Mantle pushes FX-8350 to beyond i7-4960X performance.

Shoot, this stupid war here is far worse than any console fan war could ever be. No wonder vendors think it is not worth it to put effort into PC games anymore.

Vendors would have to be incredibly stupid to think arguing about the validity of an API is somehow related to game sales. Please link the vendors that make this non-existant correlation as I'm sure such a ridiculous, baseless statement wasn't simply made up by you.
 
But they run OpenGL, so we don't need either Mantle or DX12. :cool:

I'm suspicious that we'll ever see something like Mantle on mobile and Linux. They could always surprise us all but so far the official position is that they're investigating the costs and they've already said they don't have any money to spend on Linux.

But nobody uses OpenGL 4.4 unless forced to in Windows. Well anything relevant. :)
 
Vendors would have to be incredibly stupid to think arguing about the validity of an API is somehow related to game sales. Please link the vendors that make this non-existant correlation as I'm sure such a ridiculous, baseless statement wasn't simply made up by you.

Nah, this is just one comment that I made based on the constant bickering and cheapness most PC gamers seem to have. They expect the world but are only willing to pay for next to nothing. Although, I am sure you knew what I basically meant so there. :p:D

Every time something new comes out it seems, folks just like to complain more than research into it. I personally find Mantle to be an exciting prospect and very real option as time goes on. It will also be interesting to see what Intel could do with it once it becomes available to them.

Gaming on a Surface Pro 3 would be interesting at that point or any other Intel based Windows tablet for that matter.
 
I don't like Mantle for the same reasons I don't like D3D and Metal. I think what's best for the industry is a cross-platform, open-standard, royalty-free, extendable graphics library guided by an open committee structure.

That and I think with Mantle and Metal there was a bit too much stretching of the truth re: how "heavy" APIs like D3D and OGL are and how "low-level", "to the Metal" their APIs are.

Nothing you said above proves that either Mantle or DirectX are tecnically inferior to your open crossplatform, royalty free api guided by a committee. I'd say that things guided by a committee actually end up more often than not in compromises that makes the end product less than optimal.
 
How could anybody take this seriously? We don't even have the faintest, singlest, smallest idea yet of how something like Mantle would look on, e.g., an Intel IGP, and already you're accepting the bullet point on a marketing slide about how a Mantle port is just one hop away from a console version? How could that be unless the underlying PC hardware is almost the same as what's in the console?

Yet here in this thread you have no problems claiming Mantle is inferior.
 
Nah, this is just one comment that I made based on the constant bickering and cheapness most PC gamers seem to have. They expect the world but are only willing to pay for next to nothing. Although, I am sure you knew what I basically meant so there. :p:D

Every time something new comes out it seems, folks just like to complain more than research into it. I personally find Mantle to be an exciting prospect and very real option as time goes on. It will also be interesting to see what Intel could do with it once it becomes available to them.

Gaming on a Surface Pro 3 would be interesting at that point or any other Intel based Windows tablet for that matter.

You mean the same group of people who spend as much or more for a GPU than an entire console costs? I'm not sure you need to keep provimg to everyone you don't know what you're talking about. Your comments are either totally baseless not to mention irrelevant, or the complete opposite of reality.
 
Nothing you said above proves that either Mantle or DirectX are tecnically inferior to your open crossplatform, royalty free api guided by a committee. I'd say that things guided by a committee actually end up more often than not in compromises that makes the end product less than optimal.

Well we know that Mantle is inferior because as it stands now it only runs on one architecture and it seems to be in perpetual closed beta. When it's open and adopted by all the GPU players, then we can consider it playing in the same ballpark as D3D and OGL.

And then we'll have to judge its performance once it's suitable for cross-vendor development.

But what we do know about its performance is that it's not markedly better than grandma-old D3D11 in GPU-limited scenarios. It is helpful in CPU-limited circumstances but those aren't necessarily ubiquitous and it's not obvious if it's better than D3D because it's that good, or if it's just because D3D11 is older than an iPhone 3GS.
 
Well we know that Mantle is inferior because as it stands now it only runs on one architecture and it seems to be in perpetual closed beta. When it's open and adopted by all the GPU players, then we can consider it playing in the same ballpark as D3D and OGL.

And then we'll have to judge its performance once it's suitable for cross-vendor development.

But what we do know about its performance is that it's not markedly better than grandma-old D3D11 in GPU-limited scenarios. It is helpful in CPU-limited circumstances but those aren't necessarily ubiquitous and it's not obvious if it's better than D3D because it's that good, or if it's just because D3D11 is older than an iPhone 3GS.

So you think it's inferior based mainly on subjective characteristics, not objective ones.
 
I'd say that things guided by a committee actually end up more often than not in compromises that makes the end product less than optimal.
Depends on the committee. Evolution by committee can be slow and painstaking, but can also result in high levels of stability and reliability, high levels of specification quality and diminished feature creep. Evolution by a sole controller can often yield fast iteration but low stability and low quality.

I think OpenGL has been mismanaged in some respects, but I do think they're on a reasonable evolutionary path, especially within the ES subset. There's just too much reliance on the community to develop tools, really.
 
ogl ES is pretty irelvent in this context, as it is targeted the mobile platforms.
 
To whom do you think I was referring when I said "they're on a reasonable evolutionary path"?

Obviously, it's in reference to the Khronos Group. They manage both OpenGL and OpenGL ES. Is this confusing to you?
 
again, mobile,

no relevance whatsoever to desktop.

Khronos has dropped the ball on desktop.

OGL is dead man walking.

No one wants to use it, and those that are forced to, do it reluctantly.
 
NVIDIA was able to beat Mantle with a simple driver update. No new API needed.


http://www.tomshardware.com/news/nvidia-driver-update-direct3d-optimization,26381.html

/thread

nvidia-d3d,D-J-428167-22.png

I wonder where they get these performance figures from? Seem really really poor overall.

thief_1920_1080.gif

Here's an independent review site. Even at 1080 (the res nVidia used because their arch is better at low resolution) the 780 ti doesn't beat the 290X. Although even last gen cards perform better than nVidia's results.

The fact is if mantel is a true low level api, it's vendor locked it's simple as that. Nvidia and Intel can write a wrapper to interpret mantle code for their gpu's but that wrapper will create cpu overhead making mantel pointless for them. Let alone that intel's gpu's architecturally are very far from nvidia's or amd's plus the simple fact that amd doesn't have support for mantel for the HD 5000 or HD 6000 cards. Meaning support across brands is a pipe dream at the moment.

If it's vendor locked that means small userbase, companies don't want to pay people to implement mantel let alone run long term support for fixing bugs that are specific to mantel, the user base simply would never justify the cost.

So mantel will exist as long as AMD props it up and will fall apart as soon as AMD stops absorbing the implementation cost.

AMD hasn't reinvented the wheel, all they have done is remarketed something the industry abandoned for the PC because it added too much extra cost to development. Low level api is great when you know the hardware, when you specialize, you will never get that for a generic PC game platform.

Sorry, but your understanding of Mantle is way off. It requires a driver. If the hardware supports the features all they need to do is write a driver. It's not direct to metal.

In March it was announced that Crytek was adding Mantle support for their engine. Since then we've come to know that the company is basically in the worst of financial troubles. Why would a company that was so strapped for cash take upon themselves the added financial burden of adding an API like Mantle to their engine? Conversely, how amendable would a company in Crytek's financial position be to accept funding from AMD to add Mantle?

I'm bringing this up because at the time the companies were talking about how CryEngine was picking up Mantle because of all of the amazing superior technical wonders of Mantle itself. That was the public facade put out there.

Something to keep in mind when you hear about how "developers" are so eager to implement this in their games.

So, you are theorizing that Crytek took money from AMD for Mantle because they were cash strapped? Well, they could sure use this imaginary money of yours to become real so they can keep going. We've seen no evidence of what you are claiming here. Especially not from you who are making this claim. Do you have any?

That ought to be a very interesting experience. So we're going to take an API that's made for low-level, bare-metal access to GCN architecture and nvidia / intel will write their drivers to convert those API calls to match up as best they can with their own architecture?

And this is preferred to OpenGL how...?

You're basically advocating something that would be twice (if not more) inefficient than an abstract API like D3D for non-AMD hardware.

Come to think of it, no wonder AMD thinks it's such a great idea to be vendor-agnostic.



Do you really believe that I am set to benefit financially by criticizing Mantle?

Mantle is not bare metal.

The way Mantle treats a CPU seems to be implied by AMD to suggest that a variation of it will make it's way into DX12. Which would immediately set off the "Then why do we need Mantle?" crowd. There are other platforms like mobile, Linux, hell maybe even servers that could possible benefit in some form or fashion and they DO NOT RUN DX12!

But since AMD is trying to design the API to be easily portable, it makes complete sense that they would want an universal tool to get great performance for their products. Remember the generic path will always be there for other vendors like Nvidia and Intel if they aren't interested in Mantle. So no harm; no foul.

Even if DX12 matched Mantle performance (I'm not saying it will. This is just for the sake of discussion.), besides being tied to Windows as you are saying, DX also holds back the hardware industry. We've had DX11 since 2009 and because of that rendering feature development has virtually ground to a halt over that time. I know there have been some new features added, but not enough to matter. Consider that nVidia is still on feature set 11_0 and it doesn't really matter.

Getting out from under Msft, and hopefully beyond Windows, is really what I'm hoping for with Mantle. I've got my fingers and toes crossed that when AMD releases Mantle that Intel will decide to adopt it. With only AMD supporting it adoption has been faster so far than what DX11 was. Add Intel support and I believe DX will be done.
 
Even at 1080 (the res nVidia used because their arch is better at low resolution) the 780 ti doesn't beat the 290X.
That "better at low resolution" jab doesn't really make much sense in this context.

If Nvidia wanted to show-off against Mantle, they'd choose a HIGH resolution to force a GPU-limited scenario (where Mantle and DX11 perform similarly).

Running at 1080p actually gives Mantle the advantage, as the test will be CPU-limited more-often with such low GPU load.
 
That "better at low resolution" jab doesn't really make much sense in this context.

If Nvidia wanted to show-off against Mantle, they'd choose a HIGH resolution to force a GPU-limited scenario (where Mantle and DX11 perform similarly).

Running at 1080p actually gives Mantle the advantage, as the test will be CPU-limited more-often with such low GPU load.

My point is the results look wrong. Compare the TPU graph, which isn't even using Mantle. The nVidia results are bogus.
 
But what we do know about its performance is that it's not markedly better than grandma-old D3D11 in GPU-limited scenarios.
Well it's still new and AMD has yet to announce proper GPU optimizations within Mantle. So far, it has only been said to focus mainly on CPU optimization and/or CPU <---> GPU communication. As you said, Mantle is still beta. Therefore, it is not a final product. It is very possible the non-beta versions of Mantle will offere GPU optimizations/performance beyond DX11. But, we will see!


It is helpful in CPU-limited circumstances but those aren't necessarily ubiquitous and

So I'm not sure what you mean here. A game is CPU limited, or it isn't. Sure, we can't know a game will be CPU limited until it is tested. But you either do or do not have CPU limitation. Whether it is momentary or sustained, it nets the same general outcome: The CPU cannot always feed the GPU fast enough. So far, Mantle has been a boon for fixing CPU limited setups/situations. This means higher minimum framerates, higher average framerates, and lower and more consistant frametimes. All of that nets some notable benefit to gameplay experience and 'feel'. [H]'s own reviews even note that Mantle generally feels better, even if the upper end performance may not be quite as good, in some comparisons.

On my own system, I get over 25fps higher average framerates in BF4. This suggest that my CPU limitations are cleared up so well, that my 7870 is free to perform as if had a newer/better CPU. (I have an AMD Phenom II X6) and my gameplay feels smoother and more responsive than ever.

That stuff should not be dismissed or glossed over.

it's not obvious if it's better than D3D because it's that good, or if it's just because D3D11 is older than an iPhone 3GS.

Does the distinction matter? The point of Mantle is recognizing that DX11 is congested and does it wrong, so to speak. Yeah, DX11 is old! Mantle is 'new', with more modern ideas and a finer approach. That's a positive. There's no way to spin that otherwise. Even your words highlight the most basic premise for Mantle: things could be done better.
 
My point is the results look wrong.
Wasn't commenting on the results, only the jab (which was a bit backwards).

Compare the TPU graph, which isn't even using Mantle. The nVidia results are bogus.
Depends on what those graphs are showing. They aren't labeled, so we don't know if they're displaying minimum or average framerates.

If the Nvidia graph is minimums and the TPU graph is averages, then the numbers make a lot more sense.
 
Last edited:
Your taking about a game that EA has given away for free that no one gives a shit about and really doesn't look that good. Sentiment is all AMD has to work with in this case.

Still runs on FB3 still works well I guess. Is that something that does not count when AMD wins ?

Just tried the link again from a computer, was using my ipad before... You were better off not posting that, because here's what I saw which I'm not sure how you missed.

Question was asked, first guy answers "lets leave that one alone" the other guy says "they wouldn't have to pay us anything" One guy completely wants to avoid the question, the other not saying that they weren't getting paid...

Denial on a level I wasn't expecting, that's what you're living in. Your own "source" and it's far clearer that they were paid than they weren't. If they weren't paid, the first response would not have wanted to avoid the question all together and the second guy would have actually said "they didn't pay us anything"

Thanks for playing, I know that's the best you have because if you had anything better, you'd have posted it.


You are lying here again , the conference moderator says lets leave that one alone , the guy who is hosting the Nvidia forum. Don't define him as guy , , clearly he doesn't like what is being said.

And the person you define as second guy is someone whom matters since he "helped" on Frostbite 3.

You have nothing but conjecture and you are claiming what again?

Don't come posting commentary on a story by fudzilla and then announce it as fact since the story was nothing but conjecture and the comment on it by hexus.net didn't do anything beside speculate somewhat more.
http://www.fudzilla.com/home/item/32673-amd-spent-millions-on-battlefield-4-deal

stop repeating stuff that is not the truth.
 
Last edited:
Still runs on FB3 still works well I guess. Is that something that does not count when AMD wins ?

not really a win when you "win" is at best 10fps again other then that ONE TIME every thing else is in side margin of error 3-5 fps is not a win
 
Still runs on FB3 still works well I guess. Is that something that does not count when AMD wins ?




You are lying here again , the conference moderator says lets leave that one alone , the guy who is hosting the Nvidia forum. Don't define him as guy , , clearly he doesn't like what is being said.

And the person you define as second guy is someone whom matters since he "helped" on Frostbite 3.

You have nothing but conjecture and you are claiming what again?

Don't come posting commentary on a story by fudzilla and then announce it as fact since the story was nothing but conjecture and the comment on it by hexus.net didn't do anything beside speculate somewhat more.
http://www.fudzilla.com/home/item/32673-amd-spent-millions-on-battlefield-4-deal

stop repeating stuff that is not the truth.

You clearly don't know what a lie is, since I told you exactly what happened in the video YOU posted. You've shown nothing that says DICE wasn't paid. I've showed you something that says that they were. That's where we stand. Conjecture? A lot of it, from your end. Actually, it's more like denial.

Here's another source that isn't from fudzilla that says the same thing.

http://bf4central.com/2013/10/amdamd-paid-ea-5-million-battlefield-4-deal/

I hope if you're going to post another source, it actually says what you think it says. Thus far you've shown nothing but hurt feelings
 
he says in the video, "they wouldn't have to pay use anything, i've been pitching these ideas for many many years."

Unless you've got the contract that divides what percentage is game bundles, bf4 promotion exclusivity or hypothetical "compulsory" mantle inclusion, this debate is pointless. It's ludicrous to believe for one second that amd would be paying per game basis to implement mantle, they just don't have that amount of resource and we all know it .
 
"wouldn't have to" vs "didn't pay us" not the same thing. Why would he say "wouldn't have to" if they actually "didn't"?

Answer = He wouldn't.
 
"wouldn't have to" vs "didn't pay us" not the same thing. Why would he say "wouldn't have to" if they actually "didn't"?

Answer = He wouldn't.

What about when mentioning the EA&AMD deal, tell about the copies of BF4 included in that deal, which later was bundled with the "Never Settle" promotions?
 
I've played PvZ:GW exclusively through Mantle and it is perfectly fluent, all the way. Never ever any frame pacing or latency issues. BF4 on the other hand, I haven't had a good Mantle experience with that game, with massive spikes in frametimes, sometimes freezing the game for a second, while the D3D11 version had none of these problems.

In the end it's the same engine, both use Frostbite 3.
 
again, mobile, no relevance whatsoever to desktop. Khronos has dropped the ball on desktop.
OpenGL ES is on the desktop, and has been for some time. NVIDIA supports it in their drivers, as does AMD, and OS X Mavericks supports the ES subset out of the box.

That's rather beside the point, as I was again referring to what the Khronos Group has done with both OpenGL and ES, but it's worth mentioning so you needn't permanently wallow in your own ignorance.
 
Wasn't commenting on the results, only the jab (which was a bit backwards).

Funny, how you see a jab simply when facts are stated. I was actually trying to avoid that part of your comment because it's irrelevant and argumentative. You do realize that relative to each other AMD's performance improves with resolution?


Depends on what those graphs are showing. They aren't labeled, so we don't know if they're displaying minimum or average framerates.

If the Nvidia graph is minimums and the TPU graph is averages, then the numbers make a lot more sense.

So what good are they if we don't know what they were measuring?
 
You clearly don't know what a lie is, since I told you exactly what happened in the video YOU posted. You've shown nothing that says DICE wasn't paid. I've showed you something that says that they were. That's where we stand. Conjecture? A lot of it, from your end. Actually, it's more like denial.

Here's another source that isn't from fudzilla that says the same thing.

http://bf4central.com/2013/10/amdamd-paid-ea-5-million-battlefield-4-deal/

I hope if you're going to post another source, it actually says what you think it says. Thus far you've shown nothing but hurt feelings

From your link:

The Title: "AMD paid up to $8 million for Battlefield 4 deal"

The Tag line: "AMD paid EA up to $8 million USD in order to get exclusive use of BF4 to promote its upcoming products."

From the actual article: "According to our sources, AMD paid EA up to $8 million for the rights to use Battlefield 4 to promote its hardware products. Other publications reported a more conservative $5 million number. The deal also gives AMD the right to bundle Battlefield 4 with various graphics card products. The deal is exclusive to AMD, which means you won&#8217;t be seeing Battlefield 4 bundled with Nvidia or Intel hardware anytime soon."

Here's where Dice is mentioned: "Part of the deal included AMD giving Battlefield 4 developer DICE early access to the new &#8220;Mantle&#8221; API technology. The game will support the new tech in an update in December, which is said to bring much better performance to AMD Radeon graphics cards."

I see nowhere that it states Dice got paid by AMD. I put forth that it's you who doesn't know what a lie is. "An attempt to deceive by changing or omitting the facts." The facts, as you've presented them, don't prove what you are saying. You claiming they do won't change that.

Plus, it's a pure rumor article that can't be claimed to prove anything. He said, she said isn't evidence.
 
No idea what you're asking me

My main point would is: the sponsorship agreement is never for just one subject, there's always more.
Here's two examples:
1. AMD paid EA $8mil to implement Mantle in BF4.
2. Nvidia paid Ubisoft $5mil to lock out AMD from the Watch Dogs optimizations.
Both statements have equal quantity of truth and incorrectness.
In both cases the sponsorship includes more than I've stated(game copies which are distributed with hardware and cross promoting brands, etc.).
And BTW: the sponsorship was between EA and AMD, NOT between DICE and AMD; so the video was barking at the wrong tree.
 
My main point would is: the sponsorship agreement is never for just one subject, there's always more.
Here's two examples:
1. AMD paid EA $8mil to implement Mantle in BF4.
2. Nvidia paid Ubisoft $5mil to lock out AMD from the Watch Dogs optimizations.
Both statements have equal quantity of truth and incorrectness.
In both cases the sponsorship includes more than I've stated(game copies which are distributed with hardware and cross promoting brands, etc.).
And BTW: the sponsorship was between EA and AMD, NOT between DICE and AMD; so the video was barking at the wrong tree.

Seems you haven't read my previous posts in this thread. I specifically mentioned Mantle being "part of" the deal and linked two different articles mentioning the same thing. I'm not saying nor implying otherwise.
 
OpenGL ES is on the desktop, and has been for some time. NVIDIA supports it in their drivers, as does AMD, and OS X Mavericks supports the ES subset out of the box.

That's rather beside the point, as I was again referring to what the Khronos Group has done with both OpenGL and ES, but it's worth mentioning so you needn't permanently wallow in your own ignorance.

Name one AAA title on the desktop that uses OGL ES?

like i said, its targeted at mobile, the fact that its supported on desktop is irrelevant.

ID is using OGL in thier IDtech engine, but even their own coders complain about it.

Again, my original statement stands. Khronos has mishandled OGL on the desktop, and its living on borrowed time.
 
Last edited:
Seems you haven't read my previous posts in this thread. I specifically mentioned Mantle being "part of" the deal and linked two different articles mentioning the same thing. I'm not saying nor implying otherwise.

No what you doing here is conjecture. You post things which are not credible and then project them as a fact. What is worrying that BF4 central somewhat shows that it is aware of something but can not quantify what it is. Then to make the story saucy it starts adding things.

If any article has merit then it will show the facts clearly in the headlines. Now up to 8 million and between 5 and 8 million is speculation it is not a fact you could run that story each day for the next 300 years and still not guess the right amount.

When you tack on things from sources as Fudzilla and clearly a commentary on fudzilla story by hexus you know you are not telling the truth. Bf4 central are people who copy news from anywhere I guess there was no source link unlike hexus.

When this gets posted in volume and it is true that are facts , not speculation will drive an article.

You claim I have no facts, While showing the video from the technical director of Frostbite. The person that would know what is going on you hide that, you can not deny this statement in words by defining people as guy. You blame me for having no source while my source can be named and shown speaking about not getting paid by AMD for Mantle.

How on earth do you expect anyone to read up on your hopelessly failed attempts at making conjecture truth while there is a credible source that proves that the story by all the 3 websites (same story different "facts") is false.



.
 
No what you doing here is conjecture. You post things which are not credible and then project them as a fact. What is worrying that BF4 central somewhat shows that it is aware of something but can not quantify what it is. Then to make the story saucy it starts adding things.

If any article has merit then it will show the facts clearly in the headlines. Now up to 8 million and between 5 and 8 million is speculation it is not a fact you could run that story each day for the next 300 years and still not guess the right amount.

When you tack on things from sources as Fudzilla and clearly a commentary on fudzilla story by hexus you know you are not telling the truth. Bf4 central are people who copy news from anywhere I guess there was no source link unlike hexus.

When this gets posted in volume and it is true that are facts , not speculation will drive an article.

You claim I have no facts, While showing the video from the technical director of Frostbite. The person that would know what is going on you hide that, you can not deny this statement in words by defining people as guy. You blame me for having no source while my source can be named and shown speaking about not getting paid by AMD for Mantle.

How on earth do you expect anyone to read up on your hopelessly failed attempts at making conjecture truth while there is a credible source that proves that the story by all the 3 websites (same story different "facts") is false.



.

You can keep repeating yourself, the fact remains that I've provided evidence showing DICE was paid, and you haven't been able to rebut that in anyway except to cry about it over and over again and provided a video that didn't actually say what you think it said. I have two articles claiming the deal happened, and thanks to you, I have a video of DICE NOT denying it. That's where we stand. Cry more, it won't change anything.
 
Last edited:
That was not your assertion.

You asserted Dice was paid SPECIFICALLY to implement mantle.

THAT is pure conjecture.

There is zero proof that they were allocated funds by AMD specifically for the purpose of mantle development.

You were called out for stating conjecture as fact. It is not.

The burden of proof rests with you, as you are the one making a claim. You have yet to cite any credible source of truth, only other conjecture.

The accusation stands.
 
Last edited:
No one on this planet would be so naive to think EA spent their own money to port their game for a small chip vendor.
 
Back
Top