The AMD Mantle Thread

[DiogoDX]

The pc.lab slides strikes again.:biggrin:

 

[Sisyphean]

If Mantle is what it takes to get multithreaded DX performance out of AMD cards. Just wierd they needed a new API for that instead of fixing their DX driver. But I wonder how much the coffers can hold to sponsor companies to use it.

I think there were devs asking for a low level API like Mantle for quite a while. The problem with DX are more than just whether DX is able to perform similarly to a newly released driver for a newly made API. Just observe the rumors about Maxwell. We have an environment where hardware sophistication is exceeding the ability of software developers to keep up. Is the solution just slap on more hardware power to solve this problem? No, because it further encourages slow software adaptation to hardware advances.

It's not a case of software developers being lazy either, it's just that the hardware is advancing that damn fast. DX, and the sheer size of it, will have an increasingly difficult time keeping up with features. Cutting DX out of the picture would be healthy in the long run since developers and consumers won't have to wait on MS to give (or lock features) in slow increments that segment the market. Having a low level (and hopefully open source, sometime) API means that developers will have more freedom and flexibility to include features without having to rely so heavily on a single company.

 

[frozentundra123456]

This is why I want to see anands review. Hopefully he will do a thorough test under a wide variety of conditions. So far the results are all over the place and each side is saying whatever doesn't fit their agenda is invalid.

 

[cytg111]

If Mantle is what it takes to get multithreaded DX performance out of AMD cards. Just wierd they needed a new API for that instead of fixing their DX driver. But I wonder how much the coffers can hold to sponsor companies to use it.

- Hahaha .. nice spin .. You sure looked long and hard for that angle, lets hope it sticks. (or not).

 

[ASM-coder]

With all the activity on this board, don't we have someone who plays online BF4 with an R9 290x, who
can share their current performance(FPS) stats. Pre-Mantle of course.
And then post Mantle stats later??
I know anecdotal data isn't the same as official benchmarks, but it's certainly would be better than some of the arbitrary speculation I see posted some times.

 

[cytg111]

yup my english is a bit bad and sorry for that.

Wish I could edit your post and smack you around a bit with a large trout.

 

[Carfax83]

Considering this is the R9 290 and not the R9 290X, I still don't see how PC.Labs were able to obtain a ~40% performance increase over the R9 290X...

Isn't it obvious? 1080p vs 1600p

 

[Deders]

Don't know if anyone's done this yet, Just tried the DX11 Star Swarm stress test on my hardware and with motion blur disabled it's very smooth, framerates varying from over 100 to around 40 fps.

 

[Erenhardt]

This is why I want to see anands review. Hopefully he will do a thorough test under a wide variety of conditions. So far the results are all over the place and each side is saying whatever doesn't fit their agenda is invalid.

I think you may be a bit disappointed because to me, it looks like anand doesn't even know mantle is out. There is not a single mention of it on the page. They probably didn't receive the beta press driver.

 

[Nintendesert]

http://www.anandtech.com/show/7728/battlefield-4-mantle-preview



Preview is up.



Unfortunately it's just average FPS, I want to see some graphs showing if Mantle has smoothed out the framerates too like some of the slides we saw. Next week a more indepth review though.

 

[SiliconWars]

I think you may be a bit disappointed because to me, it looks like anand doesn't even know mantle is out. There is not a single mention of it on the page. They probably didn't receive the beta press driver.

Spoke too soon

http://www.anandtech.com/show/7728/battlefield-4-mantle-preview

~10% on an ultra high-end rig in SP seems to be the norm. Dave at B3D says this -

http://forum.beyond3d.com/showpost.php?p=1824818&postcount=1061

So basically yeah, 10% is "worst case" for those who have extreme high-end overclocked single card setups, playing single player. It's only going to get better from there with MP and Crossfire etc.

 

[Slomo4shO]

Isn't it obvious? 1080p vs 1600p

Are you insinuating that there is a ~25-30% performance delta between two GPUs because of the drop in resolution?

 

[SiliconWars]

19% extra in SP with 4770K @ 4.5GHz, max settings 1080p

http://translate.google.it/translat...-mit-mantle-erster-eigener-benchmark/&act=url

I dunno but for me that's pretty huge. There will be a lot of heavy gamers with a rig exactly like that who just gained 20% for free.

 

[DiogoDX]

19% extra in SP with 4770K @ 4.5GHz, max settings 1080p

http://translate.google.it/translat...-mit-mantle-erster-eigener-benchmark/&act=url

I dunno but for me that's pretty huge. There will be a lot of heavy gamers with a rig exactly like that who just gained 20% for free.

The gain is above other sites because they are using Win7. The perf gap will be smaller in single player under Win8.1.

 

[Paul98]

Spoke too soon

http://www.anandtech.com/show/7728/battlefield-4-mantle-preview

~10% on an ultra high-end rig in SP seems to be the norm. Dave at B3D says this -

http://forum.beyond3d.com/showpost.php?p=1824818&postcount=1061

So basically yeah, 10% is "worst case" for those who have extreme high-end single card setups, playing single player. It's only going to get better from there with MP and Crossfire etc.

Yeah the results are looking great! Looks like with my setup I should never drop below 60fps with mantle, I might even be able to up resolution percentage. This is the first game with mantle, beta drivers, beta API. With more work being done on all of them. Things will get better from here, and this is a great starting point.

 

[SiliconWars]

The gain is above other sites because they are using Win7. The perf gap will be smaller in single player under Win8.1.

Gotcha. Still, a crapload of people on Win 7 yet I bet, and with even less incentive to "upgrade" now.

 

[SiliconWars]

it can give us an idea of what kind of performance gains we can expect if developers chase the low-hanging fruit offered by Mantle. What is that low-hanging fruit? For the most part that is going to be CPU bottlenecks, specifically bottlenecking in issuing draw calls. Of all of the bottlenecks that can impact a high performance GPU, keeping it fed can be the biggest bottleneck, and in turn bottlenecking in the draw call submission phase can be the biggest culprit. In the long term Mantle will also benefit GPU performance more directly by optimizing workflows within a GPU, and we already see a small bit of that today in Battlefield 4, but the bulk of the optimizations for these earliest titles have been made around the draw call bottleneck.

Ryan's article is the first I've seen mention that the GPU optimizations have barely even started, so kudos to him for that.

This is important guys - we ain't seen nothing yet.

 

[Erenhardt]

Spoke too soon

http://www.anandtech.com/show/7728/battlefield-4-mantle-preview

~10% on an ultra high-end rig in SP seems to be the norm. Dave at B3D says this -

http://forum.beyond3d.com/showpost.php?p=1824818&postcount=1061

So basically yeah, 10% is "worst case" for those who have extreme high-end single card setups, playing single player. It's only going to get better from there with MP and Crossfire etc.

Yea... perfect timing...
Full reviews will not come soon. That is a lot of benching to be done. I would like to see how mantle benefits amd CPU vs intel

 

[Abwx]

Phenom II X6 users will be happy.....

http://www.anandtech.com/show/7728/battlefield-4-mantle-preview

 

[VulgarDisplay]

As i doubt DICE benchmark that said CF had 58% performance increase is turly for AMD marketing.
http://www.hardwareluxx.de/index.ph...te-ergebnisse-im-kampf-mantle-vs-directx.html

There really not much difference in CF.

And then you click on the Multiplayer results link at the bottom of the page and get this...

Imagine that. When you cherry pick you can easily support your arguments.

 

[krumme]

As for Anand SP testing, no matter how precise, its just utterly useless if it doesnt reflect MP behavior. And we know, they are no where the same. If bf4 sp was a movie it would get 2.3 at imdb, and all play mp.

There is no exuses. At Siege of Shanghai with +60 on server you can eg. take the same tour with the scooters each time. Show the route. Die, repeat 20 times. Give us a median average and median min, and give us the variation.
Do that for eg a i5 4c 4t ib/hw at about 3.8GHz and 290 (sans x) at 1080p with and without msaa. Do the same for a 780.

I prefer less combinations instead of a useless sp bm of several combinations.

 

[ginfest]

Ryan's article is the first I've seen mention that the GPU optimizations have barely even started, so kudos to him for that.

This is important guys - we ain't seen nothing yet.

Wow, I'm surprised the mods let the blatant marketing continue. We have had 253 pages of BS and overstating the case about the next great thing, the results prove it, and now the crap continues?
Now that Mantle has been proven to be "not all that", we are told, just wait it will get better?
This was brilliant marketing move by a floundering company to sell their latest and greatest, based on promises. and it worked.
AMD will always be #2 in CPU and GPU and no amount of phoney promises will change that.

Warning issued for thread crapping.
-- stahlhart

 

[Slomo4shO]

Imagine that. When you cherry pick you can easily support your arguments.

So his cherry picked results are inferior to your cherry picked results?

AMD will always be #2 in CPU and GPU and no amount of phoney promises will change that.

Seems you need a history lesson...

Warning issued for thread crapping.
-- stahlhart

 

[Cloudfire777]

So there we have it. Unless you are running a crappy Intel i3 or AMD APU, there are very small gains with Mantle over DX. Even i5 that cost around $200 is not being a bottleneck with a powerful GPU like 290x.

Good news for low end and those who crossfire 290x or run 780Ti SLI. But that needs to be verified.
Plus Im very curious to see how CPU bound games like Starcraft will perform with Mantle. Im sure we will see better results there. If they get Mantle support that is.

 

[VulgarDisplay]

So his cherry picked results are inferior to your cherry picked results?

The point is that he selected the 1 chart out of that review that showed his point.

I'm not claiming that Mantle is giving 3000% performance increases. More the fact that people are intentionally ignoring benchmarks that will show where it really does what it's supposed to.