[DX12] Fable Legends Beta Benchmarks

[PPB]

From GTX980 being 15-20% faster at launch than R9 290X

http://www.anandtech.com/show/8526/nvidia-geforce-gtx-980-review/13

To R9 390X being faster than GTX980 in DX-12 games

Also R9 290 faster than GTX970 and very close to double priced GTX 980.
Anyone bought an R9 290 at $240-250 the last months should be commended

Edit:

Also to add that GTX 960 should start to look way overpriced at DX-12 games against the R9 380.

Wooooowowooooo, my 960 went from 40% slower to 100% slower than 290 in DX12. But but my shiny box says DX12.1!!!! IT MUST BE BETTER :awe:

Jokes aside, it is nice to see how the first truly DX12 only game behaves like. If only there were 290/390 with 20cm/8 inch PCBs, I would jump on one inmidiatedly.

 

[Phynaz]

From GTX980 being 15-20% faster at launch than R9 290X

http://www.anandtech.com/show/8526/nvidia-geforce-gtx-980-review/13

To R9 390X being faster than GTX980 in DX-12 games

Also R9 290 faster than GTX970 and very close to double priced GTX 980.
Anyone bought an R9 290 at $240-250 the last months should be commended

Edit:

Also to add that GTX 960 should start to look way overpriced at DX-12 games against the R9 380.

Nice cherry picking, combining graphs from different reviews.

How about we use the one from Techreport, how is that AMD scaling looking here?

 

[AtenRa]

Nice cherry picking, combining graphs from different reviews.

How about we use the one from Techreport, how is that AMD scaling looking here?

You do realize the graph you posted is for the CPU scaling.

 

[AtenRa]

Wooooowowooooo, my 960 went from 40% slower to 100% slower than 290 in DX12. But but my shiny box says DX12.1!!!! IT MUST BE BETTER :awe:

Really the GTX 960 was overpriced at launch and should come down in price very quickly the next months or else R9 380 will gain momentum.
Same goes for the GTX970 vs R9 390.

 

[TheELF]

From GTX980 being 15-20% faster at launch than R9 290X

http://www.anandtech.com/show/8526/nvidia-geforce-gtx-980-review/13
To R9 390X being faster than GTX980 in DX-12 games

Battlefield 4 has no in game gpu benchmark so they ran the actual game, the fable benchmark is a gpu based benchmark,they even say so themselves,so wait until an actual DX12 game is benchmarked with actual gameplay before drawing any conclusions.

 

[Makaveli]

Bleh so much for AMD crushing Nvidia in DX12. I would think after being proved wrong so many times, people would stop buying into the AMD marketing slides.

The funny thing is everyone that was telling the AMD guys it was just one benchmark you can't get solid information from it yet.

Now another beta bench shows NV in a better light and everyone has forgotten the same "its just one benchmark wait for a final game" because now it aligns with there chosen gpu.

Seems to me like both sides need to stop the nonsense.

 

[railven]

The funny thing is everyone that was telling the AMD guys it was just one benchmark you can't get solid information from it yet.

Now another beta bench shows NV in a better light and everyone has forgotten the same "its just one benchmark wait for a final game" because now it aligns with there chosen gpu.

Seems to me like both sides need to stop the nonsense.

I can dig it.

 

[tential]

From GTX980 being 15-20% faster at launch than R9 290X

http://www.anandtech.com/show/8526/nvidia-geforce-gtx-980-review/13

To R9 390X being faster than GTX980 in DX-12 games

Also R9 290 faster than GTX970 and very close to double priced GTX 980.
Anyone bought an R9 290 at $240-250 the last months should be commended

Edit:

Also to add that GTX 960 should start to look way overpriced at DX-12 games against the R9 380.

Where's my cookie for picking up a r9 290 for $200?

Looks like anandtech needs a wider range of gpus tested as it seems the 290/x benefit more than fury x?

That's all I can see from mobile right now though not sure if it's true.

 

[AtenRa]

Battlefield 4 has no in game gpu benchmark so they ran the actual game, the fable benchmark is a gpu based benchmark,they even say so themselves,so wait until an actual DX12 game is benchmarked with actual gameplay before drawing any conclusions.

There were users telling us in the Ashes Of the Singularity topic that NV got up to 20% with DX-12 in Fable. Now Fable is just a benchmark again because GCN performance is good.

Well, there is a trend developing in the DX-12 games and its only good for the competition and the actual PC gamers

 

[RussianSensation]

Bleh so much for AMD crushing Nvidia in DX12. I would think after being proved wrong so many times, people would stop buying into the AMD marketing slides.

I knew I'd read comments like these despite people not paying attention at all to how much UE4 engine has favored NV's Maxwell. But let's start with the elephant in the room:

1. "The Test
The software provided to us is a prerelease version of Fable Legends, with early drivers, so ultimately the performance at this point is most likely not representative of the game at launch and should improve before release. What we will see here is more of a broad picture painting how different GPUs will scale when DX12 features are thrown into the mix. In fact, AMD sent us a note that there is a new driver available specifically for this benchmark which should improve the scores on the Fury X, although it arrived too late for this pre-release look at Fable Legends"

So to start off that means the AT results online aren't even using the latest driver from AMD that has optimizations for this title.

As others already mentioned before me, UE4 heavily favours Maxwell in DX11. This is a HUGE factor here and it's amusing to see how certain people keep ignoring this when R9 290X > 970, R9 285 > 960 and HD7970 925mhz version > 680.

We do not see GTX970 smashing 290X in UE4 in this benchmark, why is that? DX12 FTW! and this is despite not even using AMD's optimized drivers!

UE4 970 >>>> 290X

UE4 970 > 290X

UE4 970 > 290X

UE4 970 > 290X

We go from NV cards heavily dominating GCN in UE4 games in DX11 to NV losing all of its performance advantage to GCN under DX12 in UE4. The only card that has the lead is 980Ti. R9 290X beats GTX970 by more than 13%! And this is thermal throttling reference 290X too, imagine if R9 390 was in here, it's be a slaughter for the 970. DX12 isn't a game changer for GCN? hilarious.

2. It's pretty obvious that DX12 has resolved AMD's major DX11 draw call bottlenecks as seen in low resolution benchmarks where Fury X and even 290X (!) beat 980Ti.

Can anyone here name any 1 modern AAA title where a 290X or a Fury/Fury X beats 980Ti at low resolutions?

All those people who said DX12 would not be a boon to GCN were dead wrong. It's clearly having a game changer as it has a massive impact on AMD's GCN cards as this article predicted.

Some also claimed that unless DX12 games uses bucketloads of Asynchronous Compute, that GCN would not benefit from DX12 despite many pointing out that DX12 itself removed major bottlenecks AMD due to DX11 drawcall API bottleneck. Once again those who predicted that GCN would see huge performance benefits moving from DX11 -> 12 even without heavy AC use were spot on.

3. I hope people finally start seeing the light and STOP recommending the POS GTX960. R9 290X is smashing the GTX960 by a factor of 2. Seeing this benchmarks, no one should be buying a $200 GTX960 4GB for gaming to keep for 2-3 years when R9 290 still sells for $230 in the US.

Glad to see that my predictions for the last 9-10 months of GTX960 getting destroyed even more with future games by R9 290 is coming true, despite the Greenpeace movement doing everything possible to recommend GTX960 over the far superior AMD solutions.

4. More evidence that DX12 is a huge boost for GCN is R9 285 is beating GTX960 by 24%.

So ya, in a heavily NV-favoured UE4 game engine, AMD's cards are cleaning up Maxwell, other than the amazing GTX980Ti, but that's despite not using AMD's optimized driver for this benchmark and people still are in denial that DX12 is a game changer for GCN. :sneaky:

 

[Makaveli]

Looking at the numbers RS i'm wondering about my own card.

With final game code and better optimized drivers and a overclock wonder if the 7970 can get close 60fps in ultra.

Should be no problem if you do a mix of high and ultra settings this is looking promising for me. Since i'm still on a 1200p screen still have enough gpu power. Will be able to hold out on GCN 1.0 until we start seeing the HMB2 cards where I will probably go 27inch 1440P Freesync screen after the gpu update.

 

[tential]

How does this game look? The main boon for dx12 isn't what gpu vendor is better to me. Does this game look good for the gpu horsepower necessary? Is dx12 helping in that regard?

 

[AtenRa]

Looking at the numbers RS i'm wondering about my own card.

With final game code and better optimized drivers and a overclock wonder if the 7970 can get close 60fps in ultra.

Should be no problem if you do a mix of high and ultra settings this is looking promising for me. Since i'm still on a 1200p screen still have enough gpu power. Will be able to hold out on GCN 1.0 until we start seeing the HMB2 cards where I will probably go 27inch 1440P Freesync screen after the gpu update.

I dont believe the HD7970 will be able to do 60fps on ultra. But as you said a combination of high/ultra settings will get you there.

 

[Headfoot]

Looking at the numbers RS i'm wondering about my own card.

With final game code and better optimized drivers and a overclock wonder if the 7970 can get close 60fps in ultra.

Should be no problem if you do a mix of high and ultra settings this is looking promising for me. Since i'm still on a 1200p screen still have enough gpu power. Will be able to hold out on GCN 1.0 until we start seeing the HMB2 cards where I will probably go 27inch 1440P Freesync screen after the gpu update.

I think there they are testing a stock 925mhz 7970, since you've got your ghz edition underwater I assume you've got a pretty good overclock rolling on it? Final code + final drivers + 1200 mhz core OC might get you pretty close to that 60fps mark. I'd guess between 50-55fps avg

 

[LTC8K6]

So ya, in a heavily NV-favoured UE4 game engine, AMD's cards are cleaning up Maxwell, other than the amazing GTX980Ti, but that's despite not using AMD's optimized driver for this benchmark and people still are in denial that DX12 is a game changer for GCN. :sneaky:

I wish a 980 was in there to help clarify card positions.

Presumably, it would be well above the 970?

And where would full Tonga fall?

And before I forget, with the 285 over the 960:

"GO TONGA!" :biggrin:

 

[thesmokingman]

This bench should probably not be taken as being representative of the game. Both camps are doing respectively well, some ups and downs for both sides. It seems pretty even or close overall. Regardless of which camps you subscribe to, it looks good for everyone involved. Wipes brow...


"The benchmark provided is more of a graphics showpiece than a representation of the gameplay, in order to show off the capabilities of the engine and the DX12 implementation. Unfortunately we didn't get to see any gameplay in this benchmark as a result, which would seem to focus more on combat."

 

[LTC8K6]

This bench should probably not be taken as being representative of the game. Both camps are doing respectively well, some ups and downs for both sides. It seems pretty even or close overall. Regardless of which camps you subscribe to, it looks good for everyone involved. Wipes brow...


"The benchmark provided is more of a graphics showpiece than a representation of the gameplay, in order to show off the capabilities of the engine and the DX12 implementation. Unfortunately we didn't get to see any gameplay in this benchmark as a result, which would seem to focus more on combat."

Yeah I think overall what this says is that DX12 helps AMD, but NV is no slouch at it, either.

 

[xorbe]

If I understand, DX12 helps shrink software driver overhead. So AMD stands to gain more, as NV has been noted for better efficiency previously. Also, it exposes how weak the 960 hardware really is (that 128-bit vram bus same as 750 ...), unlocking the power of the faster cards that were somewhat bottlenecked by the software api. Nothing wrong with the 960 (my htpc choice due to wattage limits) but the price is $50 high imho.

 

[ThatBuzzkiller]

Good, AMD really needs every advantage they can get to compete ...

Fury X scaling in comparison to Hawaii is pretty poor ...

They'll really need to reconsider on lowering the occupancy required to get high utilization on their next micro-architecture ...

 

[thesmokingman]

Btw isn't UE4 a heavily Nvidia biased engine to begin with which would make these results a best case scenario?

 

[biostud]

Why are the top tier cards (Fury X and 980Ti) so close to 390X and 980 in performance?

 

[AtenRa]

Why are the top tier cards (Fury X and 980Ti) so close to 390X and 980 in performance?

Lack of driver and game optimization most probable, Fable is in Beta as of now.

 

[RussianSensation]

TechReport is objective as always / sacrasm. They are using a massively factory pre-overclocked 980Ti in their benchmarks -- "For example, the GeForce GTX 980 Ti we tested is the Asus Strix model that won our recent roundup."

Looking at their roundup, it's actually Asus Strix GTX 980 Ti OC Edition. Looking at reviews online, this card is listed to boost to 1317mzh on paper specs but in reality this SKU boosts to 1380 out of the box.

TPU's review:

"Today, we reviewed the ASUS GTX 980 Ti STRIX, one of the most anticipated cards this season. It comes with a base clock of 1190 MHz, which is already good, but ASUS has also increased the Boost clock range of their card, which makes it the fastest GTX 980 Ti we have tested so far. Compared to the GTX 980 Ti reference, we see an impressive 16% performance improvement at 4K"

I guess it's a normal practice now to test some of the best factory pre-overclocked NV cards and keep AMD cards at stock speeds. No wonder AMD didn't want to send samples to this site as they continue to show total lack of professionalism when testing products fairly. We see 980Ti beating Fury X by 17%, but the Asus Strix 980Ti OC Edition is itself 15-16% faster than a stock 980Ti. Talk about a misleading review.

----

Anyway, ignoring the nearly 1.4Ghz 980Ti in the TR overview:

285 > 960
390 > 970
390X > 980

This is unheard of in an UE4 game when NV cards always dominated AMD under DX11.

Looking into frame times, in every key price segment, AMD beats NV, except for the after-market 980Ti:

We can also see from AT's review and TR's review that Fury/Fury X cards needs a newer driver to optimize their performance of there is an architectural bottleneck in this title.

Barely any performance difference between the 390X/Nano/Fury/Fury X at 4K is also very odd. This game should be retested at a later date with updated AMD drivers to see if this mysterious bottleneck disappears.

In Ashes of Singularity R9 290X/390 beat 970 and we see the same in Fable Legends. AMD is off to a very good start for DX12.

 

[monstercameron]

@rs I don't see how a driver update will do much for supposedly thin driver.

 

[RussianSensation]

@rs I don't see how a driver update will do much for supposedly thin driver.

It's right there in the AT article that AMD has a newer driver with even higher performance. It wasn't used for testing purposes. Even another 3-5% increase in performance will only extend the lead of 285/7970/390 over 960/680/970 cards. The lack of a larger performance increase from Fury/Fury X over 290X/390X is also odd unless there is some bottleneck in the Fiji architecture or possibly a driver issue.

PCgameshardware has Fury X beating 390X by only 7% at 4K.
http://www.pcgameshardware.de/DirectX-12-Software-255525/Specials/Spiele-Benchmark-1172196/

Also, the HD7970 AT used is the 925mhz version and it still managed to beat the GTX680 by almost 13%. I wish they included R9 280X or HD7970Ghz edition.

If AT ends up using this game for testing in 2016, I hope they upgrade their CPU platform too.

Intel Core i7-4960X in 3 modes:

'Core i7' - 6 Cores, 12 Threads at 4.2 GHz
'Core i5' - 4 Cores, 4 Threads at 3.8 GHz
'Core i3' - 2 Cores, 4 Threads at 3.8 GHz

I'd like to see these tests with DDR4 3000mhz on an i7 6700K @ 4.7-4.8Ghz, or perhaps they could consider upgrading to the 6-core BW-E (i7 6820K?) in Q1 2016.

TR lol, they sure do love a lopsided test. Shakes head...

They mysteriously did not test Ashes of Singularity, sighting that the game was still in beta/not ready for actual release and not indicative of DX12 performance, but then they go ahead and use a nearly 1.4Ghz boosting 980Ti in Fable Legends BETA.

Overclock3D's on Extremetech results:

"If one were to just look at these two GPUs (Fury X and 980Ti), you could easily say that both perform very similarly in DirectX 12, but if you look at the more detailed set of benchmarks below you can see that AMD clearly comes out on top in the lower end of the market.

AMD's R9 380 beats Nvidia's GTX 960 by significant margin, giving AMD a massive advantage in the Sub £200 GPU pricepoint.

"Most surprisingly at both 1080p and 4K AMD's R9 390 beats Nvidia's GTX 980, which is a GPU that is over £150 more expensive. The AMD R9 390 is a price competitor to Nvidia's GTX 980, so a R9 390 beating a GTX 980 here is a massive shock."

PCGamesHardware

1080P
280X is 19% faster than GTX770
R9 290 is 13% faster than a GTX970
390X is 9% faster than GTX980 ** Shockingly Asus Strix GTX980 is only 3% faster than the Sapphire Tri-X 290. We are talking about a $450 card barely beating a $230-$250 R9 290 **
980Ti SSC beats Fury X by 10%

The main reason the 980Ti beat the Fury X is because they used an EVGA Superclocked version. Not surprising since that's free 15-18% more performance for the 980Ti. If Fury X could overclock as well as the 980Ti, this would have been a clean sweep for AMD in Fable Legends. Great engineering from NV to manufacture a GM200 chip that overclocks this well.

The theme many of us have been repeating for months continues -- unless one is buying the flagship GTX980Ti, AMD currently has NV beaten in the key $180-500 pricing segments, specifically R9 380 2/4GB > 960 2/4GB, R9 290 >>>> GTX960 4GB, R9 390 > 970, and R9 390X is also a better buy for the price than the 980. Fable Legends benchmarks and Ashes of Singularity only help to underscore this fact, albeit both are in beta form.

Looking forward to more DX12 benchmarks but clearly DX12 is starting to pay off for AMD. It's going to be extremely hard now to objectively recommend GTX960/970/980 over R9 380/290/390/390X AMD cards that have already been winning in price/performance to start with and now in Fable Legends a $280 390 is trading blows with a $450 980.