• 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.

AMD Polaris architecture – GCN 4.0

 
Last edited:
Orcvader said:
Yup, I'm glad they've been very forward on the benchmark results.

Relooking the OP on the Reddit thread, the benchmark seems to be split into 3 different tests, the first to purposely CPU bound, the second an average of both, and the last more GPU heavy.

Which then makes me wonder why the 1080's utilization for the CPU bound test is so... high.
Click to expand...

It is likely the typical story we have seen form Maxwell onwards where nvidia cards are achieving much closer to their (lower) theoretical maximum and are not getting bottle necked by drivers, draw-call and front-end issues.

I hope I'm wrong but that actually paints a poor picture of Polaris because I had hoped those are the things that had been improved on the most:confused: And AotS should be the one game where AMD maximizes utilization through copious async shaders and close partnership with the developers, Oxide. So why can nvidia achieve excellent utilization in a scenario that the 480 is getting heavily CPU-bound on?
 
AoTS isn't a game I would recommend and have only played the actual game for probably 2 hours max but it has caused so many debates and raised so many questions, it should probably be ignored. Besides, looking at the bench thread, it looks like I am not alone and only owned by a couple of people and a dodgy jack sparrow version thrown in for good measure.

Nasha is correct and we want the games like BF, Tomb Raider, The Division, GTA V to be the type that has DX12. Sorry to those guys who like RTS games but not my cuppa:D
 
D.P. said:
It is likely the typical story we have seen form Maxwell onwards where nvidia cards are achieving much closer to their (lower) theoretical maximum and are not getting bottle necked by drivers, draw-call and front-end issues.

I hope I'm wrong but that actually paints a poor picture of Polaris because I had hoped those are the things that had been improved on the most:confused: And AotS should be the one game where AMD maximizes utilization through copious async shaders and close partnership with the developers, Oxide. So why can nvidia achieve excellent utilization in a scenario that the 480 is getting heavily CPU-bound on?
Click to expand...

Ideally surely if you're running into a CPU bottleneck, you want the GPU usage to be pretty low as it will be waiting for the CPU to catch up? Otherwise it just means your GPU is not as CPU bottlenecked as it should be :p.

We really need more DX12 benchmarks!!!
 
Orcvader said:
Ideally surely if you're running into a CPU bottleneck, you want the GPU usage to be pretty low as it will be waiting for the CPU to catch up? Otherwise it just means your GPU is not as CPU bottlenecked as it should be :p.

We really need more DX12 benchmarks!!!
Click to expand...

If you hit the CPU bottleneck yes, but these numbers are indicating the Pascal drivers and architecture don't suffer from the CPU bottleneck which is something we have seen time and again when looking at AMD and Nvidia GPUs.
 
Wish more devs put benchmarks into their games, like The Division etc, would love to see actual performance in games that people actually play, last game i owned with an ingame benchmark was "Thief" and that game sucked, so regardless of what FPS etc it achieved i wouldnt play it anyhow.
 
D.P. said:
It is likely the typical story we have seen form Maxwell onwards where nvidia cards are achieving much closer to their (lower) theoretical maximum and are not getting bottle necked by drivers, draw-call and front-end issues.

I hope I'm wrong but that actually paints a poor picture of Polaris because I had hoped those are the things that had been improved on the most:confused: And AotS should be the one game where AMD maximizes utilization through copious async shaders and close partnership with the developers, Oxide. So why can nvidia achieve excellent utilization in a scenario that the 480 is getting heavily CPU-bound on?
Click to expand...

It doesn't paint a poor picture of Polaris because IF your hypothesis is true it shows vast potential for improvement in the near future with driver updates.

Though I think your hypothesis probably isn't the explanation that said.
 
Gregster said:
AoTS isn't a game I would recommend and have only played the actual game for probably 2 hours max but it has caused so many debates and raised so many questions, it should probably be ignored. Besides, looking at the bench thread, it looks like I am not alone and only owned by a couple of people and a dodgy jack sparrow version thrown in for good measure.

Nasha is correct and we want the games like BF, Tomb Raider, The Division, GTA V to be the type that has DX12. Sorry to those guys who like RTS games but not my cuppa:D
Click to expand...

AotS is a marketing showpiece for AMD designed to cause controversy.

What I also don't get is the graphics look rubbish compared to soe thing like to braided or the latest BF games. And it is supposed to be a strategy game, and the least important thing about strategy games is graphics, so really weird to use as a showpiece,



the fable legends benchmark was nice because it broke down the performance by the different rendering workloads so you could see the comparative strengths and weaknesses of the different cards.
 
h4rm0ny said:
It doesn't paint a poor picture of Polaris because IF your hypothesis is true it shows vast potential for improvement in the near future with driver updates.

Though I think your hypothesis probably isn't the explanation that said.
Click to expand...

That same potentially that has been in Hawaii and Fiji but is never actually there?

But you are probably right, the 51% CPU-bound number is probably meaningless because in the other 49% you don't know what the actually GoU utilization was, maybe it was 99%utilized.
 
The reason why AoTS is used as a bench is because its the only DX12 game that actually can bottleneck the CPU and get a good utilisation on the GPU. What do you suppose they use? Tomb raider? Remember what low level API was supposed to do? Alleviate that CPU overhead which is why they use AoTS

I see it like making a gpu which is excellent at tessellation then not bothering to put it to use at driver level. Gotta get DX12 put to use!
 
r7slayer said:
The reason why AoTS is used as a bench is because its the only DX12 game that actually can bottleneck the CPU and get a good utilisation on the GPU. What do you suppose they use? Tomb raider? Remember what low level API was supposed to do? Alleviate that CPU overhead which is why they use AoTS

I see it like making a gpu which is excellent at tessellation then not bothering to put it to use at driver level. Gotta get DX12 put to use!
Click to expand...

exactly and amd does it best with true async compute
game developers says so
 
D.P. said:
That same potentially that has been in Hawaii and Fiji but is never actually there?

But you are probably right, the 51% CPU-bound number is probably meaningless because in the other 49% you don't know what the actually GoU utilization was, maybe it was 99%utilized.
Click to expand...

.....
 
r7slayer said:
The reason why AoTS is used as a bench is because its the only DX12 game that actually can bottleneck the CPU and get a good utilisation on the GPU. What do you suppose they use? Tomb raider? Remember what low level API was supposed to do? Alleviate that CPU overhead which is why they use AoTS

I see it like making a gpu which is excellent at tessellation then not bothering to put it to use at driver level. Gotta get DX12 put to use!
Click to expand...

I put it into the same bracket as Firestrike and Heaven benches. Ideal for gauging performance but ultimately, we all want to see how the cards fair in the actual games we play. In time, there will be several DX12 games to test but for now, Tomb Raider is a great choice to use as it is an excellent game with sweet visuals and owned by many.
 
SiDeards73 said:
Warhammer has a built in DX12 Bench now btw ;)
Click to expand...

I guess someone might want to start a new thread if discussing it further?

http://www.guru3d.com/articles-page...-graphics-performance-benchmark-review,6.html

index.php
 
You must log in or register to reply here.
Back
Top Bottom