DirectX 12

[LtMatt]

Now this is the best example you can give non Mantle users of the Mantle effect. Super low and super consistent frame times. Comfortably beating AMD and NV DX11 frame times/frame variance - smoothness. Nvidia multi gpu users, you just don't realise how much better it can get. The dx frame times from both sides look similar, but Mantle is in a league of its own. Considering that is vs a 780TI (and SLI) which is typically faster (and more expensive) than the 290X in most but not all things, thats even more impressive.







And at the highest possible settings, gpu limited Xfire still the daddy when it comes to scaling, even before you add in the Mantle effect.





Considering the 280X has not had its performance optimizations yet its also putting out some respectable numbers vs the 770.

 

[Silent_Scone]

Graph wars lol. I loved the part where Matt picked an area that was supposedly very CPU limited asking me to then cripple my CPU to find it made absolutely no difference.

It's a waste of time in that game. Thief even less so, nevermind graph wars, see if someone on a 290 can beat my dx score on the thief thread when Mantle arrives lol. Can I just say whilst the AMD defence force are out in full swing just how crap the audio is in theif. Another wonderful demonstration yet again from their investment in willing participants.

FPS is life. Frametimes were only every really yabbered Matthew when they were so drastically bad on Crossfire. My Frame times are no where near as flat as Mantle at 4k yet its smooth as butter and I get more frames. If I wanted to look at flatlines I'd work in a care home. Oh and seeing as we are being Mantle fantastic pedantic again, 290 frametimes on Dx are not the same at all lol.

 

[andybird123]

Why do all the mantle tests not include minimums?
Far more important than averages

 

[LtMatt]

Graph wars lol. I loved the part where Matt picked an area that was supposedly very CPU limited asking me to then cripple my CPU to find it made absolutely no difference.

It's a waste of time in that game. Thief even less so, nevermind graph wars, see if someone on a 290 can beat my dx score on the thief thread when Mantle arrives lol. Can I just say whilst the AMD defence force are out in full swing just how crap the audio is in theif. Another wonderful demonstration yet again from their investment in willing participants.

FPS is life. Frametimes were only every really yabbered Matthew when they were so drastically bad on Crossfire. My Frame times are no where near as flat as Mantle at 4k yet its smooth as butter and I get more frames. If I wanted to look at flatlines I'd work in a care home. Oh and seeing as we are being Mantle fantastic pedantic again, 290 frametimes on Dx are not the same at all lol.

Not graph wars i posted those as someone mentioned them (the H results) above so i went looking for them and was pleasantly surprised by the results. Also the part you kept posting was not cpu limited, as we both found out. I did ask you to try the other part i posted on Oman but it fell on death ears. What was interesting was even your 780TI SLI overclocked (and 6 core cpu clocked) was still a fair few fps slower than old 290 pro xfire at 1000/1250, in a gpu limited scenario.

As you've not tried multi gpu mantle frosty, you're not in a position to comment on it. You may think things can't get smoother, but trust me they can. The dx frame times look similar to me. Ok so Nvidia has a couple more larger spikes but aside from those you won't notice a difference. However Mantle has no spikes, that you will notice only once you've tried one then the other. As i can easily switch between the two, it becomes immeditately apparent. On DX there is no noticeable difference between the frame times on either side.

We prepared this chart using data from our testing at 2560x1600, all in game settings set to maximum values and MSAA set to 4X. We did not experience the need to show you all the frame time graphs for all the single-GPU video cards. The frame times were well enough in those, but we did want to show you the improvements using Mantle with CrossFire and how it improves frame times.

Looking at frame time performance under DirectX 11 mode, both the AMD Radeon R9 290X cards in CrossFire and the NVIDIA GeForce GTX 780 Ti have a number of frames that take longer than 20ms to deliver (some going past 30ms.) When Mantle was turned on for the pair of R9 290X cards, frame times tightened up significantly compared to DirectX 11 mode. Both DirectX 11 and Mantle provided a smooth and enjoyable experience at these settings. It is clear that Mantle can and does benefit CrossFire performance and smoothness. This is positive for AMD since CrossFire has taken a bad wrap, albeit deservingly so, in the past for un-smooth gameplay.

Why do all the mantle tests not include minimums?
Far more important than averages

No idea but when it comes to minimum fps in this game AMD are ahead, even on DX. On Mantle minimums are even higher than on DX. I agree though, it would be nice to include them.

 

[Rusty0611]

Thought this was a DX thread? Why does any passing reference to Mantle (which is natural given the thread title) have to result in pages of defending / graph spam? Horrendous lads. Let the thread be back on topic.

 

[LtMatt]

Thought this was a DX thread? Why does any passing reference to Mantle (which is natural given the thread title) have to result in pages of defending / graph spam? Horrendous lads. Let the thread be back on topic.

My graph spam was comparing DX to Mantle Rusty poo's.

 

[SiDeards73]

wow you guys are still bitching and derailing threads with AMD vs Nvidia shizzle? Mantle vs DX, yada yada, go play games with the cards you bought which were incidentally designed to play games on, not just benchmark

Just RMA'd my Tri-X and back with my trusty Ref 290, i think im going to slap an Arctic X4 on it and just be done with it, its not let me down yet.

 

[Silent_Scone]

Frame times are worse in DX in Crossfire than SLI. They are there and they always will be. I told you at the time I was being fair on the test as I could have quite easily quoted and screened the higher 140 but went with an average. No doubt you weren't so generous .

I just said its smooth as butter? Would you like to come and try and spot the difference for me?

 

[LtMatt]

Frame times are worse in DX in Crossfire than SLI. They are there and they always will be. I told you at the time I was being fair on the test as I could have quite easily quoted and screened the higher 140 but went with an average. No doubt you weren't so generous .

I just said its smooth as butter? Would you like to come and try and spot the difference for me?

Just because you say something does not make it the case though. I'm going on the data HardOCP collected and reported on, as they have both cards and took the time to actually test the frame data. Their word > your word.

wow you guys are still bitching and derailing threads with AMD vs Nvidia shizzle? Mantle vs DX, yada yada, go play games with the cards you bought which were incidentally designed to play games on, not just benchmark

Just RMA'd my Tri-X and back with my trusty Ref 290, i think im going to slap an Arctic X4 on it and just be done with it, its not let me down yet.

Agreed, but you know what Frostys like. Can never admit he's wrong, even when he's wrong. (which happens a lot)

 

[Kaapstad]

We want to see some Mantle action @4K, I think the advantage over DirectX will be even smaller than it is at 1600p.

 

[LtMatt]

We want to see some Mantle action @4K, I think the advantage over DirectX will be even smaller than it is at 1600p.

Makes sense. I did see some 4k Mantle results somewhere, ill see if i can dig them up.

Or i could just test them myself i suppose.

 

[Silent_Scone]

Nerve touched? For someone who keeps banging on about trying things you sure go off other peoples word a lot lol.

You see the blue lines above the green ones? That's because they're higher. That's higher frametimes. Go and beat the NV cards in the one game that has a benchmark due to have Mantle. Although scaling is broke so don't hurt yourself too much.

You see the thing is, you won't be able to. Like in 99% of all the actual end user results found on this very forum.

 

[LtMatt]

Nerve touched? For someone who keeps banging on about trying things you sure go off other peoples word a lot lol.

You see the blue lines above the green ones? That's because they're higher. That's higher frametimes.

Impossible to notice small variations like that. What you will notice though is those big green spikes and HardOCP even commented on it. You won't see those in Mantle and that is the difference. Something you won't be able to experience yet until/if you try it.

 

[Rusty0611]

My graph spam was comparing DX to Mantle Rusty poo's.

DX12 isn't out though so not relevant.

 

[Calin Banc]

Not sure how many times I have to say it. I hope Mantle is a success and if it is working well for all, fantastic.

Now explain why AMD users are turning settings down, as I thought the idea was to have higher frame rates so you could add to the details?

If it is working correctly, then I apologise and good stuff.

I''m turning them UP Greg while getting better performance and I'm expecting the same from DX12 or it can GTFO.

On another note Mantle also has stuff to show

Check out @Thracks's Tweet: https://twitter.com/Thracks/status/441399872843825152

Matt, I'm disappointed by you. A console refresh does not equal a better hardware put into it, it's the same specs as the previous, just build on a better process. What that means, is a need for DX12 to have all those "Mantle features" support on current dx11 hardware or else it risks (and probably will) to go down on the same road with dx10 and dx11, which means to little love from game devs.

AMD will still push Mantle on it's own accord most likely with some devs and if it can get it to be easy to implemented and reliable, can coexist with dx easily. I doubt Dx can provide the same advantages as this one, unless M$ went the extra mile from the start and implemented all the architectures that are currently on the market.

The last but not the least, AMD and nVIDIA probably knew about this DX version IF it was in development for quite some time, ergo AMD would have not work on Mantle if DX showed real progress or at least would have some "perks" of it's own. At the same time, it could mean that M$ just now or recently started to work on DX12 and that could mean years from know until some games will have the darn thing.

But leaving all this aside, I'm still waiting on "next-gen" software to make it's way on our PCs. Besides Star Citizen/Squadron 42, there isn't one.

 

[Silent_Scone]

Impossible to notice small variations like that. What you will notice though is those big green spikes and HardOCP even commented on it. You won't see those in Mantle and that is the difference. Something you won't be able to experience yet until/if you try it.

I don't get the spikes in Tri SLi at 4k. Only when I tab out.

 

[Kaapstad]

DX12 isn't out though so not relevant.

I will be very surprised if there is any hardware that will support it even in a years time. I am pretty sure that anything AMD and NVidia have got coming up on 20nm will only use DirectX11.XX as these architectures will have been planed a very long time ago.

Lets face it if AMD knew that DirectX12 was just about to launch and had the cards to support it, would they even bother with Mantle.

 

[Silent_Scone]

Roy Taylor said it would never happen



What a stupid thing to say really. I wouldn't quote him specifically too much in the future Matt. Might cripple the illusion that you have a scoobydoo.

 

[LtMatt]

One thing we can all agree on, Roy Taylor > LT > Frosty.

 

[Silent_Scone]

THIEF is AMD's™Gaming Evolved™

SCOREBOARD

1080p

Single GPU

(Scores link to posts)

1. MjFrosty-780ti@1315/1965, [email protected]

2. gregster-Titan@1267/3767, [email protected]

3. zpaf-AMD 290@1200/1600, [email protected]

4. uksoldierboy- AMD 290X@1150/1400, [email protected]

5. tommybhoy-290X@1175/1475, [email protected]

6. Roy Taylor - N/A