videocardzGTX 1080, Polaris 10/11 Ashes of Singularity DirectX12 benchmarks leaked

[csbin]

http://videocardz.com/59725/nvidia-gtx-1080-polaris-10-11-directx12-benchmarks

As always there are few things to consider. Ashes of Singularity has many presets, so many that barely any site has ever tested all popular high-end cards in all of these presets. In our charts we are using average scores listed on Ashes website, however they might be manipulated by overclocked results, therefore they are not a good material for comparisons. Additionally Fury X and Fury non-X are both listed under the same device, so the average score is probably lower than it should be.
NVIDIA Graphcis Device is a generic name for unreleased GeForce GPUs. We do now know if this name covers only GTX 1080, but also GTX 1070 and other GPUs (including mobile).
The same database also lists Polaris 10 and 11 GPUs. Unfortunately they were not tested in as many presets as GTX 1080 was. So direct comparison is problematic at this point.

 

[Ancalagon44]

"Leaked"

No coincidence that such benchmarks come out soon after Nvidia's announcement.

 

[Keysplayr]

You sure this isn't GP106?

 

[railven]

I'm not evens sure what info can be taken from this chart.

Only thing for sure is Polaris 10 is not 90% of GTX 1080 in this data set.

EDIT:

You sure this isn't GP106?

If the lower numbers for the NVIDIA Device is GP106, woof.

 

[Keysplayr]

I'm not evens sure what info can be taken from this chart.

Only thing for sure is Polaris 10 is not 90% of GTX 1080 in this data set.

EDIT:


If the lower numbers for the NVIDIA Device is GP106, woof.

I'm with you on that chart. It's a mess.

 

[BlitzWulf]

http://www.ashesofthesingularity.co...36165f-4e15-4c4f-af35-7340095285f5?ladderId=x

Heres the profile with all the benches

Have to select Benchmarks under the match history drop down

 

[MangoX]

Strange. Why doesn't Polaris like not drop any fps when the gfx quality is increased?

 

[Timorous]

In the GFX bench leak the 67DF had half the texturing performance of a 390X. That does kind of fit with the 36 CU @ 800 Mhz indication on the SiSoft page too so if that is P10 it is a very low clocked and possibly cut down (if 2560 shaders for the full fat version is true) version of it

It actually lines up with the rumoured PS4 Neo GPU spec of 36CUs @ 911 Mhz. Perhaps this 800 Mhz version was the first test and 911 is as fast as it can go within Sony's desired power envelope? Total guess / speculation on my part though.

Or perhaps it is actually a full fat (16 CU) P11 running at around 1.6 Ghz? That would also fit with half the texturing performance of a 390X.

 

[Erenhardt]

Do we really believe one guy gets all the cards and doesn't give a jack about NDA?
I would like to be that guy.

 

[Noctifer616]

On the Polaris 10 the difference between medium, heavy and average is less than a frame. Seams odd doesn't it?

 

[Keysplayr]

On the Polaris 10 the difference between medium, heavy and average is less than a frame. Seams odd doesn't it?

Yes. Yes it does.

 

[railven]

On the Polaris 10 the difference between medium, heavy and average is less than a frame. Seams odd doesn't it?

Yes. Yes it does.

AMD hasn't "activated" the cards through drivers yet. I remember that happened sometime before. Think it was the 7970...

 

[Noctifer616]

AMD hasn't "activated" the cards through drivers yet. I remember that happened sometime before. Think it was the 7970...

So it's like testing the GPU without working drivers? I guess that could explain this odd behavior, also kinda makes the testing invalid?

 

[Keysplayr]

So it's like testing the GPU without working drivers? I guess that could explain this odd behavior, also kinda makes the testing invalid?

Yes. Yes it does.

I'm so sorry..

 

[beginner99]

So it's like testing the GPU without working drivers? I guess that could explain this odd behavior, also kinda makes the testing invalid?

Or it even is intentionally gimped in drivers to lead to confusing leaks...

 

[railven]

I'm pretty sure it was the 7970. AMD was still fine tuning the drivers. In fact, if I remember correctly, the card launched and you were told to use the driver on the included disc as it was not available on AMD.com and the driver found at AMD.com had performance issues with the 7970.

EDIT: About as close as I can get to verifying what I remembered:
https://hardforum.com/threads/7970-arrived-today-which-driver-adapters.1672038/
Note this thread is in February 2012, the 7970 paper launched in December, 2011 and was buy-able January 2012.

The 11.12 RC driver was used for reviews and was the recommended driver for a while.

 

[Broburger]

I can't wait to see the official benchmarks.

 

[PhonakV30]

DirectX 12 Ashes Of The Singularity – 1080p Crazy Preset

DirectX 12 Ashes Of The Singularity – 1440p Crazy Preset

Source

Nothing new in that source.

 

[parvadomus]

DirectX 12 Ashes Of The Singularity – 1080p Crazy Preset

DirectX 12 Ashes Of The Singularity – 1440p Crazy Preset

Source

Nothing new in that source.

So 15% performance over 980TI, stock vs stock. Maybe its 5% faster than Titan X @ stock LOL
Wonder what performance will be on max OC vs max OC :sneaky:

 

[Sweepr]

- Geforce GTX 1080

- Geforce GTX 980 Ti @ 1525 MHz Core & 3950 Mhz Memory

Not bad considering Async is off for NVIDIA cards.

Personally, I think one could just as easily make the claim that we were biased toward Nvidia as the only 'vendor' specific code is for Nvidia where we had to shutdown async compute. By vendor specific, I mean a case where we look at the Vendor ID and make changes to our rendering path.

Let's all hope those Polaris 10 results are way off, because if GFXBench and AotS leaks reflect what to expect from this particular card (Ellesmere Pro?) I'm not very confident about it's performance.

 

[master_shake_]

So 15% performance over 980TI, stock vs stock. Maybe its 5% faster than Titan X @ stock LOL
Wonder what performance will be on max OC vs max OC :sneaky:

die shrink with clock boost'll do that.

is this card nothing but a tock a la intel?

 

[dark zero]

Strange. Why doesn't Polaris like not drop any fps when the gfx quality is increased?

Someone tells me that is a Laptop card using Bristol Ridge CPU...

 

[LTC8K6]

So 15% performance over 980TI, stock vs stock. Maybe its 5% faster than Titan X @ stock LOL
Wonder what performance will be on max OC vs max OC :sneaky:

No drivers for it, though...and it's not a "ti" model...

 

[parvadomus]

die shrink with clock boost'll do that.

is this card nothing but a tock a la intel?

This is not even a tock, its just Maxwell adjusted for maximum performance/area @ 16nm, given TSMC process characteristics.

Take GTX980, add 512 SP and clock it @ 1800Mhz, change GDDR5 by GDDR5X and bingo :thumbsup:

 

[xthetenth]

Let's all hope those Polaris 10 results are way off, because if GFXBench and AotS leaks reflect what to expect from this particular card (Ellesmere Pro?) I'm not very confident about it's performance.

Those results are bad enough I'm pretty confident they aren't a proper representation.