The AMD Mantle Thread

[ASM-coder]

Is there any difinitive word on whether there has been an IQ regression with mantle? The mantle screenshots that I've seen seem to have more fog.

Again, Johan(Frostbite technical director) has said it's a bug, and doesn't affect performance. That's about as definitive as you can get, aside from them fixing the bug.

 

[ASM-coder]

Mantle gets a 40% hit with 4xAA. The performance lead pretty much disappears.

I don't think there's anything off with the AA, rather that Mantle has a much bigger GPU lead so they get hit harder with the 'real cost' of the AA when there isn't a CPU bottleneck.

I think you are framing this in the wrong manner. Mantle doesn't take a hit with AA,
everything takes a hit and Mantle shows less improvement.
Even if it showed 0% improvement, that wouldn't be Mantle taking a hit.
Still, according the Hardocp benchmark with 4xAA:
"AMD Mantle is providing a 7% performance improvement."

 

[SiliconWars]

You do know that site/blog is created by a person listed 277 on AMD influencer leaderboard? So lets not use those numbers at all since their credibility is zero.

Yep, exactly same.

Hilarous, Shintai. :biggrin:

 

[MeldarthX]

Is there any difinitive word on whether there has been an IQ regression with mantle? The mantle screenshots that I've seen seem to have more fog.

The fog is a bug that dice have comfirmed; which because of the fog makes it seem there is a regression; there isn't just seems to be a bit more contrast; and the fog.....

That should be patched out soon.

 

[Slomo4shO]

[H] has a multiplayer review up with a 4.8Ghz 3770K:

 

[96Firebird]

Why no testing at 4xAA at 1080p, since AA is usually beneficial at lower resolutions?

 

[Slomo4shO]

Why no testing at 4xAA at 1080p, since AA is usually beneficial at lower resolutions?

This seems to be the reason:

We think most of you will be shocked at what AMD Mantle is meant to do and the results it provides. If you thought AMD Mantle would provide a big performance boost for GPU limited, high resolution settings, you would be mistaken. In fact, it seems AMD Mantle is built to alleviate CPU bottlenecks, rather than GPU bottlenecks. Or at least, it helps GPUs when running on slower and more low-end CPUs and lower resolutions.

Our tests so far back this up. The lower the resolution we tested at, the great performance advantage we experienced with AMD Mantle over Direct3D 11.1. As the resolution decreases, and AA, the CPU becomes the bottleneck. AMD Mantle seems to excel in that scenario, providing substantial performance improvements, even with our overclocked i7 3770K. As we increased the resolution, and AA settings to GPU dependent settings, we experienced less of an advantage with AMD Mantle. However, there still was an advantage at least, so AMD Mantle is providing a "free" performance upgrade.

They do have a thorough review in the works so maybe we shall see alternative setting there:

Preview
Please remember, this is just our preview of performance of AMD Mantle on BF4. We have literally only had a day of testing with Mantle and BF4. More time needs to be spent, and more opinions need to be gathered.

We have a full in-depth evaluation planned for AMD Mantle and BF4. It will include not only the current R9 series GPUs, but we will also be testing the previous generation 7000 series for compatibility and performance. We will also be including the competition's performance. This will take us some time to test and put together. Remember, we cannot use FRAPS, so the current method is a bit slower, please be patient.

 

[stahlhart]

Hilarous, Shintai. :biggrin:

How so? Explain. I didn't find anything particularly funny when I read it, and I could use a good laugh today.

Citing you for a callout and a personal attack would serve this purpose equally well.

-- stahlhart

 

[el etro]

BTW, how much more does the GTX 780 GHz cost? Looks like $550+ vs. ~$450 for the standard edition. I don't see many listings here.

Costs some bucks more than GTX 780 reference. It have the same pricing as Classified.

 

[96Firebird]

BTW, how much more does the GTX 780 GHz cost? Looks like $550+ vs. ~$450 for the standard edition. I don't see many listings here.

Here in the states, it is only about $20 more. At Amazon, the cheapest 780 is $510, while the GHz is $520. At Newegg, the cheapest 780 is $490, while the GHz is $520.

 

[el etro]

The fastest 780 on the planet and the slowest 290X.

We don't know the clocks of the AMD card. At ~900Mhz clocks, surely will not beat 780Ghz/780Ti. At 1050Mhz clocks beat the first and tie the second.

Maybe a less confusing representation here:

Feels good to know that I was right about the drivers though.

This seems like linustechtips graph style.

Since win8.1 when CPU bound. Mantle certainly closes that gap though, or surpasses it.

That's why I put the GameGPU test with 3970x, to remove any CPU advantage of both sides:

Spoiler

[H] has a multiplayer review up with a 4.8Ghz 3770K:

1Ghz locked 290x(XFX Double Dissipation). Important Test.

 

[SiliconWars]

How so? Explain. I didn't find anything particularly funny when I read it, and I could use a good laugh today.

Yep Shintai telling us to throw out the polish results is definitely hilarious - It's no surprise that YOU didn't find it funny though.

Citing you for a callout and a personal attack would serve this purpose equally well.

-- stahlhart

Thanks for the ammo - not that I needed any more but this sure is helpful. You'll know exactly what I mean soon. :thumbsup:


Mod callouts are not acceptable. If you have a problem with the moderation, take it to the Moderation Discussion section.

-Elfear

 

[Slomo4shO]

1Ghz locked 290x(XFX Double Dissipation). Important Test.

Considering how well the XFX cards scaled in Crossfire, it would be interesting to see the outcome with mantle considering that they have an overclock review in the works:

 

[Will Robinson]

[H] conclusion Mantle:

Mantle could potentially, drastically change the gaming world if more games used it. Lower-end computers, especially notebooks, and ultrabooks could see significant gaming performance increases. AMD isn't done with Mantle, this is just a first taste of it. We have to say, this first taste is sweet. We also experienced no issues, no crashing, it was rock stable for all the pounding we gave it the whole day. We look forward to future improvements from DICE and AMD, and hopefully more games that will implement Mantle. Remember, we have more evaluation coming, so stay tuned.

That's a pretty promising conclusion by [H]

Both the game update and the driver delivered in January.(just)

 

[ASM-coder]

This seems to be the reason:
"We think most of you will be shocked at what AMD Mantle is meant to do and the results it provides. If you thought AMD Mantle would provide a big performance boost for GPU limited, high resolution settings, you would be mistaken."

Interesting that they would say that. Dice has been very clear, and I think most posters on this thread were well informed on these points.
(And this is probably the most active Mantle discussion thread on the internet.)
We all would have been "shocked" if we saw greater than 45% improvements.

 

[Stuka87]

Interesting that they would say that. Dice has been very clear, and I think most posters on this thread were well informed on these points.
(And this is probably the most active Mantle discussion thread on the internet.)
We all would have been "shocked" if we saw greater than 45% improvements.

Yeah, anybody that was actually keeping up on things was fully aware that this was for boosting CPU performance, not GPU performance. There may be some GPU performance increases, but AMD was certainly aiming towards CPU performance for this first release.

I am personally seeing over 50% performance increase on at least 64 player Shanghai. I went from hanging around 30fps to nearly sustaining 60fps with the same settings. It is a massive increase, but I was very CPU limited.

 

[Black Octagon]

I'm impressed by what I'm reading. I feel that the talk of 'ridiculing the Titan' added unnecessary hype but that aside Mantle seems genuinely promising., especially for Crossfire users.

If this API is a permanent thing then I suppose future AMD GPU reviews will need to benchmark with Mantle enabled for those games that support it...even when comparing the GPU under review to NV cards, and AMD cards that were pre-GCN.

 

[krumme]

Why no testing at 4xAA at 1080p, since AA is usually beneficial at lower resolutions?

One could also say why no test of 3570 no oc? And with a 290 sans x no oc.
In general this mixing of oc and nonoc test is a bad habbit imho. And the 2500k 3570 and 4670 4c 4t is pretty used cpu.

I had hoped mantle would enable me to play with msaa. It doesnt look like it. Johan have said they did i plement some of the new msaa specific hardware optimizations but msaa does take a toll. I think its quite aparent now when we are not cpu bound anymore - or at least just a tiny fraction of it.

But this smoothness to the gameplay is far more important and i will take that anyday to msaa.

 

[SiliconWars]

PCGH results look amazing - not sure if they've been posted already or not.

http://www.pcgameshardware.de/AMD-R...ecials/AMD-Mantle-Test-Battlefield-4-1107754/

29% in SP at 1080p + 4xMSAA + FXAA

39% in MP same settings 4770K.

 

[ASM-coder]

I am personally seeing over 50% performance increase on at least 64 player Shanghai. I went from hanging around 30fps to nearly sustaining 60fps with the same settings. It is a massive increase, but I was very CPU limited.

I'm shocked, but in a good way, more like astounded.
A question I haven't seen asked or answered yet.
How is competitive gameplay now? Besides smoother graphics that is.
I've often wondered if players with expensive rigs have any advantage.

 

[mikk]

PCGH results look amazing - not sure if they've been posted already or not.

http://www.pcgameshardware.de/AMD-R...ecials/AMD-Mantle-Test-Battlefield-4-1107754/

29% in SP at 1080p + 4xMSAA + FXAA

39% in MP same settings 4770K.

Tested on Windows 7.

 

[Shivansps]

Again, Johan(Frostbite technical director) has said it's a bug, and doesn't affect performance. That's about as definitive as you can get, aside from them fixing the bug.

The fog by itselft its not going to affect performance by much, BUT is a common trick used to hide something else.

 

[ASM-coder]

The fog by itselft its not going to affect performance by much, BUT is a common trick used to hide something else.

Are you suggesting that it was intentional?

 

[Stuka87]

I'm shocked, but in a good way, more like astounded.
A question I haven't seen asked or answered yet.
How is competitive gameplay now? Besides smoother graphics that is.
I've often wondered if players with expensive rigs have any advantage.

On smaller maps I would average 50-60fps average, so it wasn't too bad. But on Shanghai and other CPU intensive maps there was parts that slowed down enough to really hurt my aiming. And not having that issue anymore has helped. Once AMD/Dice fix the bug where every now and then the FPS drops way down causing a stutter is fixed, I will be very happy.

 

[PPB]

I see everything still rendered, just foggy as hell post levolution on shangai, for example.

But well, I actually play the game, what would I know...