Mantle API Vs. DirectX Thief Benchmarks

[Mr.Duck]

 

I'm going to let the benchmarks talks and keep quiet...

 

P.S: Well... yeah 3 FPS improvment on 1440p much wows... those game on 720p they got what was promised... and 1080p improvments are not that bad too but still not %30

 

SOURCE:http://wccftech.com/mantle-api-directx-thief-benchmarks-direct3d-creamed/

[dragosudeki]

Now waiting for the screenshot comparison like what BF4 did. >.<

[Scionyde]

I know it's only 720p, but is Mantle really expected to give such insane gains at those lower resolutions? O_o

[Vitalius]

Now waiting for the screenshot comparison like what BF4 did. >.<

That was a bug on BF4's side. The developer said so. I doubt it will happen again.

[lhikary]

mantle at a game no one cares about,..,

[dragosudeki]

That was a bug on BF4's side. The developer said so. I doubt it will happen again.

I understand it was, though has it been fix and benchmarked again?

[Vitalius]

I understand it was, though has it been fix and benchmarked again?

It was fixed in the Feb 25th update. Assuming the sky fog is what caused it.

I still get the same FPS I did with the washed out color, so I guess that didn't matter at all to FPS. 

Which, based on how those patch notes describe it, it was just the fog being too bright making it more noticeable and diluting the other colors. 

[matin94]

where are the APU benchmarks? after all, thats what matters most, more performance in low end hardware

[Mincygrape]

If DirectX 12 comes out with a low-level api, then mantle won't matter anyway. If Microsoft screw PC over again and don't release a low-level api, then hopefully mantle does become more widespread.

[ForsakenLive]

Personally, I think they are running the benchmarks wrong, Mantle it's not going to show big performance improvements on the GPU side, but on the CPU side. I want benchmarkers to try  using different CPUs and GPUs combinations, We want to see how much a Athlon X4 750k with a R7 260X can gain in performance. I want to see different CPUs both high end and low end combined with low end and high end GPUs.

[Mincygrape]

Hold on, in the source it says 'The exact test rig was not mentioned though I am pretty sure it was a i7-4770K (downclocked)'

Who downclocks a CPU? Why would you ever do it, a lot of people overclock, so they should have tried that instead.

Underclocking the CPU means that this isn't 'real world' performance, so these benchmarks don't really mean anything.

 

Sadly, this happens with a few mantle benchmarks. They always do something weird that isn't what normal people do. For example, benchmarking a 780 Ti vs a R9 290x, and using a low end APU. Obviously in this case mantle will win, but in a real world test, with i7 4770k's overclocked and the GPU's overclocked, mantle doesn't really do much. The 780 Ti still wins.

[warmmilk]

270x did better on directx than mantle at 1080p...

 

but these results are useless without know which cpu was used and at what clock speeds... the purpose of mantle is to remove the cpu bottleneck, so whats the point of posting results without telling use which cpu is used?  + what @Mincygrape said about the cpu.  if it is indeed a downclocked HT disabled i7 then it further takes away from the validity of this test

[Vitalius]

Hold on, in the source it says 'The exact test rig was not mentioned though I am pretty sure it was a i7-4770K (downclocked)'

Who downclocks a CPU? Why would you ever do it, a lot of people overclock, so they should have tried that instead.

Underclocking the CPU means that this isn't 'real world' performance, so these benchmarks don't really mean anything.

 

Sadly, this happens with a few mantle benchmarks. They always do something weird that isn't what normal people do. For example, benchmarking a 780 Ti vs a R9 290x, and using a low end APU. Obviously in this case mantle will win, but in a real world test, with i7 4770k's overclocked and the GPU's overclocked, mantle doesn't really do much. The 780 Ti still wins.

*facepalm*

The point of Mantle is to remove the CPU bottleneck. That's it. That's all it does to improve performance. Therefore, you have to create a CPU bottleneck to show the improvement Mantle offers. 

While I do agree no one is going to get a 290X/780 TI with an APU, that is to simulate any build a person has where they have a strong GPU and weak CPU. Because they can't test every single CPU/GPU combination, so they choose the strongest GPUs with a weaker CPU and just roll with it. 

Thus, downclocking the 4770k. Making it a weaker CPU.

270x did better on directx than mantle at 1080p...

 

but these results are useless without know which cpu was used and at what clock speeds... the purpose of mantle is to remove the cpu bottleneck, so whats the point of posting results without telling use which cpu is used?  + what @Mincygrape said about the cpu.  if it is indeed a downclocked HT disabled i7 then it further takes away from the validity of this test

Yes. Because there isn't a CPU bottleneck for the 260X and 270X. There won't be a difference without a CPU bottleneck.

That's why you get diminishing returns as the resolution goes up. Because the bottleneck becomes the GPU and Mantle does nothing for that.

[warmmilk]

I figured it out... they invented mantle so that coin miners can game on their sempron cpu's

[dragosudeki]

It was fixed in the Feb 25th update. Assuming the sky fog is what caused it.

I still get the same FPS I did with the washed out color, so I guess that didn't matter at all to FPS. 

Which, based on how those patch notes describe it, it was just the fog being too bright making it more noticeable and diluting the other colors. 

Good to hear. Didn't really follow what happened after that problem. Still haven't tried BF4 myself let alone Mantle.

[cramved]

Mantle does not like 1080p and beyond

[Mr.Duck]

Mantle does not like 1080p and beyond

than they shouldnt have promised for %30+ FPS İmprovment when 80% of the market uses 1080p and above

[MrSuperb]

*facepalm*

The point of Mantle is to remove the CPU bottleneck. That's it. That's all it does to improve performance. Therefore, you have to create a CPU bottleneck to show the improvement Mantle offers. 

 

So how likely is it, that you have a crapy CPU toegether with an >500$ GPU ?

 

If they want to show performance, they should make real life examples and not some artifiical test scenarios making their software look good ...

[matin94]

Personally, I think they are running the benchmarks wrong, Mantle it's not going to show big performance improvements on the GPU side, but on the CPU side. I want benchmarkers to try  using different CPUs and GPUs combinations, We want to see how much a Athlon X4 750k with a R7 260X can gain in performance. I want to see different CPUs both high end and low end combined with low end and high end GPUs.

exactly, because mantle is meant to remove that bottleneck. high end gpu with low end cpu, or other way around would be what i want to see aswell

[theking]

mantle at a game no one cares about,..,

No one cares about? Are you serious?

[Vitalius]

So how likely is it, that you have a crapy CPU toegether with an >500$ GPU ?

 

If they want to show performance, they should make real life examples and not some artifiical test scenarios making their software look good ...

Fairly good, tbh. IMO at least.

Just wait til mass selling of AMD mining GPUs on ebay happens. Just wait.

[Sky Daddy]

My results (built in benchmark) Maxed settings & AA @ 1080p FX8320 @ 4ghz and r9 290 at stock clocks
DirectX: 
Low 27
Avg 41
High 71
 
Mantle:
Low 45
Avg 58
High 79

 

Bar Graph Concept for future benchmarks (color/professionalism may vary)

 
The increase in low fps is both impressive and important, and the jump from 41 average fps to 58 is a ~40% increase. The lowest fps on Mantle is higher than the avg fps on DX, can't see why people sould think mantle doesn't do anything...

If I were to show my high fps and only my high fps, as many other results do, the gain doesn't look very significant.

[Vitalius]

My results (built in benchmark) Maxed settings & AA @ 1080p FX8320 @ 4ghz and r9 290 at stock clocks

DirectX: 

Low 27

Avg 41

High 71

 

Mantle:

Lower 45

Avg 58

High 79

 

The increase in low fps is both impressive and important, and the jump from 41 average fps to 58 is a ~40% increase. Can't see why people sould think mantle doesn't do anything...

If I were to show my high fps and only my high fps, as many other results do, the gain doesn't look very significant.

I personally hate benchmarks that only show 1 of those FPS'. Most reviewers use Avg (even Linus). That is terrible to use because it basically tells you nothing about the experience in playing a game with that graphics card at those settings (whatever they may be).

They're all important for different reasons, but of them, AVG is the least important because it's the most deceitful.

[warmmilk]

My results (built in benchmark) Maxed settings & AA @ 1080p FX8320 @ 4ghz and r9 290 at stock clocks

DirectX: 

Low 27

Avg 41

High 71

 

Mantle:

Low 45

Avg 58

High 79

 

Bar Graph Concept for future benchmarks (color/professionalism may vary)

 

The increase in low fps is both impressive and important, and the jump from 41 average fps to 58 is a ~40% increase. The lowest fps on Mantle is higher than the avg fps on DX, can't see why people sould think mantle doesn't do anything...

If I were to show my high fps and only my high fps, as many other results do, the gain doesn't look very significant.

 

I don't think anyone was saying mantle doesn't do anything, we were saying the test was useless cause it just had a guess at what the cpu was.  your test with just 1 card and cpu is much better cause it actually tells us the cpu used and clock speed.

 

I wonder what the results would be like with an OC'd i5 or i7...

[nicehat]

No one cares about? Are you serious?

 

Dont feed the nVidia troll. In other news, wccftech isnt a source to be trusted. Other than masquerading factually inaccurate content as news, they dont really do any journalism themselves. They repost stuff from some shady corners of the internet. Fox News and The Onion have more newsworthiness than these guys. For some reason, the OP keeps using it. The original source, in german isnt a well known source either. I'll reserve judgement until someone with some real street cred benchmarks this.