I agree. I don't think current i7 setups will see much more than 10%. I was thinking more along the lines of 6 months from now when the next gen CPUs are only 5-10% faster but the 20nm GPUs are 50%-100% faster. Another 20-30% on top of that would be awesome.
-
50% 100% faster ? Is that kind of an over-optimistic prediction or..
-
-
who would put R290/GTX780 in gaming machine in pair with Pentium?
-
The 10% improvement under GPU-bound scenarios is in line with AnandTech's preliminary BF4 numbers: AnandTech Portal | Battlefield 4 Mantle Performance Preview
AMD's official performance data:
I appreciate how they included numbers from their competitor's chips as well.
More graphs from the German review: Battlefield 4 Multiplayer-Benchmarks - AMDs Mantle-API im Test: Der Prozessor-Katalysator - Golem.de
64 players, Ultra Details (Core i7 3770K, 16 GB DDR3-1333, Radeon R9 [email protected] / 2500 MHz, Windows 8.1 x64, Catalyst 14.1 Beta)
Extreme details (Core i7 3770K, 16 GB DDR3-1333, Radeon R9 [email protected] / 2500 MHz, Windows 8.1 x64, Catalyst 14.1 Beta)
The German review also mentions that frametimes are more consistent in Mantle compared to D3D11, making the game feel even smoother than just the nominal FPS increase suggests.
So far Mantle is looking like mostly CPU draw call optimizations, but as the BF4 numbers show, there is a little bit of GPU optimization as well. Some of the BF4 numbers are quite impressive considering they took an existing game already highly-optimized under D3D11.1 and were able to extract that much extra performance out of it by retrofitting Mantle support. If the numbers from the Star Swarm demo are any indication, I'm excited for future games built for Mantle from the ground up. Mantle is still in beta and this is just the beginning, so expect a lot more developments in the near future as it picks up momentum.Robbo99999 likes this. -
-
Robbo99999 Notebook Prophet
@octiceps: Yep, I think so too, the most exciting part of Mantle is going to be those upcoming games designed for Mantle that can really exploit Mantle's benefits, yet still push those high end CPU's to their limits - that might yield up some very impressive visuals or gaming experiences!
-
-
I posted specifics in the BF4 thread but the mantle driver still has a lot of issues- I am testing it out on my desktop rig with dual R9-290x cards at the moment.
-
Installed the new mobility 14.1 beta driver, but somehow it installs the good old 13.12 driver;
-
Maxwell going to be super efficient and it is coming together with Broadwell in the summer time. I think it is better thing than Mantle, since Mantle working only with few games and it does not bring too much extra performance.
-
Robbo99999 Notebook Prophet
-
[H] review out. that frametime tho.
-
What's happening with the spikes?
-
is mantle a software thing or a hardware thing? will all AMD users benefit from mantle enabled drivers?
-
will all users benefit from it ? : all people with GCN cards (as far as i know the beta is properly working only on gcn 1.1 cards)ronferri likes this. -
I think Mantle has more potential as it is still in its "infant" stage. As AMD continues to tweak it, performance is sure to improve.
-
-
Basically anything that is 28 nm here:
List of AMD graphics processing units - Wikipedia, the free encyclopedia
List of AMD graphics processing units - Wikipedia, the free encyclopedia
List of AMD graphics processing units - Wikipedia, the free encyclopedia
List of AMD graphics processing units - Wikipedia, the free encyclopedia
List of AMD graphics processing units - Wikipedia, the free encyclopedia -
Robbo99999 Notebook Prophet
-
Also as a reply to all those comments that claim that mobile Core i7s won't see much of an improvement, this is just false. Lets not forget that there are even dual core i7s, and these are only as fast as desktop i3s. The quad core i7s on the other hand are clocked lower and aren't even as powerful as desktop Core i5s, and while desktop i5s don't see much of an improvement from Mantle because they only have 4 threads, the i7s have 8 and that's where Mantle comes in. Mantle is much more optimized for multithreading than DirectX will ever be. And while right now Mantle won't show much improvement on 7970m/8970m because it doesn't work well on GCN 1.0 it eventually will and probably work even better on PI cards.
Another change that might eventually end up happening because of Mantle is Notebooks getting better battery life. Remember when AMD underclocked an 8350 to 2GHz and it showed no performance degradation under Mantle. If mantle becomes more widespread there is a possibility of lower clocked higher end chips emerging with much lower TDPs, which will be great for those slim Gaming Notebooks, as they might become even thinner and the battery last a decent amount under non GPU heavy loads.octiceps likes this. -
Mantle pretty much only works with scenarios where there is CPU bottlenecks.
Our mobile CPUs are overpowered for pretty much any GPU you can find out there. Our CPUs are 47W while our GPUs top out at 100W.
Desktop CPUs are typical 80-90W while the GPUs that made Mantle shine was around 270W.
The scenarios where you got CPU bottleneck and where Mantle played the biggest role with R9 290x was:
64 player Multiplayer in BF4
R9 290x Crossfire in BF4
Maybe with 8970M CF in multiplayer you will get some decent boosts with Mantle, the rest will be pretty dissappointing with a mobile i7 because they have no problem keeping up with our mobile GPUs. -
I just dont get the amount of negativity. Its free performance boost that can only bring more good stuff in the long run. It is still in beta - meaning not finished and not for use with general population.
Sent from my HUAWEI Y300-0100 using Tapatalkocticeps likes this. -
awww so HD 6700 series are not GCN?
-
-
If Mantle would work for ALL games... Anad even not talking about already created.
Perhaps Mantle is to become something like Steam for HL2 team. But if Nvidia offers smth similar in 1 year then all will be gone.
For gamers too because instead of 1 general Mantle we gonna get 2 different proprietary features of GPU competitors. -
Just to support my previous argument here is a chart showing Mantle performance advantage with i5 3570K + R7-260X in Battlefield 4.
And here is the Legit Reviews article which not only contains the BF4 test but also different Star Swarm Demo tests.octiceps and Robbo99999 like this. -
-
It's also has to be noted so far the Mantle games that have been tested are also AMD biased games. These games are being heavily paid for by AMD, no doubt.
While I think PS4 and XBone's AMD APU are pathetic disappointing garbage, it does make sense for anyone building another PC Gaming rig to buy AMD GPU. Clearly all games that will be ported from console will inherently be AMD optimized/biased.
I hope Mantle dies. This isn't about whether Mantle's performance gains are real or not. I do not want any API to be used on any games from a GPU manufacturer, period. NONE. I don't want Nvidia to create a competing API and I don't AMD to promote their own. Both manufacturers should be promoting and helping to develop a third party API that will be universally used by any GPU manufacturer. Doesn't matter to me whether OpenGL or DirectX is the dominant API on PC Gaming, but I'd much rather have OpenGL or DirectX than this Mantle nonsense. I truly hope Mantle dies in a fire. And Nvidia steps up and helps contribute to massive gains on either OpenGL or DirectX to put AMD's Mantle in it's place, and what it really is. AMD can claim it's open to Nvidia all they want, but it's clearly an API that is optimized to AMD's GCN 2.0. It doesn't even run on previous GCN 1.0.... Nvidia's CUDA cores are NOT GCN shaders. This is ridiculously stupid to believe that AMD's Mantle is open to Nvidia. Har har, how dumb do you think we are AMD? Shove Mantle where sun don't shine AMD and die in a fire.
This whole Mantle is open to Nvidia is just a business, marketing scheme to shame Nvidia. It's clearly not open to Nvidia since Nvidia don't use GCN 2.0 shaders, but to say it is and taunt them? It's great because it just makes Nvidia look bad. It's AMD being asshats and hopefully people wake up and realize that API from a GPU manufacturer is horrible and shouldn't be supported. Having a plethora of API doesn't make any sense. PC Gamers should be able to buy whatever hardware they want from whatever manufacturer and not have to worry about if their GPU supports whatever nonsense API the developer chose. It should be a third party like OpenGL or DirectX. Instead of creating Mantle, AMD should have contributed to OpenGL to make it better. The fact they didn't just shows the motive behind Mantle.
G-Sync sucks too. Again Nvidia should have worked with future DisplayPort that will be released near end of this year, something that can be used by Intel, AMD and Nvidia and any other GPU manufacturer. Mantle and G-Sync both suck and should di e.
Cloudfire likes this. -
-
Robbo99999 Notebook Prophet
-
columbosoftserve likes this.
-
I voted for Maxwell, because it really seems to set new standards on the performance-per-Watt metrics, reading the new GM107 details. Instead Mantle seems a big marketing bull, it works only with certain games, with certain GPUs, in certain CPU demanding scenarios. And whenever AMD is out of money won't come more mantalized games.
-
columbosoftserve Notebook Evangelist
-
Although Zymphad posted a big rant, he made a lot of valid points.
API made and distributed by a OEM is nothing but a strategy to enrich their catalogue of products and to make them more tempting for the buyer. Mantle will most likely not be open for Nvidia cards unless there are money to be made. Period. AMD have invested far too much money and effort on getting the software up and running on Battlefield 4 (5 months it took), and it does have improvements over DX which is universal and open for EVERYONE. Same goes for OpenGL. It will undoubtly lead to Nvidia making similar improvements on their Nvidia API, or worse, it will like Mantle cost them money to make the developers code with that API. Which worse case could end up as more expensive GPUs.
The good thing about Mantle though is that it will hopefully get Microsoft behind in gear so that they update the DX to give similar draw call improvements on the CPU within that code.
But like Tim Sweeney said earlier, we dont need Mantle. We already have DX and OpenGL. A third API is too much.
I personally think Mantle will become AMD`s PhysX. Used by few, since it takes up man hours, it will not be prioritized because DX which is used by all systems will be the priority, it will cost money, and it will only be useful for GCN cards. Not Nvidia cards, not Intel IGPs, not pre 7000 series. Mantle will become a pro for AMD cards just like PhysX is a pro for Nvidia cards.
And in the end, we know DX will be updated. We know Nvidia will respond. What happens then? There is a limit on how low you can go with an API anyway. They will meet a point with Mantle where software can do what it can do. The rest is up to the hardware. If Mantle can get there, so will DX and the rest. Then its RIP Mantle. But atleast it sparked up a effort to push software further. -
Dear PlanetSide 2,
Please get Mantle support, or at least DX11, so this stops happening to me:
Simple File Sharing and Storage.
Simple File Sharing and Storage.
Oh and while you're at it, 64-bit would be dandy too, so um, you know, I can run Ultra textures without crashing to desktop whenever your 32-bit client runs out of memory.
You're a great game with such potential, but this level of performance ruins it for me.
K thnx baiLink4 likes this. -
Isn't Planetside 2 also limited to dual cores?
-
-
-
And still waiting for Mantle driver... If I could change, I would change my vote, AMD just disappointing me too much. Waiting is killing!
-
-
-
-
-
Give them some time. It's not that easy to push out something as complicated as a graphics API.
I agree that their P R could be more down to earth though. -
Sent from my HUAWEI Y300-0100 using Tapatalk
Maxwell or Mantle makes you more excited?
Discussion in 'Gaming (Software and Graphics Cards)' started by JKnows, Nov 15, 2013.