• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Ashes of the Singularity Coming, with DX12 Benchmark in thread.

Calin Banc said:
A nice discussion for whoever wants to go through it - http://www.overclock.net/t/1569897/...singularity-dx12-benchmarks/400#post_24321843
Click to expand...

Very good read, thanks for that.

Anyways, as DX12 games come and go, I'm sure we'll see games that are faster on AMD , and faster on Nvidia hardware. When you have 2 so different architechttures, you're bound to have games that play more on strenghts of another. It actually suprsised me how often games are running about equal with so different architectures.
 
I feel this time AMD will have the upper hand because of the Consoles and how they all inner connect with AMD hardware and the APIs

Think of it this way you building a game for Xbox One on DirectX 12 same AMD hardware found on PC you would think the work load between console and PC would be much less now. Resulting in better AMD gaming Performance!

Nvidia will still need to have guys working on games with Devs and driver tweaking. AMD already have there work getting done for them.

This how am seeing it, could very well be wrong but I guess time will tell.
 
I think DX12 performance will be better on Day 1 considering that Devs can now see what is going on. So problems can be fixed on their end immediately.

The guys post was accurate and pretty much what i was saying about the fury. The AMD drivers just can't feed it fast enough to fill all of its shaders at lower resolutions. Them AMD architectures are highly parallel from the start. While nvidias are more of a branching architecture capable of better utilising the single data stream.

But it was also interesting to see how much of a boost the 290/390 received. It is funny to see it competing with the 980 when it was originally pegged against the Original Titan as the titan killer.
 
i7 4790k @ 4.7Ghz
Sapphire R9 290 @ 1100/1300


DX11:
aots.png

DX12:
aots2.png

All graphics settings were set as high as they could all go.
 
Last edited:
Oh wait, you used the wrong test benny, you need to do the "Full system test" otherwise it wont show your actual GPU scores.

Edit - He updated the pics since this post.
 
Last edited:
Looking good :) Performance is around what i see on other sites. Although there is some CPU bottlenecking at the GPU's are not hitting 100% at medium and heavy calls.
 
Looking at the AotS forum a lot of people are having problems getting the DX12 version to run. Most seem to be Nvidia customers. :p

I hope Nvidia sort out their drivers, recent ones from what i've seen have been pretty poor.
 
Mauller said:
Yeah, checkout AMDMatts picture here of his CPU load.

https://twitter.com/TheMattB81/status/633615796200210432

He was running the CPU benchmark, but it essentially shows that the game engine will use ALL available cores.
Click to expand...

Cool thanks.

So DX12 scales up to all 16 (maybe more cores) but yet the 6700k out performs the 5960x. :confused:

Surely a faster clockspeed and slightly improved IPC cant make up the difference and better the performance of 8 additional cores!
 
Last edited:
opethdisciple said:
Cool thanks.

So DX12 scales up to all 16 (maybe more cores) but yet the 6700k out performs the 5960x. :confused:

Surely a faster clockspeed and slightly improved IPC cant make up the difference and better the performance of 8 additional cores!
Click to expand...

No , 6700k don't overperform it. In DX12 any GPU is gpu bound with both of those cpu's. Only difference in benches is that they are different runs. And you can get like 5fps variation between runs, as benchmark changes.

In DX11 actually 6700k can perform 5960x , as it has better ipc, so it'f faster on clock to clock than 5960x.
 
opethdisciple said:
Cool thanks.

So DX12 scales up to all 16 (maybe more cores) but yet the 6700k out performs the 5960x. :confused:

Surely a faster clockspeed and slightly improved IPC cant make up the difference and better the performance of 8 additional cores!
Click to expand...

I Do not believe that DX12 is limited to 6 cores. I think that the GPU's were merely hitting the limit at how many drawcalls they can process when 6 cores were used. But that was in a very limited and lightweight test in terms of graphics. The test was seeing the drawcall throughput more than anything else.

With Ashes a full game simulation is occuring, so the game engine is splitting all workloads across as many cores as you have. Game code and Graphics.

The 5960X simply outperforms all others because of its IPC and the number of threads it can crunch at once.

This benchmark is very CPU heavy before any GPU work is being prepared. IF you look at the breakdown info it will show that the game only uses up to 60k draw calls on the heavy test. and around 11k on the normal test.

But the game itself is still GPU heavy due to the sheer number of effects and individual units that are being rendered.
 
Last edited:
opethdisciple said:
On the Haswell E CPU's are all cores/threads being utilised under DX12?
Click to expand...

Nope

When I ran this bench using DX12 HWMonitor reported that all the cores were being used on my 5960X but only at around 65%, some were higher some lower and none of them got to 100%.

The comical part about it is I was playing Civ5 (DX11) this morning and that managed to get nearly all the cores to 100% lol.
 
All cores/threads are being utilised, but you never want an instance when all cores/threads are at 100% usage. This is a good thing and DX12 is doing its magic.
 
You must log in or register to reply here.
Back
Top Bottom