The AMD Mantle Thread

[OCGuy]

AMD Mantle Currently Causes Washed Out Graphics Compared to Direct X in Battlefield 4

Read more: http://wccftech.com/amd-mantle-washed-graphics-compared-direct-battlefield-4/#ixzz2sPUDpBs1
​

This would render every single benchmark done so far completely invalid.

 

[Gloomy]

Looks like the Mantle path doesn't have/has screwy color grading.

 

[desprado]

The pics posted at WCCFtech , Mantle on the left and DX on the right :

Are u serious?

Dx is on left(Image which is using fraps) and mantle is on right.

 

[Abwx]

Are u serious?

Dx is on left(Image which is using fraps) and mantle is on right.

Ask the site , here what they wrote :

Observe the two images below, both screenshots from Battlefield 4. The first one is from Mantle and the second from Direct X.

 

[desprado]

This would render every single benchmark done so far completely invalid.

Agree..When all mantle bug are fixed then we will have our actual results.

 

[Gloomy]

Ask the site , here what they wrote :

It's wtftech, as soon as you see the retarded robot you should know the content isn't theirs.

 

[desprado]

Ask the site , here what they wrote :

U Really serious.He miss wrote it lol Mantle dont use fraps.U cannot use any OSD with Mantle.It is Not D3d.

 

[Abwx]

U Really serious.He miss wrote it lol Mantle dont use fraps.U cannot use any OSD with Mantle.It is Not D3d.

Whatever the pics i gave no appreciation because the scene
is not the same at all.

 

[senorbiscuits]

I got so frustrated with people comparing different views that I took these. I'm sure they will eventually fix this, I hope, because Mantle is destroying the beautiful blue skies! :'(

[/URL][/IMG]

[/URL][/IMG]

 

[Venomous]

Nice screenies. I'm sure a few drivers down the road they may get this fixed or provide sliders in CCC or something. Surely there will be mantle controls in CCC at some given point.

 

[desprado]

I got so frustrated with people comparing different views that I took these. I'm sure they will eventually fix this, I hope, because Mantle is destroying the beautiful blue skies! :'(

[/URL][/IMG]

[/URL][/IMG]

There is also Image quality difference to.

 

[DrDoug]

About the image quality/differences between DX11 and Mantle in BF4: https://twitter.com/repi/status/429985248995467265

It's a bug, not a feature.

 

[OCGuy]

Full explanation from a source that isnt that horrible wcc :

www.extremetech.com/gaming/175998-b...washed-out-foggy-graphics-compared-to-directx

 

[Dresdenboy]

I think it's time for you to find another hobby.

You're constantly wrong. You're constantly negative. You've posted nearly 9000 times since April 2012 and you barely contribute anything worthwhile.

Give it a thought, would you?

That might not work. I know similar behaving people in RL. It's best to leave them as they are. Wrong forum here to elaborate this further. He at least creates motivation to bring up more undeniable facts. But this might be the plan, to keep ppl being busy, hehe.

 

[DrDoug]

Full explanation from a source that isnt that horrible wcc :

www.extremetech.com/gaming/175998-battlefield-4-amds-mantle-causes-washed-out-foggy-graphics-compared-to-directx

I see that the title of their story makes it look like the "washed out" look is because of Mantle but further into the story Joel notes that Johann (Repi) has stated that the problem is on their end and not a problem with Mantle.

The drapes don't match the carpet.

 

[omeds]

I've seen reviews as recent as this month and it's not reflected in them. Until the top sites we typically use reflect those numbers, it's a one off fluke in my book. You can believe it if you want, but it doesn't make sense to me.

Which reviews are those?

Sure I'll believe it, I'm getting the same results on my own system. It makes perfect sense, if you understand the notion of CPU overhead.

 

[krumme]

I have a question -
BF4 does a lot of directcompute stuff. Does directcompute work with mantle or are they converting it to opencl?

Thats a good question if i understand it correctly.
If we look at mantle execution model dice have the option to do it in a radical new way. And also use compute functions to eg order gfx stuff before it goes into gfx pipeline like done in ps3.
But my guess is we have to wait a while for stuff like that to happen because its more complicated stuff. What we know is dice have made a completely new renderer for mantle in fb3. Its not a patch. So i think in due time we will see all the funky stuff comming. For fb4 and bf5 i think we will see it all in action. And i think the compute stuff is going to be a big thing here.

 

[itsmydamnation]

i think from all the data we can see there is something from a uarch perspective that is giving NV a big advantage that isn't seen in other games. people with agendas are then using that performance to try and claim mantle is worthless vs DX11.x/win8.1/insert junk here, yet they can't explain 780 vs 290x on a 7850K compared to a 780 vs 290x on a 4770k.

so the question is what is causing that difference? im pretty sure kepler supports some buffer formats at full rate that AMD only do at 1/2 rate. Kepler has a setup rate advantage and that would also align to NV performance scaling on 7850 vs 4770k. GCN and Kepler are simlarish in there memory hierarchy but given memory bandwdith is pretty equal i think thats less likely.

any other ideas?

 

[krumme]

The future engines will surely be build to perform on the consoles and get the most of them. To get most of the performance i think using the excecution model to the extreme is a part of it. If the cost of adopting mantle specific engines and code for the games is less than the benefits here, mantle will succeed.

We dont know how much code is similar and how much is compatible. And we still dont know the full benefits of using mantle. And what is also unknown is the compettitive benefit mantle gives vs no mantle. Also the benefits of agressive consoles optimizations because they will move to mantle pc.

Cod still sells like hotcakes on the consoles even though bf4 is far superior.

I think its basicly a question of the benefits of a technically better game and the cost to get there. Mantle is just another tool here.

 

[jj109]

any other ideas?

A combination of the following:

1. D3D wants more single threaded performance due to the serial nature of command list playback.
2. D3D has more overhead so deferred rendering will cause D3D threads to collide with game threads and each other.
3. At a certain point, draw calls just aren't the bottleneck.

 

[Carfax83]

i think from all the data we can see there is something from a uarch perspective that is giving NV a big advantage that isn't seen in other games. people with agendas are then using that performance to try and claim mantle is worthless vs DX11.x/win8.1/insert junk here, yet they can't explain 780 vs 290x on a 7850K compared to a 780 vs 290x on a 4770k.

so the question is what is causing that difference? im pretty sure kepler supports some buffer formats at full rate that AMD only do at 1/2 rate. Kepler has a setup rate advantage and that would also align to NV performance scaling on 7850 vs 4770k. GCN and Kepler are simlarish in there memory hierarchy but given memory bandwdith is pretty equal i think thats less likely.

any other ideas?

We've already gone over this several times in this thread already. It's the drivers..

And it's not just BF4 either. Game engines that can utilize multiple threads give NVidia the advantage, as NVidia's drivers are better tuned for multicore processors.

Only two engines that I'm aware of can use 8 threads. Frostbite 3, and CryEngine 3. In both BF4 and Crysis 3, NVidia has a significant lead in CPU limited scenarios using Direct3D. The Mantle pathway is still more efficient though, and will be improved upon no doubt:

But it's too late for Mantle to gain a solid foothold in the industry. AMD needed Mantle to address their shortcomings under Direct3D, but I doubt they ever intended for it to become a replacement for Direct3D..

If Mantle failed when it came to developer adoption, at least it would shine a very bright light on the problem of multithreading in Direct3D and force Microsoft to come up with better ideas.

 

[Freaksterz]

We've already gone over this several times in this thread already. It's the drivers..

And it's not just BF4 either. Game engines that can utilize multiple threads give NVidia the advantage, as NVidia's drivers are better tuned for multicore processors.

Only two engines that I'm aware of can use 8 threads. Frostbite 3, and CryEngine 3. In both BF4 and Crysis 3, NVidia has a significant lead in CPU limited scenarios using Direct3D. The Mantle pathway is still more efficient though, and will be improved upon no doubt:

But it's too late for Mantle to gain a solid foothold in the industry. AMD needed Mantle to address their shortcomings under Direct3D, but I doubt they ever intended for it to become a replacement for Direct3D..

If Mantle failed when it came to developer adoption, at least it would shine a very bright light on the problem of multithreading in Direct3D and force Microsoft to come up with better ideas.

I'd ask for another site giving those numbers but I know there isn't so..

 

[mikk]

We've already gone over this several times in this thread already. It's the drivers..

And it's not just BF4 either. Game engines that can utilize multiple threads give NVidia the advantage, as NVidia's drivers are better tuned for multicore processors.

Only two engines that I'm aware of can use 8 threads. Frostbite 3, and CryEngine 3. In both BF4 and Crysis 3, NVidia has a significant lead in CPU limited scenarios using Direct3D. The Mantle pathway is still more efficient though, and will be improved upon no doubt:

I doubt it is related only to a better threading, I would assume it is related to a lower CPU overhead as well, Nvidia seems to handle high drawcall numbers better than AMD under directx 11.

I'd ask for another site giving those numbers but I know there isn't so..

There is: pclab.pl

 

[itsmydamnation]

We've already gone over this several times in this thread already. It's the drivers..

no you haven't and no it isn't... i notice that you haven't explained the bit i pointed out as well.

the funny thing is lots of people who work with or on GPU's say DX multithreaded submission say it is fundamentally broken.

heres an Intel GPU architect saying its broken:
http://forum.beyond3d.com/showpost.php?p=1825331&postcount=1158
http://portfolio.punkuser.net/

Yeah, you can get some minor gains in a few applications, but it's typically stuff like eating *two* submission threads for barely double-digit performance improvements (in CPU bound cases)... not really that exciting and doesn't scale at all beyond 2. It's clearly something that needs a fundamental rethinking, a la. Mantle.

pffft he doesn't know what hes talking about.........

heres Johan Anderson says its broken
https://twitter.com/repi/status/428896915049414656

we support it but it is fundamentally broken in DX. check my slide #34 from 3 years ago:

we all know repi is a lying shill :awe:

and here more of the stuff im talking about, possible reasons for NV's overall superior showing.

http://beyond3d.com/showpost.php?p=1747841&postcount=531

Some time ago, I did some testing with my Tahiti to determine some of it's L2 properties. One of the things I noticed immediately, was that Tahiti is already BW bound when you just alpha blend to a 4x16 bit float HDR render target, even if you are not sampling any textures at all (just write solid color triangles). Fast GDDR5, wide 384 bit bus and a total of 264 GB/s bandwidth, and a simple untextured blend is able to spend all of that BW. I roughly calculated that it would require more than 500 GB/s BW to reach full Tahiti fill rate in this scenario. So I split the viewport to 128x128 tiles, and rendered each tile separately. This doubled the performance (reached maximum theoretical fill rate). 128x128 tile of 4x16 bit float color + 32 bit depth buffer size is 196 KB in memory. It easily fits the 768 KB L2 of Tahiti, and thus all the blending passes and depth read/writes occur completely inside the L2 (no memory BW used at all). It seems that cache optimizations are very important for these new breed of GPUs. Even simple things such as sorting objects by screen space XY location (in addition to depth) could bring nice reductions in backbuffer BW usage (= big performance gains for the BW starved APUs).

It's hard to say how the large L2 caches of GCN (and Fermi/Kepler) affect performance of current generation games. Current generation consoles do not have large general purpose GPU caches, so most developers have surely not analyzed their L2 cache behavior or changed their rendering methods to exploit the L2 caches of the recent hardware. Cache optimization was (and still is) one of the major performance improvements for CPU code. If you don't design around caches, you still get some gains from the caches, but the biggest gains require careful design. So I would expect to see biggest gains in new games that are designed from the ground up for the new GPUs.

Drivers can get in the way of performance but they dont magically create higher then peak performance as determined by the hardware. The simple fact mantle scales so well on low end CPU's when NV doesn't ( magical DX multithreading working so well!) yet scale CPU performance high enough they end up in front and AMD stop scaling at not much more then a 7850K.

There's no question that NV seems CPU limited up to around a 4.5ghz 4770K in multiplayer. but thats not the question, the question is why do they have so much more performance when not CPU limited vs Mantle on hardware that on many other games is far closer. ie 290x battling the 780 and not the 290.


also if DX multithreading works so well, why does it gen pwnd so hard by openGL extensions?
https://static.slo-tech.com/52734.jpg
https://static.slo-tech.com/52736.jpg

 

[jj109]

I doubt it is related only to a better threading, I would assume it is related to a lower CPU overhead as well, Nvidia seems to handle high drawcall numbers better than AMD under directx 11.

It's mostly better threading. This is why Mantle pulls way ahead with Intel dual cores. No matter how well the renderer distributes workload there just isn't enough limited CPU resources to go around.

I'd ask for another site giving those numbers but I know there isn't so..

People like you are hilarious. It's either dismissal based on gut feeling or linking irrelevant GPU-limited benchmarks.