Regarding the RX 480 AOTS benchmark at Computex

[zMeul]

16 minutes ago, Fulgrim said:

RX 480 Crossfire

http://www.ashesofthesingularity.com/metaverse#/personas/b0db0294-8cab-4399-8815-f956a670b68f/match-details/ac88258f-4541-408e-8234-f9e96febe303

GTX 1080

http://www.ashesofthesingularity.com/metaverse#/personas/b0db0294-8cab-4399-8815-f956a670b68f/match-details/a957db0f-59b3-4394-84cc-2ba0170ab699

• neither of those numbers match what AMD showed
• the runs were on different game versions
• the diff between the RX and GTX is barely worth mentioning - not even close to what AMD showed

to put it simple: those are not the runs AMD used

 

[TidaLWaveZ]

17 minutes ago, GlassBomb said:

I'm still generally not a fan of multi-GPU solutions. As I said before, I'm going to wait until average Joe can get their hands on cards.

 

And at this point in time, Dx12 benchmarks mean very little to me, since I'm still running Win7. And don't plan to switch for a good time.

 

Me neither, I've been disappointed with past SLi/Crossfire setups.  I edited my comment, it was really meant just to point that it's more likely that they used the 8GB cards. This would just cut the price gap down to $140 and I think once we see more benchmark results that $140 will probably be justified.

[Fulgrim]

3 minutes ago, zMeul said:

• neither of those numbers match what AMD showed
• the runs were on different game versions
• the diff between the RX and GTX is barely worth mentioning - not even close to what AMD showed

to put it simple: those are not the runs AMD used

 

Computex demo = 1080p

 

I linked you 1440p benchmarks, which are irrefutable. 

[zMeul]

7 minutes ago, Fulgrim said:

Computex demo = 1080p

 

I linked you 1440p benchmarks, which are irrefutable. 

irrefutable in what case? 

GTX1080 at 1140p crazy gets 58FPS avg (normal batches)

and you're telling me 2x RX480 result was at 1080p ?!?! because otherwise, I sincerely do not understand

[PlayStation 2]

All smoke, no substance. 

AMD, why didn't you show us something related to, I dunno, Hitman? Dirt Rally? Star Wars Battlefront? Any game within the last year you had your hand in?

[Tedol]

32 minutes ago, Notree said:

Nvidia fan?  ...

yep

[LeStringMan]

Every one in this thread is thinking of multi gpu all wrong, for instance i have crossfire R7 360, 360+360=720, so obviously i have the power of an nVidia Geforce Gt720, meanwhile 480+480=960 so obviously the 1080 is still faster, and to my justify my earlier point that Fermi is the best nVidia architecture because of it's async hardware support 580+580=1160, 1160>1080, confirmed on LTT forums that SLI GTX 580s is the best 4k gaming setup /thread.

[No Nrg]

16 minutes ago, ivan134 said:

Dumb. How is number of people who play it relevant to testing the dx 12 support of cards? Would number of people playing it change the benchmark scores? If you think this is an outlier, you are welcome to go test Hitman, Quantum Break and Total War: Warhammer to see whether Nvidia sucks or not at asynchronous compute. People said Aots was one game and we should wait. We got 3 more and people are still saying wait. Don't worry we have Deus Ex, BF1, Mass Effect Andromeda, Battlezone and Watch Dogs 2 still coming up. How many more exactly do you need again?

 

 

None of the games you mentioned have been tested on the RX 480, so I'll wait and see. Given the 1080 runs 81fps in Hitman & 80fps In Total War in 1440p that's not too shabby. Quantum Break is poorly optimized at best, no one even uses it for benchmarks I can find.

 

I'll wait till the other games you mentioned are even released before using them as proof in an argument.

[Tedol]

2 minutes ago, LeStringMan said:

Every one in this thread is thinking of multi gpu all wrong, for instance i have crossfire R7 360, 360+360=720, so obviously i have the power of an nVidia Geforce Gt720, meanwhile 480+480=960 so obviously the 1080 is still faster, and to my justify my earlier point that Fermi is the best nVidia architecture because of it's async hardware support 580+580=1160, 1160>1080, confirmed on LTT forums that SLI GTX 580s is the best 4k gaming setup /thread.

hahaahaahahah, good one)))) 

[Fulgrim]

15 minutes ago, zMeul said:

irrefutable in what case? 

GTX1080 at 1140p crazy get 58FPS avg (normal batches)

and you're telling me 2x RX480 result was at 1080p ?!?! because otherwise, I sincerely do not understand

Irrefutable in the sense that they are benchmarks done within the game, and uploaded directly to AOTS servers. You can see the game settings, as well as the framerates for batch types etc.

Are you suggesting they are refutable? And if so, please explain how.

[LeStringMan]

8 minutes ago, No Nrg said:

 

 

None of the games you mentioned have been tested on the RX 480, so I'll wait and see. Given the 1080 runs 81fps in Hitman & 80fps In Total War in 1440p that's not too shabby. Quantum Break is poorly optimized at best, no one even uses it for benchmarks I can find.

 

I'll wait till the other games you mentioned are even released before using them as proof in an argument.

holy crap RIP gtx 960 in that total war bench.

[zMeul]

3 minutes ago, Fulgrim said:

Are you suggesting they are refutable? And if so, please explain how.

I explained in my original post

[DocSwag]

Saw this, and it definitely is interesting.

Still waiting for actual benchmarks for more games. Afterall, most games are still DX11.

I don't care too much, since I'm going Vega anyways (unless the 1080ti absolutely murders Vega. I have a freesync monitor so I'm willing to take, say, a Fury X over the 980 ti, so if Vega is 10% worse than GP102/100 I'll still go Vega).

 

WHY DO WE ALWAYS HAVE TO WAIT I JUST WANT STUEPID BENCHMARKS!!!!!!!!!!

[GlassBomb]

3 minutes ago, DocSwag said:

WHY DO WE ALWAYS HAVE TO WAIT I JUST WANT STUEPID BENCHMARKS!!!!!!!!!!

Because they want to build the hype.

[DocSwag]

2 minutes ago, GlassBomb said:

Because they want to build the hype.

I know....

Well, it sure ain't working on me. I'm just getting annoyed instead of hyped lol

[Daiyus]

7 minutes ago, zMeul said:

I explained in my original post

OK, so they aren't the EXACT ones from the show, but they're both at Crazy 1440p settings in a standardised benchmark showing the RX 480 CF matching up to the GTX 1080. It's all there in black and white.

 

Now that's a single game that we know AMD generally performs well in compared to nVidia, but the fact is these RX 480's are no slouch at their price point. I'm waiting for more benchmarks in different games, but even so I doubt I'll be disappointed. These cards weren't ever meant to take on the GTX 1080, they're entry level cards for crying out loud, but at least the scalability is there to seriously push some pixels if you want it after you've put together your budget build. What I find amazing is that for ~£200 I can pick up a card that will destroy any game at 1080p Ultra and run VR. That'll suit me until I can afford to go 4K with a flagship card and a new screen.

[Tedol]

8 minutes ago, GlassBomb said:

Because they want to build the hype.

aaa... you know, people... want hate... yep

 

[CommandMan7]

@Fulgrim Thank you for the Clarification, this was much needed.

 

As for some educated guessing on the 480's performance in mainstream / DX11 titles:

 

A single Fury X is 18% faster than a single 980 Ti in Ashes of the Singularity 1440p.

A single 980 Ti is 17% faster than a single Fury X in Battlefield 4 1440p. 

A single 1080 is 61% faster than a single 480 in Ashes of the Singularity 1440p.

 

That gives Ashes of the Singularity a 35% bias towards AMD. If you do some napkin math, 

 

A 1080 should be 61% faster + 35% lost in Ashes of the Singularity = 96% faster.

A 1080 gets 105 FPS in Battlefield 4 1440p.

Divide 105 FPS by 1.96, and you get 54 FPS.

 

That means the $200 480 is 96% slower than a $600 1080 for 300% less cost.

 

Conclusion: The 480 is half as fast as a 1080 for one third of the price, even in DX11.

 

If messed up my calculations somewhere, I haven't taken statistics and this is just an approximation. A very rough approximation, but still a very telling result.

 

Sources:

Spoiler

Single 480 Benchmark Ashes of the Singularity:

http://www.ashesofthesingularity.com/metaverse#/personas/b0db0294-8cab-4399-8815-f956a670b68f/match-details/8b748568-fc96-4e48-9fed-22666a7149f5

Single 1080 Benchmark Ashes of the Singularity:

http://www.ashesofthesingularity.com/metaverse#/personas/2a44f7bb-1f6f-4dbb-9ed9-603e917a5ec3/match-details/54cb7f38-8dd8-49d5-8be9-dbef57cd6074

Single Fury X and Single 980 Ti Battlefield 4:

http://www.tomshardware.com/reviews/nvidia-geforce-gtx-1080-pascal,4572-6.html

 

[zMeul]

7 minutes ago, Daiyus said:

These cards weren't ever meant to take on the GTX 1080

no? interesting because AMD showed on stage that two of thier RX480s can beat a GTX1080

while the actual results show barely a 0.1 FPS diff in average - normal batches

 

plus the difference in scene quality on AMD's RX runs

 

---

 

I do not have the game not do I intend to purchase it - can anyone confirm without any doubt that the bench run can't be modified and would show differently on-line?

do you see the core clocks these video cards ran at?

 

there are 1440p crazy runs of GTX1080 that show way higher avg FPS than AMD's run - and keep in mind, that linked GTX1080 run was on a older version of the game

[Fulgrim]

10 minutes ago, zMeul said:

no? interesting because AMD showed on stage that two of thier RX480s can beat a GTX1080

while the actual results show barely a 0.1 FPS diff in average - normal batches

 

plus the difference in scene quality on AMD's RX runs

 

---

 

I do not have the game not do I intend to purchase it - can anyone confirm without any doubt that the bench run can't be modified and would show differently on-line?

do you see the core clocks these video cards ran at?

 

there are 1440p crazy runs of GTX1080 that show way higher avg FPS than AMD's run - and keep in mind, that linked GTX1080 run was on a older version of the game

The reason why there is a difference between the GTX 1080 & RX 480 CF benches are explained in the OP, you tit. You're trying to dismiss the official statement from AMD, as well as the two AOTS 1440p benchmarks i linked you, as a comparison between the two GPU setups (albeit not the specific ones used in the computex livesteam), that are using reference models of either card.

[CommandMan7]

2 minutes ago, zMeul said:

no? interesting because AMD showed on stage that two of thier RX480s can beat a GTX1080

while the actual results show barely a 0.1 FPS diff in average - normal batches

 

plus the difference in scene quality on AMD's RX runs

 

---

 

I do not have the game not do I intend to purchase it - can anyone confirm without any doubt that the bench run can't be modified and would show differently on-line?

do you see the core clocks these video cards ran at?

 

there are 1440p crazy runs of GTX1080 that show way higher avg FPS than AMD's run - and keep in mind, that linked GTX1080 run was on a older version of the game

Here are 2 Perfectly Identical Benchmarks of the 480 and 1080:

 

(2x) 480 Benchmark - 60 FPS - Extreme 1440p

1080 Benchmark - 64 FPS - Extreme 1440p

 

Every setting and between these 2 benchmarks are perfectly identical. So you are correct, (2x) 480 cannot beat a single 1080, but it comes damn close and for a lot cheaper.

 

[Daiyus]

2 minutes ago, zMeul said:

no? interesting because AMD showed on stage that two of thier RX480s can beat a GTX1080

while the actual results show barely a 0.1 FPS diff in average - normal batches

 

plus the difference in scene quality on AMD's RX runs

 

---

 

I do not have the game not do I intend to purchase it - can anyone confirm without any doubt that the bench run can't be modified and would show differently on-line?

do you see the core clocks these video cards ran at?

Dude, we're on PC's; anything can be tampered with if you want to. I'd say official benchmarks from the databanks of the developers of the game are about as official as you'd get from anywhere outside benching it yourself. Argue about the on-stage demo's all you want, but in a standard benchmark independent of the show with identical settings we're seeing the parity.

 

This is where my opinion comes in; I believe they decided to show off the scalability of their new cards merely to show off how cheaply you can get similar performance to a flagship card. They're effectively calling nVidia out on their ridiculous pricing as well as saying "hey, you can build a kick ass system on a budget, but you can then buy another card cheap and see similar performance to a flagship card".

 

If I were new to the game and didn't know much about SLI/Crossfire and I was planning my first budget build that'd certainly make me sit up and listen; why should I cough up £620 now just for a GPU when I can build a complete system for that and throw £200 more at it later to get comparable performance?

 

Sure, for enthusiasts that are already on flagship cards then the RX 480 is irrelevant, they're already happy to splurge that cash on a flagship card as it's something they're confident in, just like car junkies don't think twice about a £1000 exhaust system, but for entry level users it's a nice upgrade path to have the option of. 

[TidaLWaveZ]

6 minutes ago, CommandMan7 said:

Here are 2 Perfectly Identical Benchmarks of the 480 and 1080:

 

(2x) 480 Benchmark - 60 FPS - Extreme 1440p

1080 Benchmark - 64 FPS - Extreme 1440p

 

Every setting and between these 2 benchmarks are perfectly identical. So you are correct, (2x) 480 cannot beat a single 1080, but it comes damn close and for a lot cheaper.

 

 

~$140 price difference.  May be a deal breaker for certain budgets, but IMO justifies a single card setup.

[zMeul]

1 minute ago, Fulgrim said:

The reason why there is a difference between the GTX 1080 & RX 480 CF benches are explained in the OP, you tit. You're trying to dismiss the official statement from AMD, as well as the two AOTS 1440p benchmarks i linked you, as a comparison between the two setups, that are using reference models of either card.

watch your language buddy

 

oh yes, I'm dismissing outright Robert's BS excuse with the snow

like I said in the other thread, it doesn't explain the difference in the mountain ridges details - they are flatter and muddier on the RX run

the excuse of YT compression doesn't hold either since the source is exactly the same

 

don't you find it funny that RX480 NDA lift and release date is on the exact same day?! just like it happened with R9 390(X) when it was shown to be just a rebadge with more VRAM

but don't you worry, on the 29th you'll know the truth

[GlassBomb]

Just now, TidaLWaveZ said:

~$140

Part of me is asking why they didn't try to use a 1070 and tell everyone "our $400 solution is better than their $400 solution".

 

But I guess it sounds better when you compare it to the 1080.