What engine does dirt 4 use and how much do we know about primitive shaders in current games?
Is this the type of performance we could see if everything lines up right on a Vega card?
Not cherry picking just wondering why vega could have such a massive win in this title vs such massive losses in UE4 games.
>>61910081
who cares it's not as good as GRID
>>61910109
i am not concerned with the game but rather the game engine which is the EFO engine 4.0 also used in F1 2017
>>61910142
And here is gets destroyed
Thing is both engines are very well designed engines. It seems like something might be working in dirt 4 where as its not working in UE4.
We know primitive shaders and tile-based rasterization was not working at all on the FE card and some reviewers were sent custom drivers that supposedly has some of that working.
Maybe some of the game engines dont have the proper updates yet? Member when vulcan came out on doom and it was a 60% gain in performance on AMD cards. We could see something similar going on here
Oh and the Dirt 4 results are not isolated to that one reviewer.
All reviewers have seen nearly double 1060 performance so much so it runs into CPU limitations
>>61910197
PUBG is pretty unoptimized in general.
>>61910081
This game just seems to run better in general on AMD GPUs https://www.youtube.com/watch?v=jdBpSHpxXWM
>>61910081
the engine and prior engine before this game has always ran better on amd gpus, one of the features that cripples nvidia performance is global illuminations, gcn has always been better at this because of their compute prowess
>>61910236
>>61910081
Where did they get a 8gb 1080ti?
>>61910197
Fuck I was gonna get Vega specifically for pubg
Welp off to get a 1080ti
>>61910335
PUBG runs like ass on just about any system
>>61910416
Blame the retarded devs who prioritize fucking cosmetics over fixing the damn game
>>61910431
Don't they have for-pay loot crates or some shit these days?
>>61910452
Sure Doo and retards fall hook line and sinker for it.
Not to mention they ban people for the game being a buggy POS
>>61910311
Actually though.
>>61910465
>>61910452
>>61910431
>>61910416
AMD defense force. These guys will pretend to play ashes of the singularity just because it benches good with amd.
>>61910489
>defending PUBG
>>61910489
Actually I have a 390x but pubg is optimization like shite.
>>61910197
anandtechs review shows the vega 56 behind in warhammer dx12, wonder whats going on.
>>61910500
>badmouths any game/software that amd doesn't come out ahead on.
It just gets old at this point.
>>61910081
Wow 1080p? WOW BEST CARD EVER!
>>61910540
It was worse before they updated the engine. I remember the poojeets shitting on it until the new engine dropped. Then they were all about pubg untill the vega benches.
>>61910550
The game runs like garbage no matter what hardware you use, on top of looking like straight shit.
Not even an AMD fanboy, but using software that's literally unfinished to bench things is moronic.
>>61910564
you know AMD usually wins/ties out in higher resolutions right? unless you have a 1080 TI (you don't) you should refrain from posting your asspained opinions.
it's possible these games are using nvidia gameworks. Given it's track record for being a complete piece of shit for AMD cards to deal with maybe it's just that VEGA drivers can't deal with nvidia gameworks stuff very well yet?
>>61910431
>Blame the devs
>Who let their artists do art
WHAT
>>61910564
AMD does even better at higher resolutions you retard
>1080ti "8GB". Cool memes bro
>>61910608
but it's AMD sponsored game
>>61910564
not even benchmark needs to be 4k 177Hz you fucking autist, most people will be playing at 1080p anyway.
>>61910648
>he thought he would get 11 gyggabytes ram from "3.5" nvidia
lol
>>61910602
Not him but okay. lmao
Tell me why no AMD card is able to handle a 1440p ultrawide let alone touch a 4K monitor.
>>61910694
He thought Vega wasn't an abortion of a gpu. Topest of keks!
>>61910719
It came out so it's not as abortion.
>>61910663
at 1080p you are not testing the card but the cpu....
>>61910081
AMDrones waited for this LMAO
>>61910743
It came out stillborn... It's already dead.
B-b-but muh HBM2 memory .. i-its gud!
>>61910672
>moving goalposts
your argument went from resolutions to vega 56 performing badly in two titles, do you have an actual argument or are you going to keep asspain posting?
>>61910702
depending on the title the fury x held its own pretty well in 1440p and more so 4k, the difference in performance between the 980 TI vs fury x was very low compared to the difference at 1080p.
of course the 1080 TI doesn't have any competition right now but i'd rather have a vega 64 with a freesync monitor over a 1080 TI with no gsync monitor.
>>61910431
There's different teams for different parts of the game. Making cosmetics isn't slowing down updates just a way to keep the lights on while their working.
>>61910795
The 2 other images are not in 1080p. The one it did good in was 1080p. Can't you read the images?
>>61910197
is dirt 4 also an gaming evolved title? dirt 3 was, that I know for sure, but I never played or even watched let's play series on the new one.
maybe they had dibs on some proto vega features during dev cycles.
>>61910812
of course i can but if you think that the factor in worse performance is resolution you're fucking retarded. but of course you can prove it's because of resolution and not the game titles, right? (no)
>>61910702
not him, but vega 64 seems to handle 2k rather well. It depends on what your standards are. Even the 1080TI struggles to hit 40fps in alot of games at 4k.
>>61910489
fuck retarted shill. You already had metro to skew the averages in your favor for too long now. How does that outlier feel now when it's against your "beliefs".
>>61910812
>Comparing across games
>Comparing across different engines entirely
Are you literally a mental midget? Do you even understand what you're arguing?
>>61910836
https://www.techspot.com/review/1468-amd-radeon-rx-vega-56/page4.html
>>61910810
Bullshit, it means corporate made a decision to hire too many artists and too few programmers. Obviously you can't assign artists to engine optimization, but the got where they are by hiring too many of them.
>>61910836
https://www.techspot.com/review/1468-amd-radeon-rx-vega-56/page4.html
>>61910878
Your image shows Vega outperforms a 1080 and nearly ties a 1080Ti at 1440p.
THE WHOLE POINT OF THIS FUCKING THREAD IS THAT VEGA 56, BY EVERYTHING WE KNOW, SHOULD NEVER BE BEATING A 1080. GET THIS THROUGH YOUR SKULL.
The question is "what's different about DiRT?" NOT ABOUT "hurr my cerd is ferster xD"
>>61910081
Parallelism. Games that takes advantage of the 4k ALU instead of instructions shortcuts will have better performance in vega. GPU Open also allow better access to the metal, if developers start working their ass of to make proper use of hardware we will see Vega getting nice results.
Also, Drivers for GPU 102 in gaming are a mess. They should be at least 30% better than a 1080 but they almost never reach it.
>>61910778
>I don't need new technologies!
>Please sell the same shit I already bought a year ago!
>>61910908
My point was 1080p TEST THE FUCKING CPU NOT GPU...
How come noone is mentioning the ENTIRE vega lineup is literally just furyx/fury but overclocked?
>>61910197
PUBG runs like shit in any GPU, less shitty in Nvidias.
>>61910889
so the 1080 barely wins by 3 fps over the vega FIFTY SIX, the mid end card top fucking kek, in 4k and you consider this proof that amd can't hold its own in 4k?
>>61910489
Check the configuration and look at that sweet performance on PUBG at 1080p
>>61910908
OP here
this is what I am asking
What did dirt do(engine side) that the other 25 games did not do.
HardwareUnboxed said he ran the test again and checked with screen caps to see if something was not rendering in the game and could not see anything wrong.
The card is all new design so maybe parts of the pipeline are not working on most current games?
>>61910950
Holy fuck you shills, I *NEVER* said it was. bad card, I was making fun of using 1080p for tests..
Go back and quote where I said the RX Vega 56 was a bad card.
>>61910908
The whole point is a 1080ti is 11GB not 8GB .. find a REAL test that knows REAL hardware specs!
>>61910930
Because literally no one here knows that Vega was in development years ago, maybe before polaris so they could do nothing except pushing it out and fix stuff on Navi.
For more info on why 1080p is bad for *GPU* testing https://www.techspot.com/news/68407-tackling-subject-gpu-bottlenecking-cpu-gaming-benchmarks-using.html
>>61910972
>>61910929
1080p is not a cpu test unless the cpu is a bottleneck
If the 1070 1080 1080 ti and Vega56 all scored the same its clearly a CPU limit but if scores are higher on certain GPU the GPU is still the limit.
Based on the 129 min FPS compared to the 159 average it appears the vega card is hitting the CPU limit and could run much higher average frames if the CPU could keep up. As you see most of the GPU's 1% lows are about 60% of the average frame rate number but the vega's is much closer to the average number. Even at 1440 the 56 is 12% faster than a 1080. Very weird.
>>61910998
well you sure as hell were implying that 1080p test doesn't matter. maybe you didn't know that AMD is notorious for losing benchmarks in 1080p while catching up in performance at 1440p/4k.
in other news
>vega 56 is literally neck and neck vs the 1070
nice.
>>61911023
1080p is not bad for benching GPUs.. its simply that most GPUs cant max 4k or even 1440p.
CPU bottlesnecks dont change much with resolution. I would test 720p but still at ultra settings for a CPU test. Some graphics settings like ambient occlusion are heavy CPU bound.
Also testing very high res or high textures can show other GPU bound issues like running out of vram or bandwidth issues in the pipeline ect
1080p is fine for testing if you still see scaling between the GPUs
IF all the GPUs benched show the same frame rate its clearly a CPU limit. This is why you never see ARMA 3 benched for GPUs because its always CPU limited
>>61911023
wide variance on the same exact system apart from the gpus themselves.
I think you're reading too much from that article or you don't get statistics very good.
>>61910972
What performance are you getting with the current update?
>>61910839
Bullshit, the 1080ti can be made to run a smooth 60 fps in almond every game on the market on either high settings or very high with variable resolution scaling. I can't think of a single one; even Dishonored 2 makes the cut after I overclock and reduce AA
>>61911207
makes you almonds activated
>>61911207
not really on 4k
some games sure
>>61910550
You fucking mongoloid , PUBG is crap in general , 80 fucking fps with a 1080ti it s crap for a shitty game like pubg
>>61910081
>>61910236
hardware unboxed joined techspot not long ago. those 2 are the same chart.
>>61910489
>Implying AotS is a bad game
Its not a bad game, but there's no way you play Metro any more than you do AotS... and yet that has been used as a benchmark since it first came out.
>>61910985
same way what doom does
they have their own culling,prim shaders and scheduling for both sides not relying on drivers
so if ALL developers be like Id we'd have 200 fps games everywhere
expect same results from new wolfenstein and farcry5
that fp16 25 tflops going to boost fps by a huge margin
funny thing it's really good arch for future pc gaming if developers adopt fp16, which is open standard
but just you wait when nvidia will come out of gimpwoods and push their own proprietary shit on us, they've been to silent recently about that
>>61910081
All the Dirt and GIRD games are make by Codemasters who optimise their games for AMD. It's not different from Doom and Vulkan.
>pic is a screenshot from Dirt 3 intro
>>61913368
>it's okay when nvidia does it
oh wait it's not okay, because doom is also best performing nvidia game by accident
while everything nvidia sponsored in the past barely works on nvidia, not to mention amd
>>61910810
>vaulting shown at E3, many months ago
>vaulting still not available despite the EA title, meant for testing and funding a game
>MTX system implemented so devs can beat the cash cow
>game still poorly optimized
>still infinite issues
>implement a forced FPS game rule
>change a few server .ini's
>BIG UPDATE TOOLS
>>61913426
>implement a forced FPS game rule
that's the best thing that happened though
also >>/v/
>>61910081
That game loves lotsa ALUs.
Vega is still heavily bottlenecked somewhere.
>>61910197
>PUBG is made with Unreal Engine 4
>Unreal Engine is developed by Epic Games
>Epic Games and Nvidia are partners for decades
>Unreal Engine has been optimized only for nVidia's cards and has been that way for decades
>Games come out with Unreal Engine
>People wonder why AMD sucks so bad
Look at all the Unreal Engine 4 games and check their bench with AMD/Nvidia. Its an obvious bias.
>>61913521
There's also the whole fiasco about async compute being enabled on console Unreal Engine but specifically disabled for desktop variant (due to nvidia).
>>61913521
sweeney on the stage
expect good things soon
>>61911252
phone poster should die
>>61913550
I'm actually waiting to see how Nano fares. I hope Vega Nano performs between Vega 56 and Vega 64. The previous R9 Nano performed only around a 3584 core Fury. The R9 Nano had 175w TDP. Assuming they are using the same cooling solution on Vega Nano, there will be an extra 25watt headroom for overclocking.
>>61910431
>>61910335
>supporting early access steam games
You are what's wrong with PC Gaming
>>61910929
So he used 7 different CPUs?
>>61910109
Gay related immunideficiency disorder?
>>61910416
runs fine for me with a haswell i5 and a 1070. i use the reshader mod + smaa injection and it runs and looks really good
>>61910081
You know what's cherrypicking? The fact that nobody bothered comparing Vega to the 980 Ti, which would be even more humiliating considering I bet it'd be 95% as fast while also being two years old.
>>61910944
>1080 outperforms 1080ti
Wut?
>>61910081
>game optimized for consoles running amd hardware
>runs good on amd hardware
damn op, this is insane
>>61916483
the game came out in march and this bench also came out at that time. it's been like 5 months since then and performance and general optimization has become a lot better now.
>>61910839
>2k
>>61913368
cool, can't wait for the AMD optimised and FP16 supporting Far Cry 5. The 1080Ti gonna be crushed by a $399 card
>>61913649
>The R9 Nano had 175w TDP. Assuming they are using the same cooling solution on Vega Nano, there will be an extra 25watt headroom for overclocking.
The full-cover water blocks for it were cute as fuck too.
Could make for aesthetic as fuck ITX builds or for throwing a few of them into a kikeripper build as a budget workstation.
There's really no such thing as an AMD optimized game; they don't use any proprietary rendering methods that Nvidia cards don't also do. The difference is that and cards implement some of these universal features better like async compute and they have better vulkan/dx12 drivers. Nvidia has better dx11 drivers, but by like double the margin over AMD than the difference between their dx12/vulkan drivers. An Nvidia optimized game has either physx or game works, an "amd optimized game" is just a regularly developed title that uses all available modern gpu rendering technology without forgoing features solely to make it look like Nvidia isn't behind in many ways
T. 1080ti owner