AMD Mantle API [updating]

[DavidGraham]

Yeah, the point is to focus on the important questions first, which are:

-With a good CPU at max settings, Is Mantle significantly faster in MP or just barely?
-With a good CPU at max settings, Is Mantle suffeciently faster to allow for R290X to beat 780Ti in MP?

A 3 test bed scenario as previously described would be sufficient to answer these questions .. I would guess a good i7 3770 or i7 2600 would be enough , a high level i5 is also good.

I don't like the prospect of testing Mantle with weak CPUs because that point is moot. why the heck would anybody be happy with a low CPU that only handles 2 or 3 Mantle games, while chocking and limping on the rest! if we are going to talk about benefits to the average player, we should stay away from extreme unrealistic cases.

 

[lanek]

Yeah, the point is to focus on the important questions first, which are:

-With a good CPU at max settings, Is Mantle significantly faster in MP or just barely?
-With a good CPU at max settings, Is Mantle suffeciently faster to allow for R290X to beat 780Ti in MP?

A 3 test bed scenario as previously described would be sufficient to answer these questions .. I would guess a good i7 3770 or i7 2600 would be enough , a high level i5 is also good.

I don't like the prospect of testing Mantle with weak CPUs because that point is moot. why the heck would anybody be happy with a low CPU that only handles 2 or 3 Mantle games, while chocking and limping on the rest! if we are going to talk about benefits to the average player, we should stay away from extreme unrealistic cases.

Dont forget , all the MP test with 780TI windows 8.1 have been made on high quality, no AA, 1080p.... there's no way you push 170% difference in Ultra with a 780TI between win7 and win 8.1 in BF4... ( i can tell you, everybody will use windows 8.1 if it was possible to bring 170% faster on games with it )..

I have no idea how Nvidia have manage to push a 70% gain in fps with the high quality in BF4 with win8.1 ...

 

[Andrew Lauritzen]

I have no idea how Nvidia have manage to push a 70% gain in fps with the high quality in BF4 with win8.1 ...

Win8.1 has some stuff that allows you to reduce CPU overhead (like Mantle, obviously not to the same extent) that BF4 makes use of as well.

 

[lanek]

Win8.1 has some stuff that allows you to reduce CPU overhead (like Mantle, obviously not to the same extent) that BF4 makes use of as well.

I know, but +70% performance is an huge leap ( on some review it is even 2x the fps )..

Ok, you dont buy a 7800TI for play on high at 1080p anyway.

And you should see the same gain with lower model ( if gpu bottlenecked, even in low details )

 

[Andrew Lauritzen]

Ok, you dont buy a 7800TI for play on high at 1080p anyway.

Sure, but by the same token none of the >60Hz results are really relevant even if it's OMG 2x 100->200fps! Ultimately if it was a multiplayer test though I wouldn't read too much into the results as I've argued earlier in the thread. The simultaneous spec mode suggestion is the best I've heard on the topic but to my knowledge no one has done that.

And you should see the same gain with lower model ( if gpu bottlenecked, even in low details )

I assume you mean CPU bottlenecked. And sure, but I will point out that there are two independent factors in comparing the CPU side: both the driver and the CPU arch. It's quite possible that Intel CPUs handle the sort of code that DX drivers do better than AMD CPUs, so it's not necessarily fair to extrapolate gains on AMD CPUs to similar Intel ones.

Really the question is if Mantle enables someone to buy a cheaper config to get a solid 60Hz in BF4 than was previously possible.

 

[Gipsel]

Sure, but by the same token none of the >60Hz results are really relevant even if it's OMG 2x 100->200fps!

Quite the opposite. 60fps in BF4 MP are kind of low and most frequent player say it feels sluggish (especially with DX, as not only the frame time distribution is worse but it appears also the absolute lag if you want to believe people who tried Mantle, the same fps feel smoother with Mantle). Some may even shoot for consistent 120 or 144 fps to match their monitor refresh.

 

[Blazkowicz]

With Quake 3 at least once you could afford a high framerate (and can see it, > 100Hz refresh) you always wanted more. At least for me. 100 fps average wasn't cutting it, about 150 fps average was the sweet spot.

I just played against bots btw. Had fun playing on Voodoo2 before, even though I only got like 55fps with vertex lighting (CPU limited), that's when I actually played the game on LAN with human opponents..

 

[DavidGraham]

Dont forget , all the MP test with 780TI windows 8.1 have been made on high quality, no AA, 1080p....

No, at least two tests has been made with Ultra 4XAA. and we are talking about SP , Mantle's gains there has been limited to about 10%.

 

[neliz]

Updated PurePC.PL benchmarks:

http://www.purepc.pl/karty_graficzn...t_graficznych_battlefield_4_mantle_vs_directx

780Ti / 290X / 770 / 280X

CPU & OC

Power Consumption tests, loads of resolutions/settings. impressive work.

 

[mc6809e]

Updated PurePC.PL benchmarks:

http://www.purepc.pl/karty_graficzn...t_graficznych_battlefield_4_mantle_vs_directx

780Ti / 290X / 770 / 280X

CPU & OC

Power Consumption tests, loads of resolutions/settings. impressive work.

Very nice!

This is interesting:

Radeon R9 290X (Mantle)
Core i3-4130, 3400 MHz, 2/4
83.5fps

Radeon R9 290X (DirectX 11)
Core i5-4670K, 3400 MHz + TB, 4/4
79.8fps

 

[sir doris]

Updated PurePC.PL benchmarks:

http://www.purepc.pl/karty_graficzn...t_graficznych_battlefield_4_mantle_vs_directx

780Ti / 290X / 770 / 280X

CPU & OC

Power Consumption tests, loads of resolutions/settings. impressive work.

That's very interesting, how did they conduct their CPU tests? (translation not working for me at the moment)

 

[neliz]

That's very interesting, how did they conduct their CPU tests? (translation not working for me at the moment)

Translation seems fine for me.. anyway, everything you need is in the tables 1920x1080 ultra, no MSAA Win7x64.

Though one interesting note is that they only noticed the "stutterbug" on native dual-core chips, not quad cores.

 

[itsmydamnation]

i think he means how did they make it cpu bound and gpu bound

 

[neliz]

i think he means how did they make it cpu bound and gpu bound

That's what I said, all the settings are in the graphs' tables

 

[Malo]

Apparently turning off MSAA alone is enough to make it CPU bound.

 

[CarstenS]

Apparently turning off MSAA alone is enough to make it CPU bound.

Hm, I don't know about their scene, but in our (deliberately CPU-focused) CPU-benchmark, which is done without MSAA, FXAA and AO disabled as well as post-processing and texture-filtering at low, I get an additional 10-ish-fps, if I lower the rendering resolution via the in-game slider down 50% from the 720p at which we test CPUs normally.

 

[Svensk Viking]

I don't like the prospect of testing Mantle with weak CPUs because that point is moot. why the heck would anybody be happy with a low CPU that only handles 2 or 3 Mantle games, while chocking and limping on the rest! if we are going to talk about benefits to the average player, we should stay away from extreme unrealistic cases.

You're thinking too much as an enthusiast. I have alot of friends with Phenom II X4 and X6 CPUs that primarily play Battlefield 4, and with Mantle they have been given a reasonable upgrade path for the graphics card. Same goes for all folks with the first generations of core i5 and i7 CPUs, and core i3s.

I think the opposite of what you think, I think it's silly of most reviewers to just test their Ivy bridge i5 and i7s and call it a day

 

[Deleted member 13524]

I don't like the prospect of testing Mantle with weak CPUs because that point is moot. why the heck would anybody be happy with a low CPU that only handles 2 or 3 Mantle games, while chocking and limping on the rest! if we are going to talk about benefits to the average player, we should stay away from extreme unrealistic cases.

What you're calling a "low CPU" is pretty much everyone with a mid-tier desktop or any laptop with a GCN graphics card.

The only "extreme unrealistic case" I see in way too many reviews is those tests using overclocked 500-1000€ Intel CPUs, since less than 1 gamer in every 200 uses such a CPU.


I do understand that graphics cards reviewers want to avoid having their CPU as a bottleneck, but what they don't understand is that most of them end up alienating 99.5% of the PC gamers by not giving them a realistic example of what to expect with that particular graphics card.
Intel CPUs between 2.0 and 3.3GHz together with AMD CPUs end up making 70% of the whole pie.

 

[function]

More than 50% of people in the Steam survey have two or fewer CPU cores.

And some of the 3, 4, 6 and 8 core machines will be from AMD's 25.6 % of the sample group, who will benefit from avoiding the single thread bottleneck.

There are far more gamers with a lot to gain from Mantle than Joe F. Nerds with their overclocked i7 watercooled machines who have nothing to gain.

Besides, just because my 2500K has the headroom to wade through the current API garbage doesn't mean that I wouldn't rather it was doing things that would make the games better.

I don't get the hate for Mantle from "enthusiasts".

 

[Alexko]

Hm, I don't know about their scene, but in our (deliberately CPU-focused) CPU-benchmark, which is done without MSAA, FXAA and AO disabled as well as post-processing and texture-filtering at low, I get an additional 10-ish-fps, if I lower the rendering resolution via the in-game slider down 50% from the 720p at which we test CPUs normally.

Careful there, or you'll get banned by Andrew…