx and below and obviously nVidia cards. Steam HW Survey (Aug) - I don't know how "Other" breaks down, but GCN cards - Radeon 7xxx series - only accounts for like 5%... There's a reason why BF4 will still have a DX code path, they don't want to just throw away a huge part of the market. This number will obviously go up, but that will take a while as this is a chicken and egg scenario.

clone wrote:

it's mentioned in the TR "live" blogpost and if you type in a general search using the keywords Mantle is an open API you find several links, anyway here is one that mentions it.

I've seen it mentioned but I was wondering what the exact quote was. I've since found it. As others have said, AMD needs to define "open." I'm prety sure we'll learn more at the up-coming developer confrence.

PixelArmy wrote:

Mantle ports will be the "lazy" ports... The ones devs do just because they can.

With the added benefit of possibly being cross-platform (Console, Windows, OSX and Linux/SteamMachine).
Another thought on this, the consoles are what, 7700 class-isg GPU's? So, it's possible that low/mid-end, inexpensive GPU's can hit the same target and experience as the consonle on almost any OS. That's not a big deal until you think about the fact that within a few years, AMD will likely have APU's that are at least as good, if not better, than the console hardware in SFF, maybe even laptops.

.

clone wrote:

substitute lazy for more efficient and inexpensive and they'd do it because they already did.

trying to pass off a better optimized more efficiently written software application as lazy just because the low level API is shared with console is flawed..... Mantle is supposed to make any of the best D3D optimized apps less efficient by it's very nature.... that's not lazy, if it takes off it'll be amazing as developers have more direct access than they've never had before.

I didn't say Mantle programming was lazy, in fact I agree it will take skill, but again that's not what I said. I said the porting as you described was lazy. Specifically the process of porting an existing console Mantle game to PC. The idea that you've completed a Mantle version for the console and bam! Most of your PC port will just mirror large portions of code... That is only true if you are forsaking the majority of the market you're porting to and only creating the port for PC GCN users. Until GCN numbers get up there, you will have to have a DX or OpenGL version.

Look at the hardware numbers for GCN on steam! http://store.steampowered.com/hwsurvey

Repackaging 90% of your code to create a PC port that will only work for 10% of the PC market (GCN users and that's being generous) sounds doesn't sound like much effort to me.

(Intel supporting Mantle if it takes off a little?!? At risk of sounding like a broken record, GCN, GCN, GCN, I think you're missing the part where this is tied to GCN. No nVidia, no intel, no radeon 5xxx, no radeon 6xxx).

I think it's likely that AMD will provide tools that let you DX/Mantle code at the same time.

Probably horribly unoptimized DX code, but functional so that devs don't have to duplicate their work (or can just focus on optimization).

PixelArmy wrote:

clone wrote:

substitute lazy for more efficient and inexpensive and they'd do it because they already did.

trying to pass off a better optimized more efficiently written software application as lazy just because the low level API is shared with console is flawed..... Mantle is supposed to make any of the best D3D optimized apps less efficient by it's very nature.... that's not lazy, if it takes off it'll be amazing as developers have more direct access than they've never had before.

I didn't say Mantle programming was lazy, in fact I agree it will take skill, but again that's not what I said. I said the porting as you described was lazy. Specifically the process of porting an existing console Mantle game to PC. The idea that you've completed a Mantle version for the console and bam! Most of your PC port will just mirror large portions of code... That is only true if you are forsaking the majority of the market you're porting to and only creating the port for PC GCN users. Until GCN numbers get up there, you will have to have a DX or OpenGL version.

Look at the hardware numbers for GCN on steam! http://store.steampowered.com/hwsurvey

Repackaging 90% of your code to create a PC port that will only work for 10% of the PC market (GCN users and that's being generous) sounds doesn't sound like much effort to me.

(Intel supporting Mantle if it takes off a little?!? At risk of sounding like a broken record, GCN, GCN, GCN, I think you're missing the part where this is tied to GCN. No nVidia, no intel, no radeon 5xxx, no radeon 6xxx).

i strongly suspects that Mantle will end up being an extension of DX11.2, porting over most (if not all) of its high level function set, as well as adding a set of low level functions that can interface closer to the GPU itself. think of it as another feature level above DX11.2.

also going by the slides, it appears that there will be a Mantle driver that interacts with the GPU, while the Mantle API interacts with the Mantle driver (isn't this how DirectX works as well?). other manufacturers need only to provide the appropriate Mantle driver, and their GPUs will work with Mantle as well. this at least works conceptually.

clone wrote:

sschaem wrote:

Benefit, you bypass DirectX11 or OpenGL entirely on the PC, removing a lot of latency, bloat, inefficiency and limitations.

JohnC wrote:

Of course

you guys might want to look outside the PC space for why Mantle may take off and take off running.

it looks like Mantle is the low level API for the new generation consoles (most notably Xbox one), if this is the case which it does indeed appear to be then if Mantle gets used in the new generation consoles it'll make for a much easier cross platform port.

Hell, i'd point to the PS4 more than the xbone. if they make a game on xbone using purely directx, they can port it to PC pretty easily. No need to use Mantle. On the PS4... there is no directx. Mantle may be the primary API for all PS4 developers.

In any case, the biggest difference between Mantle and Glide is that 3dfx didn't have their hardware in 80% of the console market.

willyolio wrote:

In any case, the biggest difference between Mantle and Glide is that 3dfx didn't have their hardware in 80% of the console market.

Interestingly enough, the Gamecube/Wii/WiiU are all using ATI tech as well. Not GCN though so doesn't benefit from Mantle.

You could pretty much say AMD owns the entire console market.

If AMD is smart, and there should be a decent chance they already thought of this, they will leave Mantle open, or at least have cheap enough licensing to be worth it for NVidia. If it actually has real world implications on performance, NVidia might actually have to jump on the bandwagon to stay competitive in some spaces, if only to support developers with it, especially if it's true it will be the main api for consoles.

Look at what NVidia did with PhysX and also it's "TWIMTBP" moniker. They have some games that support it, they also played foul with it, making it somewhat of an half-assed measure since not everybody can use it. So most developers really don't have much reason to implement PhysX unless they have a good reason for using it, which means, features that make good use of it, getting NVidia support on development, or see a very target market where people use NVidia. And they will still have to make a code-path that don't use it if any other people should be able to play the game. Thus far I still don't see any real adoption for PhysX except for... gee we have some good eye candy if you have it, but if not, it doesn't impact your gameplay. As for TWIMTBP has been used as leverage to get devs extra support in making sure they run nice on their cards, but not really offer any benefit for anybody else, and at other times, being directly detrimental if they use functions they know wont work as well on AMD cards.

That said, I'm definitely going wait for more info before saying either good or bad about it, but getting more performance out of the hardware is a good thing as long as it doesn't impact stability. With the new consoles out we might actually see a new wave of evolution in game graphics and physics, something that has been surely lacking during the last console cycle except for a few games here and there.

Also looking at the presentation posted in the other thread, AMD and DICE/EA has been in bed together for a good while and I've been really happy with my 7970 all through a thousand hours of BF3, my bet is that with BF4, The new Star Wars and Mirrors Edge on Frostbite probably using mantle, I for one will definitely get on the next gen cards as soon as I can. A single 7970 is still not enough to run everything on high with 60fps minimum on my 30" monitor in native 2560x1600.

God no, here comes the return of the era where I have to use a launcher program and select which renderer I would like to use with my game. The mentions of "Glide" are spot on.

In an age where user-friendliness is improved upon and stressed year after year, the above scenario will simply not do. Mantle will either increase development time and costs by giving developers "one more thing" they need to check against, or generate unsightly PC requirements with future games (such as jewel cases stating you "must have" a this-or-that AMD videocard to play the title, or similar).

The oft-mentioned big-house game engines are pretty much the only ones who can take advantage of this, because they have the manpower and fiscal resources to make sure their cross-platform general game engines work with that "one more thing".

What's with the pessimism, I would think that era is long gone since basically everything has to have a 3D capable card now days. My guess is that it will be either out of these two things... they wont let you choose at all, if you have it, it will use it, or they might just hide it into the graphics menu and make it an optional choice just like the high res texture packs that Crysis did. or like the current games that allows you to specify which directx version you wan't to use.

As for the big dev houses, I'm not so sure of that. If Mantle is the default API for consoles, pretty much any developers will have to support it for multiplatform, which would make DX that "extra bit" that wont be supported as well.

.

clone wrote:

games on Xbox won't use Direct X because it's too inefficient

It's hilarious because it's true.

The chance of Mantle adoption forcing anyone to use specific hardware is approximately zero, though. Aside from the fact that an AMD monopoly on graphics hardware simply will not occur for a multitude of reasons, developers have not forgotten the DX10 debacle and likely never will.

.

NovusBogus wrote:

The chance of Mantle adoption forcing anyone to use specific hardware is approximately zero, though. Aside from the fact that an AMD monopoly on graphics hardware simply will not occur for a multitude of reasons, developers have not forgotten the DX10 debacle and likely never will.

It's not forcing anyone to do anything. It's an advertising point. It's simply extremely likely that Mantle (in some form or other) will be used by both PS4 and XBOne, and PC ports will likely have whatever code left in there. AMD gets an advertising point and can say their cards perform better than nVidia.

clone wrote:

as for the DX 10 debacle are you talking about the way MS ended legacy support in DX when they moved to it, cutting off WinXP in the process?

That and the push to get developers to stop using DX9 entirely to force Vista upgrades. Several dev shops including one of Redmond's own labels went down in flames because they drank the kool aid. Even now, with XP singing its swan song, only some of the more eye-candy AAAs are lacking DX9 support.

.

BTW - with regards to Frostbite 3 - It's very BF4 centric demo reel, but it does have some shots of other things in it. I wish they had a pure tech demo out that would only show off what it can do, without being game centric. Although I guess we will see a bit more with Mirror's Edge 2 and the new Star Wars too.
http://www.youtube.com/watch?v=R9yVV6g3q7g

I also looked at a few different ´comparison clips, like this one. Although that is only Frostbite v2
http://www.youtube.com/watch?v=Xka9c6iG6g0

CryEngine seems to have a visual edge, but I wonder how much of that is attributable to the engine as compared to the art direction and the people doing the resources. In the same way I definitely feel that Frostbite has a better impact on gameplay, especially in BF3 with destruction, etc. While I like Crysis 3 MP, at times, I just get so annoyed that a rocket launcher just doesn't make a dent in something in stuff like crates, etc. But again, how much of that is to the engine and how much is based in the game development part of it. Especially considering how much destruction Crysis 1 had...

There is one more thing I would like to add. Despite all the criticism of Microsoft and how inefficient DirectX can be. DX11 is bullet proof in terms of stability. Users only run into problems with drivers and games. When was the last time anyone got a DX error when their game crashed? Mantle can only match this level of stability, we have yet to see if AMD can deliver.

.

michael_d wrote:

There is one more thing I would like to add. Despite all the criticism of Microsoft and how inefficient DirectX can be. DX11 is bullet proof in terms of stability. Users only run into problems with drivers and games. When was the last time anyone got a DX error when their game crashed? Mantle can only match this level of stability, we have yet to see if AMD can deliver.

You realize your post makes little sense, right?

If AMD is making the driver that crashes or the API that crashes, what's the difference to the end user?

BobbinThreadbare wrote:

michael_d wrote:

There is one more thing I would like to add. Despite all the criticism of Microsoft and how inefficient DirectX can be. DX11 is bullet proof in terms of stability. Users only run into problems with drivers and games. When was the last time anyone got a DX error when their game crashed? Mantle can only match this level of stability, we have yet to see if AMD can deliver.

You realize your post makes little sense, right?

If AMD is making the driver that crashes or the API that crashes, what's the difference to the end user?

Go back and read what you wrote. The one who is not making sense is you!!! End-user needs to know original source of the problem so that they could begin to troubleshoot it. This logic applies to just about to anything in life. If your TV has no signal due to cable service outage you are not going to call TV manufacturer and ask them to fix it are you?

clone wrote:

Mantle can only match this level of stability, we have yet to see if AMD can deliver.

AMD already has in the past when they provided the graphics chips and low level API for both Xbox 360 and all of the Nintendo consoles.

one could argue that it's DX that is the problem that causes AMD and Nvidia to constantly upgrade their drivers.

Another nonsensical statement.

Consoles have standard hardware in all units, it is a lot easier to write and test software with one set of specs.

Nivida and AMD update their drivers to fix their own bugs, optimize games and launch new products.

.

We ask AMD: Why will Mantle be different?

By VR-Zone Sam Reynolds. Good read.

AMD’s Ritche Corpu

"If you think about it, Mantle is truly a collaboration and I’ll tell you [DICE’s Johan Andersson] was at the forefront of that at the very beginning. A lot of the feedback on the development of Mantle came from him. We also solicited feedback from a lot of other partners that we haven’t announced yet. At this stage, Battlefield 4 and FrostBite 3 are the closest to deliver something today. I think, as I mentioned before, the goal would be to provide the spec and SDK publicly."

WiseInvestor wrote:

We ask AMD: Why will Mantle be different?

By VR-Zone Sam Reynolds. Good read.

AMD’s Ritche Corpu

"If you think about it, Mantle is truly a collaboration and I’ll tell you [DICE’s Johan Andersson] was at the forefront of that at the very beginning. A lot of the feedback on the development of Mantle came from him. We also solicited feedback from a lot of other partners that we haven’t announced yet. At this stage, Battlefield 4 and FrostBite 3 are the closest to deliver something today. I think, as I mentioned before, the goal would be to provide the spec and SDK publicly."

Are you another PR mouthpiece? We seriously don't need any more of those.

Don't like the story don't read it than.

No one is forcing you to be more knowledgeable for what will happen. And soon.

If Mantle gains enough traction, this and Gallium 3D will be enough to make me buy a Radeon next time.

michael_d wrote:

BobbinThreadbare wrote:

michael_d wrote:

There is one more thing I would like to add. Despite all the criticism of Microsoft and how inefficient DirectX can be. DX11 is bullet proof in terms of stability. Users only run into problems with drivers and games. When was the last time anyone got a DX error when their game crashed? Mantle can only match this level of stability, we have yet to see if AMD can deliver.

You realize your post makes little sense, right?

If AMD is making the driver that crashes or the API that crashes, what's the difference to the end user?

Go back and read what you wrote. The one who is not making sense is you!!! End-user needs to know original source of the problem so that they could begin to troubleshoot it. This logic applies to just about to anything in life. If your TV has no signal due to cable service outage you are not going to call TV manufacturer and ask them to fix it are you?

In both cases AMD would be the original source of the problem. AMD would need to figure out where the problem is to fix it, but from the end user pov, it wouldn't make a difference if it's their API or driver.

BobbinThreadbare wrote:

In both cases AMD would be the original source of the problem. AMD would need to figure out where the problem is to fix it, but from the end user pov, it wouldn't make a difference if it's their API or driver.

No, the D3D runtime is supplied by MS. IHVs only provide the user-mode and kernel-mode drivers that the D3D runtime eventually talks to. It's been years since the D3D runtime has had critical issues, partly because each version of D3D has successively thinned out the API to reduce "API overhead." The bulk of the modern D3D logic is implemented inside IHV user-mode drivers. Now, writing compliant WDDM drivers is a separate beast altogether, and the complexity is ratcheted up by the need to support the massive D3D9 API (especially the entirety of the old fixed-function pipeline logic).

Zoomastigophora wrote:

No, the D3D runtime is supplied by MS.

I'm fairly sure he was referring to Mantle there...in which case yes, all blame lies on AMD when things go wrong.