AMD's RX 480 hits half the performance of a Titan X (Pascal) at 8K resolution

[patrickjp93]

10 minutes ago, Prysin said:

that makes no sense, we all know AMD scales so much better with resolution.

 

I bet AMD wins at 128k by 0.0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000125%

That's within statistical error at best and is almost certainly within error of measurement at worst.

 

 

[patrickjp93]

Why do we CARE?! Half of SHIT is still SHIT!

[GlassBomb]

I'm just going to throw in that these silly benchmarks mean very little. Simply because of how unreasonable the tester were with the video settings.

 

"Hey, we got this new hot game. Let's benchmark it at 8k! WITH HIGHEST SETTINGS POSSIBLE" = Everything on ultra/highest possible setting.

 

I would much, much rather see reasonable settings being ran. For 8k resolution, that would mean low, medium settings at the very best.

 

I can't take the writer of that article serious in any kind of manner. Not when he's running the benchmark as he did, along with throwing out "Gears of War 4 is one of the best games released on the PC this year" for a game that has only been out for a couple of days.

 

Bleh - Whatever, I'll just leave it at that.

[Belgarathian]

5 hours ago, Drak3 said:

Allow me to explain it again.

 

8K is the measure of horizontal pixels. This measurement is useless without the vertical measure of pixels or the aspect ratio.

 

The industry standards are 3:2 (15:10), 4:3, 16:9, 16:10, and 21:9.

 

Multiple monitor setups can have horizontal pixel counts of 7680, thus being 8K. When using 4K panels that have aspect ratios of 16:9, the resolution of the two monitors is 7680x2160 and have an aspect ratio of 32:9. When using 1440p monitors that have aspect ratios of 16:9, the resolution of three monitors is 7680x1440 and have an aspect ratio of 48:9.

Nvidia has a name for these multiple monitor setups, it's called Surround. AMD also has a name, Eyefinity. Under both, multiple monitors are used as a single display, which is the same idea behind older ultrawide (21:9) monitors.

That is incorrect. Based on my pixel height of 2160, my monitor has 2160 rows of pixels. That is the only information you have to go off of, and you cannot reinvent the method of measuring the resolution to justify the notion that horizontal resolutions are the only determining point of total pixel counts.

If the aspect ratio is 16:9, then I have 3840 columns of pixels. That gives me a resolution of 3840x2160. That is only if my monitor is 16:9.

Forget it, never mind. Clearly not following what I'm talking about. 

 

Basically - If you were to benchmark similar performance, the card has to drive the same number of pixels. In this case, they've gone with a 16:9 benchmark which is ~33.18m pixels total. Someone else earlier was talking 7680x2160, which while technically 8K, is half the resolution of 7680x4320 and should therefore return 2x the performance. 

[Prysin]

3 hours ago, patrickjp93 said:

That's within statistical error at best and is almost certainly within error of measurement at worst.

 

 

not when you have 0.00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001FPS in the first place

[That Norwegian Guy]

They used 'ultra' for Anti-aliasing.

 

On 8K

 

Needless to say, they are idiots unfit to benchmark a toaster.

[TidaLWaveZ]

13 hours ago, dexxterlab97 said:

Yo listen up here's a story
About a little guy that lives in a blue world
And all day and all night and everything he sees
Is just blue like him inside and outside
Blue his house with a blue little window
And a blue corvette
And everything is blue for him and himself
And everybody around

Always blue! Always blue! Always blue! Always blue!

 

[Fetzie]

17 hours ago, ddswh1pk0s said:

honestly that graph is completely useless considering none of the cards can even achieve cinematic 30 fps

That 30 FPS = cinematic is bullshit.

 

Movies are filmed at 24 FPS

[Enderman]

18 hours ago, Prysin said:

 

I think you missed the fact that in English, if you say "X scales with Y" without stating "exponential" or "logarithmic" or "squared" it is implied that it is linear.

 

My first post was talking about how "performance doesn't scale [linearly (implied)] with price".

Another hint for you would have been my second post, where I said specifically that it is "exponential", making it even more obvious that the first post was referring to a linear correlation.

 

I understand that English isn't your first language, but there is honestly no need for all the profanity and name-calling that you use to defend yourself.

[Samfisher]

On 10/13/2016 at 8:59 AM, dexxterlab97 said:

agree. AMD cost 1 of 6th of the price of TITAN XP

Titans were never made to compete on price anyway.  It's like comparing a 7950 with a Nvidia Tesla card cos HEY THAT CARD EXISTS so we must compare to it.. Wat.

[Guest]

Just now, Samfisher said:

Titans were never made to compete on price anyway.  It's like comparing a 7950 with a Nvidia Tesla card cos HEY THAT CARD EXISTS so we must compare to it.. Wat.

because someone mention something about price. I agree. It would be better comparing 1080ti with an amd

[Inkz]

On 10/12/2016 at 8:52 PM, bcredeur97 said:

yeah AMD's GPU architecture is great right now... if only they would make a higher end card lol

I was literally going to type the exact same thing when I read that title. I was like, "it would be great to also have a card that beats or competes similarly at a lower price point, not just half"

[Kenoxabe]

This benchmark is irrelevant.

 

 

The response(s) in this thread, however, is awesome.

 

The fight between Price VS die shrink VS cars VS graphs.

 

 

Keep em coming, people. 

[MageTank]

20 hours ago, patrickjp93 said:

Why do we CARE?! Half of SHIT is still SHIT!

Better question is: Why does tweaktown (in their own comment section) think this is a perfectly accurate test, when no consumer gaming card in the world has the bandwidth or frame buffer large enough to handle 8k with maxed out AA? They act as if this is a completely fair comparison to make, claiming that at 8k, the RX 480 is half as fast, therefore, future AMD cards will be even better in comparison, but completely neglect their very flawed testing methodology.

 

Give me the worlds fastest card, and I can still bottleneck it to a point in which even a decade old GPU will match its performance. Point is, we cannot use selective data to base an entire point on. With their logic, the 1070 is far more amazing a GPU, because it performed less than half a frame per second slower than the 1080, but costs a lot less. We know this isn't true in the vast majority of situations, where the 1080, on average, is 20-30% faster than the 1070. We also see the Titan Pascal being 30% faster than the 1080. 

 

If my 9800GT (It died, may it RIP) gets 1FPS at 16k full AA, and the Titan X (Pascal) gets 1FPS at 16k full AA, does that mean my 9800GT is just as fast as the Titan XP? Tweaktown seems to think that's how it works. 

[patrickjp93]

41 minutes ago, MageTank said:

If my 9800GT (It died, may it RIP) gets 1FPS at 16k full AA, and the Titan X (Pascal) gets 1FPS at 16k full AA, does that mean my 9800GT is just as fast as the Titan XP? Tweaktown seems to think that's how it works. 

It's not?!

 

Also, anti aliasing is so stupid at 4K and higher. There's no visual benefit anymore!

[Prysin]

4 minutes ago, patrickjp93 said:

It's not?!

apparently not. Weird that. BUT by that logic, a 9800GT costs what on the used market? 20$?
So 1FPS with 1GPU, then you just need 60 of em for a decent playable experience

[Shakaza]

3 hours ago, Enderman said:

I think you missed the fact that in English, if you say "X scales with Y" without stating "exponential" or "logarithmic" or "squared" it is implied that it is linear.

 

My first post was talking about how "performance doesn't scale [linearly (implied)] with price".

Another hint for you would have been my second post, where I said specifically that it is "exponential", making it even more obvious that the first post was referring to a linear correlation.

 

I understand that English isn't your first language, but there is honestly no need for all the profanity and name-calling that you use to defend yourself.

Is that a fact? I never knew that before. I had always thought that the raw statement "X scales with Y" without any context meant "X scales with Y" without any extra meaning tacked on. After all, it seems to make sense that way(maybe it's just me, but we'll see). In your earlier example, you said

Quote

Since when does performance scale with price??

Since never.

Now, what I want to know is what you meant by that. By your definition, you were saying that performance does not scale linearly with price(but may scale in other ways). What I thought you said was that performance doesn't scale with price in any form, which may or may not be true depending on context.

 

For your next point, let me explain to you a little thing called clarity. To me, it appears that @Prysin was confused because, from his perspective, you were putting words into his mouth and claiming he said things that he didn't. Therefore, like any rational human being, thought you were being incredibly dense, so he attempted to spell out his intended meaning so you would finally understand. It wouldn't have been a problem in the first place if you had been more clear about your message. Language is a means to transmit information to others, not a set of rules to follow. Perhaps more scholarly people will disagree with me, but I always find it best to be completely explicit when I am communicating. Having to make assumptions about what others are trying to say is very cumbersome and detracts from the initial purpose of the conversation. I can't tell you how you should or should not engage in conversations, but I recommend clarity as a first priority to avoid unnecessary confusion or conflict.

 

Also, I think you may have missed the point that Prysin was trying to get across in the first place, but I'll just leave it at that. 

[patrickjp93]

3 minutes ago, Shakaza said:

Is that a fact? I never knew that before. I had always thought that the raw statement "X scales with Y" without any context meant "X scales with Y" without any extra meaning tacked on. After all, it seems to make sense that way(maybe it's just me, but we'll see). In your earlier example, you said

Now, what I want to know is what you meant by that. By your definition, you were saying that performance does not scale linearly with price(but may scale in other ways). What I thought you said was that performance doesn't scale with price in any form, which may or may not be true depending on context.

 

For your next point, let me explain to you a little thing called clarity. To me, it appears that @Prysin was confused because, from his perspective, you were putting words into his mouth and claiming he said things that he didn't. Therefore, like any rational human being, thought you were being incredibly dense, so he attempted to spell out his intended meaning so you would finally understand. It wouldn't have been a problem in the first place if you had been more clear about your message. Language is a means to transmit information to others, not a set of rules to follow. Perhaps more scholarly people will disagree with me, but I always find it best to be completely explicit when I am communicating. Having to make assumptions about what others are trying to say is very cumbersome and detracts from the initial purpose of the conversation. I can't tell you how you should or should not engage in conversations, but I recommend clarity as a first priority to avoid unnecessary confusion or conflict.

 

Also, I think you may have missed the point that Prysin was trying to get across in the first place, but I'll just leave it at that. 

Positively or negatively?

 

Not to be an ass, but precision and specificity matter.

 

I otherwise agree with your post.

[MageTank]

22 minutes ago, patrickjp93 said:

It's not?!

 

Also, anti aliasing is so stupid at 4K and higher. There's no visual benefit anymore!

Yeah, I have been saying this to people I game with on Overwatch. They have 4k monitors, run 200% render scale AND ultra AA, and brag about how sharp the image is, when it looks no different than my 1440p ultra AA setup. They also wonder why they are lagging with high end setups. My 1070 can't even handle 1440p at 200% render scale AND AA, so their 1080's are struggling hard. 

[Thony]

Next: RX480 performes only 10% worse at 20K than Titan X !

[TheRandomness]

On 13/10/2016 at 2:01 AM, Marvzl1357 said:

And so the wait for the RX 490 continues...

Well, if the 485 is actually going to be a thing for the desktop that'd be quite interesting (lower power draw same performance).

[DXMember]

On ‎2016‎.‎10‎.‎13‎. at 4:18 AM, ARikozuM said:

So... You're saying that the R9 Fury X will be succeeded by... the RX Fury X... and then the RXI Fury X?

 

[-BirdiE-]

On 10/12/2016 at 9:50 PM, Drak3 said:

8K in 32:9 would be 7680x2160. It's just the horizontal pixel count, and doesn't mean anything without the aspect ratio or vertical pixel count.

Not true at all...

 

1080p 21:9 monitors are 2560x1080..

1440p 21:9 monitors are 3440x1440...

 

They call them ultraWIDE monitors because they are WIDER than their traditional 16:9 counterparts. If 2560x1080 was considered a 1440p monitor, then it would be the same width as a 1440p 16:9 monitor... and that's not ultrawide at all.. it would be ultrashort.

[Misanthrope]

Riveting.

 

Also if you go all the way to 16k it has the SAME performance!:

 

-Titan XP: 3 FPS average

-480: 3 FPS average

 

Just as meaningful numbers.

[The Benjamins]

1 hour ago, -BirdiE- said:

Not true at all...

 

1080p 21:9 monitors are 2560x1080..

1440p 21:9 monitors are 3440x1440...

 

They call them ultraWIDE monitors because they are WIDER than their traditional 16:9 counterparts. If 2560x1080 was considered a 1440p monitor, then it would be the same width as a 1440p 16:9 monitor... and that's not ultrawide at all.. it would be ultrashort.

whats confusing is 21:9 content does not follow this, 1080p 21:9 movie is 1920 x 800.

 

3840 x 1600 is 4k 21:9 LG just came out with a monitor with this res.