AMD Mantle update...

[MaddMuppet]

Johan mentioned there are no number comparisons yet, not even in APU13. How sad. The only demo is by the Oxide engine developers, which is a bit pointless without any comparisons. Hope they post the demo on youtube, regardless.

One would think a demo isn't far away. After all, the BF4 mantle patch is due before the end of this year.

[blindbox]

@MaddMuppet: It shows my impatience.

Bit more details on Mantle below. Are there any slide dumps somewhere, or anyone who at least took a photo of every slide? Right click open in new tab to enlarge if it's too small.

[kromosto]

Open to everyone and with linux version released and steamos..m we are looking to the future. Very very well done.

Sent from my SM-N9000Q using Tapatalk

[Darakian]

That paves the way for battlefield linux I am excite!

[kuroikenshi]

^^^ I thought Magadascar the movie already had battle penguins...

[blindbox]

http://www.hardware.fr/news/13450/ap...ec-mantle.html

And here I thought nobody is willing to post the slides. Yes!

Seems like Oxide mentioned strategy games. I'd love to see > 1000 unit limit without any lag.

Oxide is praising Mantle to cloud 9 it seems. Quite hoping to see a game use their engine soon, but that's unlikely.

EDIT: I'm wrong. Their demo, StarSwarm is coming Q1 2014, seems to feature 2 AIs, over 20k units ('100k batch', just interpolating them saying 10k units ~ 50k batch) fighting each other out.

EDIT2: 3 times faster, but I got no clue what they're comparing it to. I think Mantle is going to be a huge boon to strategy games. AMD FX 8350 downclocked to 2 GHz is still not CPU-bound apparently.

[Final8ty]

https://twitter.com/ryanshrout

[blindbox]

http://www.pcper.com/news/Graphics-C...MD-APU13-Event

[stuffme]

https://twitter.com/ryanshrout

Couple of things about those tweets.

1. https://twitter.com/repi/status/4008...485696/photo/1 If oxide can now have 100k draw calls with mantel, how many draw call they are getting with dx11.1?
2. "Oxide wants to see 300k batches by 2015, 1 million by 2018." What does this mean?
3. I guess AMD hit the jackpot with mantel, now being able to compete intel with it's inferior CPU. Has this been AMD's plan all along when they desided to cut down those FPU units? Hard to believe, but then again, it's also hard to believe that AMD would magically get this lucky when thinking about gaming and their slow CPU's. Too bad this mantle wont help AMD CPU's in real world applications.
4. Just speculations for the future. Shouldn't excavator embedded even more CPU and GPU making FPU's in CPU's even more legacy compute units?

[spursindonesia]

So, Mantle API ---> extensions ---> GPU mArch, it's not locked to GCN only, not even AMD only. And it can be used continuously as GPU mArch evolved, as long the extensions are developed. Correct ? If that so, BRAVO AMD !

[cezar]

Couple of things about those tweets.

1. https://twitter.com/repi/status/4008...485696/photo/1 If oxide can now have 100k draw calls with mantel, how many draw call they are getting with dx11.1?
2. "Oxide wants to see 300k batches by 2015, 1 million by 2018." What does this mean?

From: http://www.hardware.fr/news/13450/ap...ec-mantle.html

1. 15k
2. What is this 'batch' you can see on the 1st photo there:
http://www.hardware.fr/medias/photos...G0043524_1.jpg

[Craftyman.]

They communicated a long time ago the days of competing with Intel on the high-end desktop segment were over. But of course, being able to produce successful APUs requires work to be done on the CPU side of things. If you cease to invest in R&D to continue pushing the CPU architecture, your APU will become an uninteresting product too. This is particularly a concern now that Intel is picking up on the GPU development (Iris Pro). So I am sure there is still focus on the CPU architecture part of the APU. They will just not focus on building "the fastest CPU".

By the way, with AMD reducing focus on claiming the title of the fastest CPU, Intel is pretty much the only company left to preach high-end desktop. That's not good for us.

but from what I understand with mantle the CPU's performance won't be so much of an issue. It seems they want to take a ton of the load off of the CPU? Intel has had free reign with the top performance in desktop sector for years now. I'm not saying it's good but I don't think it will get much worse in terms of year to year performance gains. just my opinion.

[behrouz]

Will Mantle remove Directx from game? how about OpenGL ?

[vario]

Will Mantle remove Directx from game? how about OpenGL ?

Mantle wont remove anything, it will ADD a third API option.

From what im seeing, mantle makes much better use of large number of cores, thats why PS4 and ONE can get away with weak cores, just alot of them.Thats good also for highend intel and me

[Tuvok-LuR-]

"With Mantle, an @AMDFX 8350 underclocked to 2GHZ, the game is still GPU bound on a R9 290X - no loss in frame rate"

[vario]

"With Mantle, an @AMDFX 8350 underclocked to 2GHZ, the game is still GPU bound on a R9 290X - no loss in frame rate"

Another AMD tweet:

RT @cavemanjim: "CPU benchmarks indicate an @AMDFX 8350 is equal to i7 4770K" #APU13 #Mantle #Oxide

Which makes the AMD decision to not do 8 core SR`s even more absurd :-/ .

[cezar]

These tweets are from the Oxide demo (look at the hardware.fr article mentioned above):

http://www.hardware.fr/medias/photos...G0043521_1.jpg

[GeorgeStorm]

If this gets a decent amount of support and is adopted well then this could be great for consumers
The cpu market would become very competitive (for gaming).

Going to keep an eye on this

[kromosto]

Mantle wont remove anything, it will ADD a third API option.

short term you are right but if it succeeds in long term i believe nor directx neither opengl can survive.

[vario]

These tweets are from the Oxide demo (look at the hardware.fr article mentioned above):

...

Thats what i mean, we must have to remember that while chip is larger and hotter, PD chips ARE comparable with 4 core i7`s when workload is properly MT.
All this makes it probable that system with AMD 8 core and R290X is gonna be faster at mantle games than 4 core i5/i7 with comparable nvidia card`s.

Interesting times ahead.God damn, AMD make at least 6 core Sr on Fm2!

[spursindonesia]

short term you are right but if it succeeds in long term i believe nor directx neither opengl can survive.

IF (with caps) Mantle would be successful, then it will actually be a widely de facto industry standard, won't it ? I don't see any problem with that, knowing that Mantle is deep down GPU agnostic, just like what was suggested in the slide. It will be up to the other IHVs (nVidia, Intel, even mobile graphic IHVs like PowerVR, ARM, etc) to built their own extensions and adapt/optimise their GPU designs to get the best performance possible with Mantle.

[radier]

I guess bright days will come for my brothers MSI GX60 witch decent 7970M bootlenecked by Trinity APU.

MORE !!!!

[kromosto]

IF (with caps) Mantle would be successful, then it will actually be a widely de facto industry standard, won't it ? I don't see any problem with that, knowing that Mantle is deep down GPU agnostic, just like what was suggested in the slide. It will be up to the other IHVs (nVidia, Intel, even mobile graphic IHVs like PowerVR, ARM, etc) to built their own extensions and adapt/optimise their GPU designs to get the best performance possible with Mantle.

Of course it is also not a problem but a profit for me but i am sure Microsoft won't be so happy

[Craftyman.]

A funny thing occured to me just now... Isn't this a bit like 3DFX's Glide API? Or is that different?

[Piledriver]

3. I guess AMD hit the jackpot with mantel, now being able to compete intel with it's inferior CPU. Has this been AMD's plan all along when they desided to cut down those FPU units? Hard to believe, but then again, it's also hard to believe that AMD would magically get this lucky when thinking about gaming and their slow CPU's. Too bad this mantle wont help AMD CPU's in real world applications.

So AMD made a CPU geared towards multi-threading, and then worked on something that makes games make better use of it's CPU multi-thread capabilities? Where's the jackpot? Mantle won't help in real world applications? It doesn't need to, AMD cpu are already competitive price\performance with Intel in multi-thread apps, there's multiple reviews benchmarks proving that. Plus there's the GPU to help in real world apps, open-CL Handbrake saw a performance jump above 100% in a trinity APU. Kaveri with proper software support will be a beast, and AMD is working a lot with devs in these days.