[DX12] Fable Legends Beta Benchmarks

[iiiankiii]

I think you shouldn't test oc cards vs stock cards in some situation certainly.

The 980ti cards are 670 usually and the fury x is 650 or higher as well. So I really see no issue using an aib 980ti card that's 20 over the msrp. It's what the vast majority of people care about.

Using a 980ti reference card that very very few people are interested in, in a test means nothing to me. Show me the $670 980ti that are vastly superior that's what matters to me. The only reason people are whining is because it makes the 980ti look great.

Is it unfair if people don't use amd cards that are aib models when it makes sense? Sure... But we should tell them to make their tests better. Not to make other people's tests worse because we're mad at some "slight" from years prior....

It's more about consistency than anything really. I want a test that includes the reference models because it's exactly that... a reference. It's the starting point. If you're going to include an overclocked version, include the reference card, too, so the readers can have a better understanding of how each cards stands relative to one another. Plus, we can compare it to other reviewers to get an even better understanding of the cards' performance. Chances are, other reviewers won't have the same aftermarket OC'ed card in their reviews; but, more than likely, will have the reference design. That's it.

 

[Headfoot]

If you don't use the reference model for referencing, what's the point of even having it? This debate has happened over and over and every reviewer is well aware of what they're doing.

There is a clear and obvious way to prevent any appearance of bias, and that is to include both stock reference and a common aftermarket model. Just show both and the problem is solved. The only sticky issue is where there is no true "reference" model, like the 460s back in the day and the Fury today.

You dont even have to add more bars to your graph. Just put two end-cap lines on each bar, the first being stock and the second being representative aftermarket.

 

[thesmokingman]

If you don't use the reference model for referencing, what's the point of even having it? This debate has happened over and over and every reviewer is well aware of what they're doing.

There is a clear and obvious way to prevent any appearance of bias, and that is to include both stock reference and a common aftermarket model. Just show both and the problem is solved. The only sticky issue is where there is no true "reference" model, like the 460s back in the day and the Fury today.

True, but I think an unbiased reviewer can tackle the issue in a fair manner. However, finding an unbiased reviewer is the tricky part.

 

[Piroko]

Update on the pcgameshardware.de bench, they did CPU scaling as well now.
Seems like Nvidia doesn't play well with slow CPUs but also shows much better scaling with faster ones.
Also seems like Fiji is CPU bottlenecked even with a 4.5 GHz 4770k.

 

[railven]

Update on the pcgameshardware.de bench, they did CPU scaling as well now.
Seems like Nvidia doesn't play well with slow CPUs but also shows much better scaling with faster ones.
Also seems like Fiji is CPU bottlenecked even with a 4.5 GHz 4770k.

I wonder if the MOAR COARS strategy is finally paying off. Did anyone use a 6+ core AMD CPU?

 

[Techhog]

Showing a constant 1050mhz clockspeed does not mean it does not throttle/has aggresive power gating at a lower level, that can add latencies inside the chip itself.

Can you explain more in-depth, please?

 

[tential]

It's more about consistency than anything really. I want a test that includes the reference models because it's exactly that... a reference. It's the starting point. If you're going to include an overclocked version, include the reference card, too, so the readers can have a better understanding of how each cards stands relative to one another. Plus, we can compare it to other reviewers to get an even better understanding of the cards' performance. Chances are, other reviewers won't have the same aftermarket OC'ed card in their reviews; but, more than likely, will have the reference design. That's it.

Hey even better my friend. As readers we want the most information possible!

 

[Azix]

Fable Legends bench.
The press may not get the same version.

The press results are similar to your async off results. 70-90fps

your async on result is over 100fps. This could mean nvidia convinced them to turn off async for all gpus in the press benchmark like they tried to do with ashes.

Testing with maxwell 2 would tell us something, if there was a performance issue with async on.

The async compute & DMA is ready and enabled in the AMD/NV drivers - if the hardware support it.

I have tested the benchmark on a Fury X:
1080p async on: 105,8 fps
1080p async off: 87,8 fps

I don't have Maxwell so I can't provide data on it.

pcper

anand

TR

ET

 

[.vodka]

If Fury gains that much, maybe AC on the rest of the lineup would be enough for even the 370 to beat the 960, for the 285 to get uncomfortably close to the 970, for the 290/390 to beat the 980, and the 290x/390x to also get uncomfortably close to the 980ti... as we've seen more or less in the ashes results.

Interesting.

I also find it hard to believe that every publication out there agreed to nV's request of not testing with AC on, if it happened that way.



Zlatan, can you get hold of a maxwell card to do your own testing? Maybe someone can lend you a card.

 

[Azix]

If Fury gains that much, maybe AC on the rest of the lineup would be enough for even the 370 to beat the 960, for the 285 to get uncomfortably close to the 970, for the 290/390 to beat the 980, and the 290x/390x to also get uncomfortably close to the 980ti... as we've seen more or less in the ashes results.

Interesting.

I also find it hard to believe that every publication out there agreed to nV's request of not testing with AC on, if it happened that way.



Zlatan, can you get hold of a maxwell card to do your own testing? Maybe someone can lend you a card.

The press don't have many settings options if at all. It would be lionhead/Microsoft who decided what the advanced settings would be for the press distributed benchmark.

 

[Qwertilot]

True, but I think an unbiased reviewer can tackle the issue in a fair manner. However, finding an unbiased reviewer is the tricky part.

You can try to be fair, but not having a reference card does just make it very hard.

Never mind Fury, take the 3xx stuff. Ok, no reference model is believable there as rebrands, but what do you use?

A 2xx reference model fair when they claim to have tweaked them? Some after market - so very likely factory OC'd by some variable amount - model instead?

Not great. Have to hope AMD will be back using reference models with the die shrunk stuff.

As well as the other stuff, the other thing that reference cards are good for is showing the energy efficiency of the architectures at a suitably chosen clock speed.

A touch ironic perhaps that a 'reference' FuryX - clock speeds/TPD between the nano and X - would probably be a rather nice card once some 3rd party put a proper cooler on it

 

[tential]

You can try to be fair, but not having a reference card does just make it very hard.

Never mind Fury, take the 3xx stuff. Ok, no reference model is believable there as rebrands, but what do you use?

A 2xx reference model fair when they claim to have tweaked them? Some after market - so very likely factory OC'd by some variable amount - model instead?

Not great. Have to hope AMD will be back using reference models with the die shrunk stuff.

As well as the other stuff, the other thing that reference cards are good for is showing the energy efficiency of the architectures at a suitably chosen clock speed.

A touch ironic perhaps that a 'reference' FuryX - clock speeds/TPD between the nano and X - would probably be a rather nice card once some 3rd party put a proper cooler on it

Why would amd use reference models? It's the largest waste of money ever it's mind boggling that they would introduce reference models again after how well the 300 series did..

I like amds current approach. Use reference models on a select number of skus, then let aibs handle the rest.

 

[PhonakV30]

The async compute & DMA is ready and enabled in the AMD/NV drivers - if the hardware support it.

I have tested the benchmark on a Fury X:
1080p async on: 105,8 fps
1080p async off: 87,8 fps

I don't have Maxwell so I can't provide data on it.

So your Bench has option to turn off/On AC ? some says the benchmark doesnt allow configuration? can you provide screenshot of that option ? Thanks.

 

[Qwertilot]

Why would amd use reference models?

Not to sell (Has anyone really done that en mass?). Its something they both seem to have done traditionally though so must think its worth doing.

The reason for us to want them to do it, is a service to the community. It gives people a stable place to benchmark against/compare. Otherwise (as noted above by RS) you really do get a bit (lot!) of mess.

Purely self interest? Not so obvious. It does help make the AIB's look good of course and you want to keep them onside. There is also the point that putting out a reference card lets them show it at the most efficient point of the clock/efficiency curve.

The AIB's often enough crash well over this of course. Imagine no reference 980 or 980ti, and people judging them purely by the most extreme factory overclocked versions? Yes it'd look a chunk faster on the charts but the power draw......

 

[stahlhart]

The async compute & DMA is ready and enabled in the AMD/NV drivers - if the hardware support it.

I have tested the benchmark on a Fury X:
1080p async on: 105,8 fps
1080p async off: 87,8 fps

I don't have Maxwell so I can't provide data on it.

Thread cleaned.

Zlatan: in the future, document metrics such as those quoted with screen shots.

-- stahlhart

 

[N7SpectreElite]

Stahlhart. Thank you for you post.

 

[brandonmatic]

ExtremeTech updated their benchmarks for the Fury X and 980ti. Apparently Windows 10 power management was throttling the 980ti at low resolutions for some reason.

Corrected Benchmarks:

Original Benchmarks:

 

[ShintaiDK]

Interesting with the power management.

 

[RussianSensation]

Interesting with the power management.

This has been a know Windows "issue". By default Windows sets Balanced power management instead of High Performance to prioritize laptops. This setting should never be used for desktops, nevermind for high-end GPU/CPU benchmarking.

Too bad many non-professional PC gamers probably have Windows using the Balanced setting.

For years I tested NV and AMD GPUs and Intel's high CPU over clocks and the conclusion remains the same from Windows 7->8->10: Balanced setting does not allow for maximum CPU/GPU boost and % utilization over the same loads as a High Performance setting does. Even running distributed computing on my i7 laptops wipes out a 10-15% chunk of my CPU performance.

 

[railven]

Well damn, did any site cover all the bases and provide a decent bench?

At this point every other page on this thread is updated benches with new info.

 

[dzoni2k2]

This has been a know Windows "issue". By default Windows sets Balanced power management instead of High Performance to prioritize laptops. This setting should never be used for desktops, nevermind for high-end GPU/CPU benchmarking.

Too bad many non-professional PC gamers probably have Windows using the Balanced setting.

For years I tested NV and AMD GPUs and Intel's high CPU over clocks and the conclusion remains the same from Windows 7->8->10: Balanced setting does not allow for maximum CPU/GPU boost and % utilization over the same loads as a High Performance setting does. Even running distributed computing on my i7 laptops wipes out a 10-15% chunk of my CPU performance.

Never had any issues with balanced settings. My GPU and CPU ramp up as they should and there is 0 performance difference for me in games or benchmarks.

 

[N7SpectreElite]

Windows tweaking class 101... Always high performance setting if you care about performance.

Any gfx benchmarking website that doesn't do that as standard is now on my take with a grain of salt list... Lol.

 

[tweakboy]

Tell me a Fable ......

 

[tweakboy]

When will people stop benchmarking and instead of playing a game ?

 

[Erenhardt]

Where can I download the benchmark?