AMD’s demo showed two rx 480’s in Ashes of the Singularity gettings 62.5 fps, with the performance scaling to 183% the performance of a single rx 480 (source, AMD’s spokesperson reddit.com/r/Amd/comments/4m692q/concerning_the_aots_image_quality_controversy/ ). The formula to find the fps of a single card is 1x (the performance of the first card) + .83x (the average % utilization of the second card) = 62.5 fps, and solving that gets x = ~34.2 fps.
So the often repeated claim that the rx 480 performs as well as an r9 390 to a gtx 980 is true, but ONLY for dx 12. As you can see from pic related, the r9 390x only gets 57% of its dx12 performance in dx11 at 1080p high settings. IF (and that’s a big if) the rx 480 scales the same way, that means that it would get 19.6 fps in the Ashes of the Singularity benchmark AMD showed if it was in dx11. Because the gtx 980 gets around 40 fps in the AotS benchmark in dx11, that means that for the vast majority of games, the rx 480 would be around HALF as fast as a gtx 980.
btw if you're confused by the apparent discrepancies between my image and my post, high settings are not crazy settings. It goes high -> extreme -> crazy.
David Price
Thats still a lot of assumptions given we know GCN 4 is different to GCN 2 in many ways (the extent of these differences is unknown).
Cameron Walker
Too bad no other benchmarks yet.
I don't even care about this game, I just want to know if polaris will be a good mid range so I can justify not buying nvidia stuff.
Time will tell.
Carson Young
It will probably be a good mid range card and have a good price/performance ratio. I just think the hype that it's going to be a gtx 980 for $199 is off the mark by a lot.
Robert Mitchell
Oh and by the way, using the benchmark search tool on the AotS website doesn't differentiate between the r9 390 and 390x I think. It just says "r9 390 series" with no option to select which one, the x or the non x card. So I could have used a 390x instead of a 390 in my OP.
Leo Allen
weren't there benchmarks showing the 480 getting more than the 390x?
Asher Ross
It's about on part with the Fury.
Ryan Price
Apparently everybody "knows" how fast the 480 is but oddly nobody can seem to agree exactly what fast is.
It's ""fixed"" now, few weeks after the first 1080 benchmarks.
Dominic Diaz
Last addition to my OP I promise.
Both of these are at crazy settings, 1080p, in AotS.
dx11 960 - 18 fps 980 - 36 fps 1080 - 60 fps
dx 12 960 - 18 fps 980 - 40 fps 1080 - 65 fps
So the 960 gets 50% of the 980's performance in dx11 and 45% the 980's performance in dx12 in the AotS benchmark. IF (and that's another big if), the gtx 1060 has the same ratio of performance to the 1080 that the 960 does to the 980, then the 1060 will get 30 fps in dx11 and 29.2 fps in dx12. I think that's a reasonable way to estimate the 1060's performance because the 760 also gets around 50% of the performance of the 780, so this is a trend with Nvidia. So *if* I am close to the actual numbers with my ballpark estimates here, the gtx 1060 will perform slightly worse than the rx 480 in dx12, but will outperform it in dx11 by quite a large margin. The 960 launched at $199, and if we add $30 to roughly accommodate the price increase that the 1070 and 1080 got over the launch price of the 970 and 980, then the 1060 will be ~$230 (which is also the price of the 8gb rx 480) and will have a very similar price/performance ratio as the rx 480, except it'll be a much more well rounded card for dx11.
I honestly think this is why AMD set their NDA to lift on launch day while Nvidia lifted the NDA on the 1080 and 1080 well before their launch, Polaris probably just isn't what they were hoping it would be. I'm hoping that's not the case and it's like the Doom NDA, everyone was worried they were hiding something by making the NDA for reviews lift on launch day but the game turned out great. Either way, AMD needs get better marketing ASAP.
Henry Clark
You know, on the one hand they keep showing us benchmark with astonishing results, but the other those are all games optimized for amd.
I want benchmarks of other games, and while I dont think itd perform less than the 390 or 290, there is no proof at the moment to back that statement.
And I would really prefer to not buy nvidia's stuff, knowing their ethics
>This video contains content from PONYCANYON. It is not available in your country. REEEEEEEEEEEEEE. I live in the US, this never happens to us!
Matthew Phillips
>provide a hack driver for press to verify their claim >hack driver is mainly a performance driver at the cost of quality >press driver is used in benchmark and is seen as the true performance
>weeks later >get called out for "bug" >fix it with new driver >performance drops >no one cares about drop in performance
Jack Morris
Business as usual goy- I mean guy.
Christian Wood
Oi vey delete this.
The only reason nvidia can get away with this shit so much is bc consumers keep buying them.
The only way to unfuck this is truly for amd to step up.
Landon Gutierrez
>often repeated claim that the rx 480 performs as well as an r9 390 to a gtx 980 is true, but ONLY for dx 12.
The "claim" is based off common sense and comparing the Polaris Ashes demo bench to existing 390x Ashes DX12 benches you shit stain. The comparison to a 980 isn't based on worthless Aots dx11 benches, but on how the 390x compares to the 980 in general.
>As you can see from pic related 5 seconds in Aots benchmark tool shows your turbocuck image is pure fantasy
>vast majority of games even disregarding your bait posting, AtoS dx11 on AMD doesn't provide a performance comparison for the vast majority, nor can 390x performance tell you anything about tell you anything about Polaris.
Why speculate when you have the card itself benchmarked ? It's GTX 980 range performance on AotS.
Luke Sanchez
In dx12. People want their GPUs to be able to play the games they already own and older games they'll pick up in steam sales in the future. dx11 is still very relevant.
Oh please. AMD fanboys have been touting dx12 as the nest big thing because it gives AMD a bigger performance boost than Nvidia, which is legitimately because Nvidia fucked up on dx12. Because dx12 improves the performance of any card that supports it 100%, the dx11 performance of those cards will be behind their dx12 performance.
Chase Perry
So I guess I shouldn't sell my r9 390 for the 480 yet...brb taking it off of ebay.
Christian Ward
i havent been entering these garbage threads but i must know; do people really think $700 card is going to be beaten by a sub $500 when actual game benchmarks are released?
Nolan Evans
It can't hurt. The 480 probably won't be that much faster than the 390, the main advantage is power consumption/heat.
Tyler Thompson
>the dx11 performance of those cards will be behind their dx12 performance. and? DX11 performance on AMD isn't behind Nvidia in general which the OP is trying to claim using AotS to make Polaris look weak.
>Why speculate when you have the card itself benchmarked ? >It's GTX 980 range performance on AotS. You obviously didn't even check because 3900 score is trash tier.
Jacob Thompson
R9 390 is around similar performance level as 980 in DX12.
In DX11 titles, 390 is 1 tier lower in many DX11 games. With release driver, I suspect its going to be 970-390 in DX11 games. With mature drivers, it might be bit higher, so in between 390 and 980 in DX11.
Ofcourse this is assuming the 480 has no DX11 hardware improvement at all. However there are reports suggesting some good DX11 improvements on AMD Polaris. Reduction in driver overheads, hardware improvements to hand dx11 games, etc. This is purely speculation.
>The 480 probably won't be that much faster than the 390 but will it be faster at all? I'm so confused
Ian Scott
If one card has better dx12 performance than dx11 performance, and the dx12 performance is equivalent to a card that has similar performance between it's dx12 and dx11, then the first card has to have worse dx11 performance than the second card. I used the fps, not the games score because I don't quite understand how the scoring works and fps is a more practical standard than semi synthetic benchmarks (if that's even accurate to say...).
Evan Sanchez
We just don't know. All we have is rough ballpark estimates and speculation.
Easton Jones
What happened to that 57% crossfire dx11 performance? 19.5 for 1 card? Yeah. 57% of 19.5 is around 11 fps. So 2xRX480's = 30.6 fps in dx11. Rip op's maths.
Nicholas White
Why don't we wait for more benchmarks before proclaiming how good a card is? I know it's fun to speculate but all of these wild claims with little proof from AMDummies and the massive damage control and mental gymnastics from Nvidiots is stupid.
Jace Harris
I'll admit I probably used a game with a bigger than normal difference between dx11 and dx12 for the AMD cards, I saw the benchmark used am i7 6700k so I wanted to use that one to be consistent because the gtx 980 and r9 390 AotS benchmarks I used also used that processor (just to avoid the processor being the bottleneck). Obviously I couldn't find benchmarks of the i7 6700k with the other cards because nobody runs an i7 with a gtx 960 and uploads the benchmarks to AotS.
Elijah Brown
They showed it running Hitman at 1440p 60fps which is faster than 390.
Austin Lee
>Hitman Is that DX12? Well I'll go for it, at least I'll have a cooler card with full DX12 if I can even sell my 390
Nicholas Watson
Kek I'm laughing hard while reading all the mindfuck of people trying to guess what the actual performance of polaris could be, because AMD accidentally released a benchmark showing how a single rx480 performs in ashes right after the presentation, but no one knows
From this benchmark we can deduce that it will surely perform at 980/fury levels, even in dx11 games.
Owen Torres
>EDIT: Dang it. Polaris 10 have "terrain shading samples: 8 Million" while the rest have 16 Million.
Grayson Miller
What does the ti branding mean on Nvidia cards? I thought it was reserved for the x80 cards but the rumors/leaks for the 1060 mostly call it the 1060ti.
Chase Gonzalez
I think it stands for titan and is a more powerful version
Andrew King
Apparently it stands for titanium. IDK what the branding signifies about the cards though.
Michael Lee
It means they're more powerful than the normal edition, it's old ass branding that comes from the GeForce 4 days, like ATi had the XT and PRO branding
Lucas Evans
I don't think it will matter that much, in the worst case scenario it would perform same as a 390x
Justin Morris
Sup Forums IS NOT YOUR PERSONAL REVIEW SITE PLEASE FUCK OFF!!! Sup Forums IS NOT YOUR PERSONAL REVIEW SITE PLEASE FUCK OFF!!! Sup Forums IS NOT YOUR PERSONAL REVIEW SITE PLEASE FUCK OFF!!! Sup Forums IS NOT YOUR PERSONAL REVIEW SITE PLEASE FUCK OFF!!! Sup Forums IS NOT YOUR PERSONAL REVIEW SITE PLEASE FUCK OFF!!! Sup Forums IS NOT YOUR PERSONAL REVIEW SITE PLEASE FUCK OFF!!! Sup Forums IS NOT YOUR PERSONAL REVIEW SITE PLEASE FUCK OFF!!!
William Peterson
I compared DX11 performance of RX 480 to 980 in AOTS with the same setup and game version. Literally same performance.
How do we know which actual card (470, 480) this 67EF:C3 shit is?
Cameron Williams
Don't forget XL, XXL, XTX, SE, LE, PE, GT, GTO, and GHZ
Nicholas Jackson
this is the 470, the 480 is the C7
Dominic Lee
Dunno. I just know from other leaks that the letters and numbers that look like that are Polaris. I'm going to take that absurdly low score with a grain of salt though, I'll blame it on an experimental driver or something.
Matthew Foster
C3 is second cut Polaris 10, C7 is first cut, iirc C9 or C10 was the full with 2560 shaders.
Wyatt Young
Have you tried ignoring what you don't like?
Aiden Fisher
Why do people give a shit about ashes of singularity particularly??
I've literally never seen this game played/spoke about/displayed before this. I saw it on steam once for a second but never looked past that.
Why don't people benchmark on things people like, play? And are used to or know how it actually performs.
It seems this game is more of a fucking cherrypicked case just for benchmarks rather than being an actual game anyone would give a shit about.
It's not even on the top 100 steam charts (2k players is the lowest..)
Xavier Clark
>Why do people give a shit about ashes of singularity particularly?? It's one of the only games that was built for dx12 from the ground up. It also supports multiple GPUs through dx12, so you can have two AMD cards and use them both without crossfire and stuff. That's why it's being used, it's like the only game like that so far.
Elijah Peterson
The test system used by AMD had an i7 5930K. Thats 3.5ghz stock.
Not directly comparable to a higher clocked chip.
Christopher Morris
It's the best Total Annihilation type game since Supreme Commander: FA; and it's made on an engine that can easily handle quite literally thousands of units at a time with damn good graphics and physics.
Just because faggots are missing out doesn't mean it's bad.
Jayden Moore
You said the magic words "supreme commander". Okay, now I have to check it out.
you shut your ignorant cock-slobbering whore yap, and you never open it again
Logan Ross
Guys guys. Here is comparison of RX 480 in DX11 to nvidia cards. >RX 480 (1440p crazy) = 2500 score >GTX 970 (1440p crazy) = 2600 score >GTX 980Ti (1440p crazy) = 4000
Jeremiah Gonzalez
that's really bad, I don't get it
Jason Hall
Because its a marvel of engineering. First dx12 game. First async game. Scales with every cpu core. Has multicore AI multigpu tech meaning you can pair an amd and a nvidia gpu
64 bit.
Daniel Sanders
Thats 390 / 980 level.
Joshua Lewis
I think the more important question is, what are developers going to do with all this power now? Even conservative estimate for AMD and Nvidia's low end put them at 970 tier (without the 3.5gb) minimum. Are they going to start making games more taxing so that low end cards have to play games on medium/low settings again?
Jace Nelson
The CPU is a 3.3ghz Sandy Bridge Xeon. Everyone needs to stop trying to directly compare results when the test systems are vastly different. That isn't how benching works.
Liam Long
They'll still make games to run on shitty 750ti.
Adam Young
huh, i looked for a 390 on the benchmarks and it was higher
Jonathan Kelly
>I live in the US, this never happens to us hahaaha get fucking rekt!
William Taylor
BEEP AUTISM ALERT
Mason Peterson
wont know till benches, at least in synthetics, it beats out a fury non x, but in gameplay we have benches that beg the question, why the fuck you do that?
so many unknowns...
but not a lot of if onlys.
Nathan James
benchmarks in a game place it in a different spot then a synthetic benchmark, so we have a range where it can fall
Cooper Parker
I have an i5 4570 and a 1080p 60hz monitor. Will my cpu bottleneck that card if I choose to upgrade?
Ayden Robinson
depends on the application...
Adam Murphy
Let's say, shooter games, Witcher 3, Far Cry 4, Quantum Break. I don't usually play RPGs, MOBAs or strategy games.
Joshua Richardson
Just a reminder that 5 years ago, people called dx11 a meme because "all games use dx9 or dx10; dx11 performance doesn't matter".
There were also far fewer benefits to dx11 than there are to dx12.
In both this case and the dx9/10 vs 11 case, the cynics aren't wrong. Presently most existing games and engines don't use the new standard. To that I ask how IDE and EISA are doing.
Andrew Turner
Fuck all of this, I just want to train neural networks on it. Does anyone have the benchmarks for that?
Caleb White
AMD is betting the farm on DX12/Vulkan performance but NVIDIA has the warchest to pay off devs and producers to cling to DX11 for a little longer which will hurt AMDs sales in the short term and potentially ruin them overall.
Christopher Long
get a 144hz monitor,
Jordan Martinez
>that means that for the vast majority of games, the rx 480 would be around HALF as fast as a gtx 980.
Who cares? Most every new games is going to be DX12 or Vulkan.
Older Dx11 games already run well enough.
Jeremiah Thomas
You have this all wrong. There may be some mild improvements on performance on the 480 over a 390. But it's all about making the cost of entry and power usage a lot lower for 1080 gaming. It also allows the option to CF them at a much lower cost to perform on par with a more expensive Nvidia single card solution. True it may or may not cause microstutter on DX11 games (That seems debatable from what I have read about it). But it's all about the long game with AMD. They are punting on DX12 for the future. I know that means things will probably be somewhat shaky in this transient period. AMD will take a hit on DX11 like they always have but it's all about evening up the field. When AMD get their Vega offering out the door the game will change even further and Nvidia will have to really deliver with their 1080Ti to stay relevant. So yes you Nvidia fans can beat your chests about lack of DX11 performance on AMD cards RIGHT NOW. But give it a couple of years and you may be having to admit you got it all wrong and paid too much to have the latest and greatest.
Jordan Young
>that isnt how benching works I'm tired of seeing top of the line (or close to it) $1100-1500+ GPU-less systems used for benchmarking, when the majority of people play on older dual and quad cores, not overclocked, with overall much worse specs
8350s and 3770ks still need to be benched. Dual core i3s need to be benched. Not 8 core, 4Ghz 64GB RAM monster units with triple RAID SSDs and waterproofing, being compared with the same generation 6 core at 4.2Ghz
It gives people an unrealistic expectation of their own real world performance, and many are either too stupid or too lazy to figure out how the FPS scores translate to their own machine, if something of that complexity can even be compared with any accuracy to begin with.
Josiah Miller
>waterproofing you know what I meant Im fucking tired, suck a dick.
Dominic Cook
>larger than average nose >likes shekels
hmm I smell something
Jaxson Wood
Wait a second, what the fuck? What is this 67EF:C3 about? The Polaris leaks had the 67DF:C7 - which we now know to be the 480 - and 67DF:C4 - which is presumably the 470. I have never heard of an EF:C3 before now. Where the fuck did this come from? That framerate looks really low. Is that the 460, or even the 450? What the fuck is going on?
Jaxon Flores
I remember the aots benchmark being on 1440p.
Adrian Powell
if its good, the engine could be ported for other games
