The AMD Mantle Thread

[Despoiler]

About the crossfire results:
http://battlelog.battlefield.com/bf3/forum/threadview/2955064773334491797/

A) Why is the 3970X running at 3.5Ghz when tested and not 4.0GHz?
B) Why is it running in 60Hz mode when they tested with DX but 120Hz when tested with Mantle?
C) Why does it only get 78FPS with 290x CF running DX with 1080p with 4xMSAA when Sweclockers got 102FPS with 290x CF running 1440p and 4xMSAA?
http://www.sweclockers.com/nyhet/17...-r9-290x-crossfire-och-r9-290-i-battlefield-4


Something is very fishy here

Not really.

a)It's reporting the base clock like windows device manger does. The other scenarios with different processors they report the base clock as well.
b)Different machines with different monitors. Why does it matter since both are v-sync off?
c)Look at the system spec for the B4 scenario vs swede clockers. They aren't the same. Also, the B4 scenarios are at specific snapshot in the game whereas swede clockers is a benchmark run.

 

[Shivansps]

This looks like CPU bound to you guys?

 

[bystander36]

Precisely the point, you have no data to make any conclusive verdict on performance of a setup that has yet to be observed.

So far every benchmark on GPU bound setups gave very little benefit. I gave no conclusive verdict either. Why are you so upset about my statement?

Are you saying that an i7 3770k @ 4.0 with a single R9 290x on a 4k screen is going to be CPU bottlenecked while using as high of settings as he can? Probably not.

I said that he probably would not see much improvements in his case and you certainly gave no information to show otherwise.

I did not, in any way, try to insult Mantle. I simply stated a fact based on all the benchmarks given. Mantle helps CPU bound situations and barely improves GPU bound situations. Is that not true?

I took that info, and gave the opinion that it probably will not help him, because he has a fast CPU, a single GPU, and a 4K resolution.

 

[Gloomy]

Looks Direct3D-bound.

 

[cmdrdredd]

What?!

At least the scree shots and numbers that AMD posted shows a 58% increase in frame rate. What else could the 50% mean if you say its not FPS?

Look again and read what I said. "All cases" That's important here. Not everyone will magically see 50% increases. It won't happen. It's very specific setups to get maximum benefit and from the looks of it, you already have more than enough performance to begin with (CF 290x).

 

[Stuka87]

Look again and read what I said. "All cases" That's important here. Not everyone will magically see 50% increases. It won't happen. It's very specific setups to get maximum benefit and from the looks of it, you already have more than enough performance to begin with (CF 290x).

Ahh yes. I did gloss over the "all cases" part. My apologies. It will obviously vary depending on setup and what is going on in game.

 

[Stuka87]

This looks like CPU bound to you guys?

You have one core pegged (Thread is bouncing between cores a bit). So yes, you are CPU bound as the demo obviously is not multi-threaded.

 

[Shivansps]

NO, ive seem less complex DX9 games pegging it at 90 all the way to 100%, if the bench is not using MT, last time i checked DX11 supported it, them the benchmark is not done properly.

Looks Direct3D-bound.

no no no, dont even start making excuses, even if using DX11 single threading rendering i dont see 1 core at 100%.

I gona start thinking the benchmark has been manipulated to make Mantle looks better, but nah, AMD does no do that sort of things

 

[Imouto]

Pfft,

If only such claims could be independently tested right now by a review website such as guru3d or PCPer to verify such findings. As is, it sounds like an extreme cherry picked result. But, I don't know. A single GPU in GPU limited situations giving 0-5% gain while crossfire giving 58% gain?

It's about halving CPU frame times. If you have 12 ms for the CPU and 16 ms for the GPU you're GPU limited. If you add a second GPU and your GPU frame times drop to 8 ms but your CPU frame times remain the same 12 ms you're CPU limited. With the same dual GPU system your performance will improve with better CPU frame times until they best your GPU frame times.

Yeah. Okay. Whatever. Sure. Probably a result to put AMD in the best light possible in the weirdest CPU limited situation possible. Is it possible? Maybe, but I don't think that is a representative or "typical" result, and I highly doubt it. What we need is for independent reviewers to test such claims.

Isn't this "weirdest CPU limited situation" the same as reviewers all around the world maxing all the sliders when testing a game? I thought it was fine when testing impossible/retarded amounts of tessellation, AA and such.

 

[cmdrdredd]

Ahh yes. I did gloss over the "all cases" part. My apologies. It will obviously vary depending on setup and what is going on in game.

This is why I really want to see what the average increase is. In other words. When someone has an overclocked i5 and a single 290, what can they expect vs say a 780 with the same CPU. This will be important to me and I'm sure many. Crossfire and SLI are still pretty niche and people with those setups are sitting pretty anyway and may not really see the benefit. They may feel it in terms of a smoother experience with fewer variations in the highs and lows. If I had a pair of 290x's in my system I wouldn't be worried about lack of performance. I think it'd be pretty great as it is. Getting a smoother experience is more important to me than outright FPS once there's enough FPS to sustain smooth motion in the game. Keep the mins from dropping and keep frames from skipping etc.

 

[AznAnarchy99]

I want to see reviews using an old i5-2500k overclocked.


That being said, I want to see reviews. If the only thing Mantle gives us is a MS that actually improves DX more than it has, it's a win for gamers. If we get more from that, it's gravy to me.

That's what Im thinking for myself. I'm still running a first gen i7. I wonder if this will help me out a ton.

 

[Cloudfire777]

Not really.

a)It's reporting the base clock like windows device manger does. The other scenarios with different processors they report the base clock as well.
b)Different machines with different monitors. Why does it matter since both are v-sync off?
c)Look at the system spec for the B4 scenario vs swede clockers. They aren't the same. Also, the B4 scenarios are at specific snapshot in the game whereas swede clockers is a benchmark run.

The two CPUs are pretty equal in performance actually. The only difference is that the 3970x is running @4GHz while 4960x@4.4GHz.

That should not mean 30% more FPS on higher resolution...
Both DICE and Sweclockers did a multiplayer run.

 

[Despoiler]

You have one core pegged (Thread is bouncing between cores a bit). So yes, you are CPU bound as the demo obviously is not multi-threaded.

Also, he is doing it wrong for comparison's sake. Please provide the text dump of the benchmark run. Something like this (not mine).

High



== Configuration ==========================================
API:	 DirectX
Scenario:	 ScenarioAttract.csv
User Input:	 Disabled
Resolution:	 1920x1080
Fullscreen:	 True
GameCore Update:	 16.6 ms
Bloom Quality:	 High
PointLight Quality:	 High
ToneCurve Quality:	 High
Glare Overdraw:	 8
Shading Samples: 64
Shade Quality:	 Mid
Motion Blur Frame Time:	 16
Motion Blur InterFrame Time:	3
Detailed Frame Info:	 Off
================================================== =========


== Results ================================================
Test Duration:	 120 Seconds
Total Frames:	 5667

Average FPS:	 47.22
Average Unit Count:	 3547
Maximum Unit Count:	 5328
Average Batches/MS:	 746.21
Maximum Batches/MS:	 2043.45
Average Batch Count:	 17333
Maximum Batch Count:	 60286

 

[Despoiler]

The two CPUs are pretty equal in performance actually. The only difference is that the 3970x is running @4GHz while 4960x@4.4GHz.

That should not mean 30% more FPS on higher resolution...
Both DICE and Sweclockers did a multiplayer run.

Well and the part where they are totally different architectures.....

Dice took a snapshot at that exact point the game where the CPU usage spikes. It's not a benchmark run.

 

[frozentundra123456]

Looks like Mantle may be able to save people CPU/MB upgrades and allow a sizable performance increase with just a video card. Not bad :thumbsup:

Only if you want to play the 0.01 percent of games that support mantle.

 

[Shivansps]

This is it.

===========================================================
Oxide Games
Star Swarm Benchmark - ©2013
C:\Users\Shivan\Documents\Star Swarm\Benchmark_14_01_30_2141.txt
Version 0.95
01/30/2014 21:41
===========================================================

== Hardware Configuration =================================
GPU: AMD Radeon HD 5800 Series
CPU: GenuineIntel
Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz
Physical Cores: 4
Logical Cores: 4
Physical Memory: 8569073664
Allocatable Memory: 8796092891136
===========================================================


== Configuration ==========================================
API: DirectX
Scenario: ScenarioAttract.csv
User Input: Disabled
Resolution: 1920x1080
Fullscreen: False
GameCore Update: 16.6 ms
Bloom Quality: High
PointLight Quality: High
ToneCurve Quality: High
Glare Overdraw: 0
Shading Samples: 8
Shade Quality: Mid
Motion Blur Frame Time: 16
Motion Blur InterFrame Time: 8
Detailed Frame Info: Off
===========================================================


== Results ================================================
Test Duration: 120 Seconds
Total Frames: 1744

Average FPS: 14.53
Average Unit Count: 3418
Maximum Unit Count: 5619
Average Batches/MS: 57.52
Maximum Batches/MS: 514.40
Average Batch Count: 4116
Maximum Batch Count: 11913
===========================================================

 

[OCGuy]

So I was correct, Mantle will not help my 290 and Haswell CPU. Can't wait for actual reviews on this.

 

[Slomo4shO]

So far every benchmark on GPU bound setups gave very little benefit. I gave no conclusive verdict either. Why are you so upset about my statement?

Every GPU bound setup? What sample size do you have? 1? So I am now upset because I refute your baseless speculation?

PC Perspective has shown 4 benchmarks:

• Core i7-4960X CPU + R9 290X GPU
  • 1080p, Ultra Preset, 4xAA: 9.2% improvement with Mantle
  • 1600p, Ultra Preset, 4xAA: 10% improvement with Mantle
• Core i7-4960X CPU + R7 260X GPU
  • 1080p, Ultra Preset, 4xAA: 2.7% improvement
  • 1600p, Ultra Preset, 4xAA: 1.4% improvement
• A10-7700K CPU + R9 290X GPU
  • 1080p, Ultra Preset, 4xAA: 40.9% improvement
  • 1600p, Ultra Preset, 4xAA: 17.3% improvement
• A10-7700K CPU + R7 260X GPU
  • 1080p, Ultra Preset, 4xAA: 8.3% improvement
  • 1600p, Low Preset: 16.8% improvement

Dice provided 3 benchmarks in their blog:

If other benchmarks exist, I would be happy to see them.

But of what is known to be available, only one of these benchmarks can conclusively be deemed GPU bound. Considering that none of the benchmarks show performance details of quality changes, by what basis are you concluding the performance based on the question on hand?

Do you folks think that Mantle will improve 4K performance? I have BF4 @ 150% rendering; everything else maxed. Would be nice to hit 200% without a performance impact

Giving blind advise is just as ignorant as willfully accepting opinion as facts:whiste:

I'll wait for the drivers to be release before I further this discussion.

 

[Nintendesert]

Only if you want to play the 0.01 percent of games that support mantle.

There are tons of threads on this forum and the PC gaming forum where people upgrade over a single game. BF4 despite its bugs caused a lot of upgrades. As long as that's the one game someone really wants they may very well upgrade around it.

 

[el etro]

Imagine the system advices:


Before Mantle: $140 CPU(i3) + $200 GPU(270x)

After Mantle: $70 CPU(750k) + $250+ GPU(7950 Boost/cheap 280x)


---------------


Before Mantle: $230 CPU(i5) + $300 GPU(280x)

After Mantle: $110 CPU(FX 6300) + $400 GPU(290)


---------------


Before Mantle: $230 CPU(i5) + $400 GPU(290)

After Mantle: $160 CPU(FX 8320) + $550 GPU(290x)


------------------

Before Mantle: $600 CPU(i7 4930) + $800GPU(2 x 290)

After Mantle: $300 CPU(i7 or 9590) + $1100GPU(2 x 290x)

 

[OCGuy]

^ I hope the suggestion above that people should recommend weak CPUs is just misguided enthusiasm.

 

[smackababy]

^ I hope the suggestion above that people should recommend weak CPUs is just misguided enthusiasm.

That suggestion might be good, if you only play BF4.

 

[AnandThenMan]

^ I hope the suggestion above that people should recommend weak CPUs is just misguided enthusiasm.

Misguided enthusiasm is thinking you know how others should build their systems. Different people have different needs depending on budget. If you need a very fast CPU and only a basic GPU, that doesn't mean any of the above configs are invalid for others.

 

[PPB]

^ I hope the suggestion above that people should recommend weak CPUs is just misguided enthusiasm.

No, its the possibility mantle opens for people shelling even more cash on the gpu than ever.

This only proves those cpus arent weak, rather that the api we all are using right now (dx) is weaksauce. Mantle might open the eyes of the people in charge of making that weak api get better (MS).

Ps: and please dont even bring up dx's dumb brother. If it was actually half as good as some people claim, OGL wouldnt have such a hard time competing against DX.

 

[Paul98]

Looks like those who got 7950X2 when they were under 200 will really get their moneys worth.