Mantle drops Performance?

[wxyzsupermod]

I have crossfire r9 270's and an i5 4670k all stock in bf4 if I use mantle my fps on high and ultra sits at 44 fps if I switch to dx11 I get 84 but some spots on the map are flickering black this is fixed by mantle but I can't deal with the fps drops and lag down to 15 when in a firefight so I have to deal with the glitches what is going on?

 

[DMTron]

Idk what the problem is but same thing happened to me with thief, but my fps dropped by 10

 

[bgunner]

For BF4 you need to have 3GB or more of dedicated video memory. Unless both of your 270's have this then you will have issues in this game with mantle. This is a known limitation for this one game with mantle. As for the the black spots on in the game with DX 11, try different driver versions to see if this will help the remove the graphic anomaly's.

EDIT: Also just FYI Mantle does not like Overclocked GPU's. the API will crash the game, any game, it you OC the GPU.

 

[thor220]

Mantle works better with an AMD CPU. In addition, most games don't support mantle with a crossfire setup. They figure that only people with low end rigs are going to need the boost that mantle brings. You most likely aren't using your 2nd card at all.

 

[bgunner]

Mantle works better with an AMD CPU.

Is there any reading material on this subject as this is the first I am hearing this? I has anyone done a though test of this because I would like to read the results of this.

As for using it on low end rigs, I do not have a low end rig and I use it in both Sniper Elite 3 and Thief to help maintain a 60 FPS minimum. Although because of Thief's graphic demands on maximum settings I am unable to maintain 60 FPS at all times either way. But I do get a slight improvement using mantle in Thief.

 

[thor220]

Here you go

http://www.techspot.com/review/793-thief-battlefield-4-mantle-performance/

It may be because AMD designed mantle or that their processors are just slower in single-threaded.

 

[barto]

Here you go

http://www.techspot.com/review/793-thief-battlefield-4-mantle-performance/

It may be because AMD designed mantle or that their processors are just slower in single-threaded.

I personally don't care for that site. I've always felt that the data is fake. There's no way an A8-7600 can compete with an FX8350 or an i7. Mantle does improve performance but it the stronger the CPU the less performance boost there will be. IMO that article proves absolutely nothing.

 

[iron8orn]

mantle is really meant for a weaker cpu with a strong gpu.. like if you have a athlon with a 280x the gpu should handle the off load of the cpu and more fps would be seen.
a 4670k does not need mantle and the 270 is not the strongest of gpu's for the extra load so you get worse performance..
not sure if i am 100 percent correct but seems to make some since..

what driver are you using? omega?

 

[bgunner]

Here you go

http://www.techspot.com/review/793-thief-battlefield-4-mantle-performance/

It may be because AMD designed mantle or that their processors are just slower in single-threaded.

I personally don't care for that site. I've always felt that the data is fake. There's no way an A8-7600 can compete with an FX8350 or an i7. Mantle does improve performance but it the stronger the CPU the less performance boost there will be. IMO that article proves absolutely nothing.

After reading the article for my self I agree that the only thing this article really says is that the Mantle API only helps in situations that the CPU becomes a bottleneck. A powerful CPU sees very little performance improvement at all.

As for the data aspect, Did you notice that they used faster ram on the i7 cpu than the rest of the rigs? For any test to be valid, IMO, the setups need to as equal as possible. You want to test how well a i7 4770 performs against a FX 8350 you need all other parts to run at the same exact speed. The same applies with GPU's the rest of the system parts besides CPU's need to run at the same speeds or you inherently add in an extra variable that can skew the results.

 

[thor220]

Here's a link from tom's
http://www.tomshardware.com/reviews/amd-mantle-performance-benchmark,3860-8.html

As you've guys have already said, it's only really meant for weaker CPUs. As an additional disadvantage, devs. must support multi-GPU setups if they are to be utilized.

 

[iron8orn]

would think the same things will be seen with dx12.. a stronger gpu will be needed for more load

 

[bgunner]

would think the same things will be seen with dx12.. a stronger gpu will be needed for more load

Actually you are correct that the DX12 API is going to allow for better hardware access. Now as for Mantle vs DX12 only time will tell which will be better at allowing applications to access the hardware. Right now there are mostly rumors concerning DX12 and which GPU's and OS's will be supported. Bahh humbug I want conformation ! lol

 

[renz496]

isn't that mantle was suppose to be more lower level than DirectX? then why it needs more VRAM in BF4 than DX11 counterpart?

 

[bgunner]

isn't that mantle was suppose to be more lower level than DirectX? then why it needs more VRAM in BF4 than DX11 counterpart?

Just because the API allows better access to the hardware does not mean that requirements for things to run better can not be different. Mantle's main design was to help lower end CPU's game better not help lower GPU requirements. The BF4 issue is a bug and was the very first game to support Mantle so you need to expect it to be a little buggy.