Mantle decreasing performance in BF4?

[TrickyWitcher]

My brother has an R9 280 3GB. I switched from DX11 to Mantle. His FPS dropped exactly by 10. It's odd...

[helpingaskewcokemansarcasm]

How did you ascertain this, did you replicate the benchmark run correcly after switching APIs?

[Notree]

 Using the latest drivers? That is odd. I gained very little fps in mantle so I just use dx 11 anyways. Mantle aint all that great unless you have a lower end cpu. At least I havent seen any huge benefits for people using i5/i7/single gpu.

[thorpj]

My brother has an R9 280 3GB. I switched from DX11 to Mantle. His FPS dropped exactly by 10. It's odd...

Mantle doesn't always increase performance, this is expected

[JoshM]

Make sure his drivers are up to date.

 

Mantle doesn't always increase performance, this is expected

yes but in a game like bf4 this is very odd...

[thorpj]

Make sure his drivers are up to date.

 

yes but in a game like bf4 this is very odd...

In BF4, it is even less surprising. I've been through every performance issue out there when it comes to BF4. Solution to this is simple - use dx11.

[xAcid9]

Did u restart the game? d:

[typographie]

My brother has an R9 280 3GB. I switched from DX11 to Mantle. His FPS dropped exactly by 10. It's odd...

What CPU? Faster CPUs (Core i5's and i7's) that can already cope with DX11's overhead don't have much to gain, so there's no benefit to Mantle on that hardware. It wouldn't be outrageously surprising if there was even some decrease.

[TrickyWitcher]

What CPU? Faster CPUs (Core i5's and i7's) that can already cope with DX11's overhead don't have much to gain, so there's no benefit to Mantle on that hardware. It wouldn't be outrageously surprising if there was even some decrease.

He has an i5 4670K, like me.

How did you ascertain this, did you replicate the benchmark run correcly after switching APIs?

Yes, I left other settings on exactly what they were. Gave me 10fps less.

[xAcid9]

I get similar performance between Mantle and DX11 when i benchmarked iit n Single player.

[Monarch]

What CPU? Faster CPUs (Core i5's and i7's) that can already cope with DX11's overhead don't have much to gain, so there's no benefit to Mantle on that hardware. It wouldn't be outrageously surprising if there was even some decrease.

 

From what I've seen, no i5 can cope with AMD's DX overhead in BF4. Even i7's struggle. High-end i7 like 5820k might be able to mitigate this issue but still, that's why there's a huge difference between AMD's Mantle and AMD's DX.

 

 

 

@OP

Mantle can significantly help in a CPU-bound situation. If your GPU hits its limit before the CPU does then you won't really see an improvement. Checking GPU usages would help. But Mantle doesn't decrease performance. Do it again, monitor and record your GPU usage and report back.

[xAcid9]

From what I've seen, no i5 can cope with AMD's DX overhead in BF4. Even i7's struggle. High-end i7 like 5820k might be able to mitigate this issue but still, that's why there's a huge difference between AMD's Mantle and AMD's DX.

 

 

 

@OP

Mantle can significantly help in a CPU-bound situation. If your GPU hits its limit before the CPU does then you won't really see an improvement. Checking GPU usages would help. But Mantle doesn't decrease performance. Do it again, monitor and record your GPU usage and report back.

 

I take any numbers regarding AMD DirectX overhead from PCLab.PL with a bowl of salt.

Their number is exaggerated. I tried BF4 using friend's 290x Tri-X with my Xeon E3-1230V2 in both DirectX and Mantle, Mantle only give 5 FPS extra at best in my case and the 290x is definitely faster than a 780@1100mhz which i had in my shop that time. 

[Monarch]

I take any numbers regarding AMD DirectX overhead from PCLab.PL with a bowl of salt.

Their number is exaggerated. I tried BF4 using friend's 290x Tri-X with my Xeon E3-1230V2 in both DirectX and Mantle, Mantle only give 5 FPS extra at best in my case and the 290x is definitely faster than a 780@1100mhz which i had in my shop that time. 

 

Pclab is right. I'm getting the same results on my PC (4670k+290X). Bf4 DX is unplayable for me, average fps around 60. It never drops below 60 with Mantle.

I did a benchmark in Star Swarm and I've seen the same issue as in BF4. Here's how GPU usage looks like:

 

 

[xAcid9]

Pclab is right. I'm getting the same results on my PC (4670k+290X). Bf4 DX is unplayable for me, average fps around 60. It never drops below 60 with Mantle.

I did a benchmark in Star Swarm and I've seen the same issue as in BF4. Here's how GPU usage looks like:

 

 

 

You can't compare StarSwarm and BF4, StarSwarm was design to show CPU bottleneck for DirectX11 with tonnes of drawcalls way pass DX11 comfort zone. 

[Monarch]

You can't compare StarSwarm and BF4, StarSwarm was design to show CPU bottleneck for DirectX11 with tonnes of drawcalls way pass DX11 comfort zone. 

 

But both StarSwarm and BF4 manifest the same issue. DX/Mantle GPU usages are the same. 

 

This guy got 56.08 fps with GTX 970 in StarSwarm.

I got 37.94 fps with R9 290X (DX).

 

===========================================================

Oxide Games

Star Swarm Stress Test - ©2013

C:\Users\NJ\Documents\Star Swarm\Output_14_12_20_1300.txt

Version 1.10

12/20/2014 13:00

===========================================================

 

== Hardware Configuration =================================

GPU: AMD Radeon R9 200 Series

CPU: GenuineIntel

Intel® Core i5-4670K CPU @ 3.40GHz

Physical Cores: 4

Logical Cores: 4

Physical Memory: 8472465408

Allocatable Memory: 140737488224256

===========================================================

 

 

== Configuration ==========================================

API: DirectX

Scenario: ScenarioFollow.csv

User Input: Disabled

Resolution: 1920x1080

Fullscreen: True

GameCore Update: 16.6 ms

Bloom Quality: High

PointLight Quality: High

ToneCurve Quality: High

Glare Overdraw: 16

Shading Samples: 64

Shade Quality: Mid

Deferred Contexts (D3D11): Disabled

Small Batch Optimized (Mantle): Enabled

Temporal AA Duration: 16

Temporal AA Time Slice: 2

Detailed Frame Info: Off

===========================================================

 

 

== Results ================================================

Test Duration: 360 Seconds

Total Frames: 13661

 

Average FPS: 37.94

Average Unit Count: 4182

Maximum Unit Count: 5547

Average Batches/MS: 672.54

Maximum Batches/MS: 1534.21

Average Batch Count: 20389

Maximum Batch Count: 125874

===========================================================

 

So we're talking Nvidia's vs AMD's DX. You can't blame it on drawcalls. With Mantle I got 61.45 fps.

Pclab was right.

[Columbo]

With the Omega drivers Mantle runs very well. AMD has said Mantle is still in beta stages. It's geared towards the 3gb and 4gb cards. I play a lot of BF4 and I use mantle. It works very well. I am looking forward to the new AMD cards coming out early next year. Both AMD & Nvidia should be gearing their GPU's for 4K gaming.