[Benchlife] R9 480 (Polaris 10 >100w), R9

[IllogicalGlory]

Pro Duo isn't even going to be released to consumers. It's a total non-factor to gamers, and probably means little even to AMD at this point. Although, I think they got a supercomputer contract using them, so there's that at least.

Besides, who needs a completely balanced stack? What went between the $3000 Titan Z and the $550 980?

 

[JDG1980]

Remember overclocker's dream ? yeah. nobody can trust that guy again.

That was Joe Macri, not Robert Hallock.

 

[KaRLiToS]

Remember overclocker's dream ? yeah. nobody can trust that guy again. i seriously don't get amd fanboys, instead of waiting they always overhype every little piece of info. and when the product doesn't perform "wait for drivers hurr durr'' i mean. can't you guys just wait for release and stop hyping ?


This crap is doing more brand damage than nvidia fanboys doo with their no.1 motto " amd drivers suck"

Robert Hallock is no Joe Macri.

But you are right, it's better to be careful, wait for the info and real benchmarks to come out instead of hyping things up and get disapointed afterward. Been like that for 3 years already. Hope Zen, Polaris and Vega will be up to the task.

 

[IllogicalGlory]

That was Joe Macri, not Robert Hallock.

And more importantly, while misleading and underhanded, "overclocker's dream" is a pretty abstract term, and it can't strictly be called a "lie", even if it practically speaking it is. The demo was either running at ultra settings or it wasn't, and I don't think they'd have the balls to outright lie about that.

 

[tential]

The problem is twofold:



1. There is a pent up desire for cards on a new node.



2. People want to ignore how long we will be on the next node.

But why are we obsessed with die size and memory tech?

HBM didn't all of a sudden make chips super fast.
Having similar die sizes didn't make the GTX 980Ti and Fury X perform similarly.
GTX 970 having less memory bandwidth than the gtx 780Ti didn't just kill the card.
etc.

Also, yes we're going to be on this node for AWHILE.

Also, I even see people applying the same perft/watt jump that Polaris is getting to Vega.

Yet, Vega is listed as a NEW architecture with even HIGHER perf/watt. We don't know about Vega yet either then, and if AMD is using 2 architectures in relatively quick succession of each other to compete with 1 architecture, Pascal, that's also something we should take into account.

 

[JDG1980]

Isn't it clear by now that 2.5x perf/watt gain is over Tonga?

Tonga peaks at about 180W in gaming. (It can go higher under torture tests like FurMark, but that's largely just due to where they set the power limit.) If we assume the 2.5x perf/watt gain is over Tonga, then that would mean R9 380X gaming performance at 72W (180 / 2.5). This sounds about right for Polaris 11 if pushed to the maximum PCIe power limit, though we might not see that at first if all the good dies go to Apple and other laptop vendors. There should be at least one Polaris 10 desktop SKU that pushes the power limit to 150W (the maximum for one 6-pin connector), if not more. Double Tonga performance would beat any 28nm card, even Titan X, but that seems too optimistic for a Pitcairn-sized chip on 14LPP. More likely we'll see performance slightly above Hawaii and below Fiji at 1080p, though falling behind at 4K.

 

[kraatus77]

Robert Hallock is no Joe Macri.

But you are right, it's better to be careful, wait for the info and real benchmarks to come out instead of hyping things up and get disapointed afterward. Been like that for 3 years already. Hope Zen, Polaris and Vega will be up to the task.

Sorry my mistake. i thought it was Hallock.

 

[3DVagabond]

That was Joe Macri, not Robert Hallock.

Don't let facts get in the way of a good thread crap.

 

[kraatus77]

Don't let facts get in the way of a good thread crap.

I already said sorry dude, it was an honest mistake from my side.

 

[Hazy24]

https://youtu.be/J3IYhWVxCco?t=6m58s

^ AMD's Robert Hallock confirms Hitman was running 1440p on Polaris 10 with maxed settings, 60 fps locked, something Fury X cannot achieve at that area/view.

This is how they quoted him in their latest interview (18th April 2016)

Question 3b: The Hitman demo was quite interesting – albeit rather short. Can you confirm Hitman was played at max settings, DX12 1440P on prototype hardware (I believe Polaris 10) and was actually capped at 60FPS. This had lead folks to speculate the GPU was / is capable of much more, especially on final production hardware. Are you able to comment on this?

RH: You are correct that this was the first public showcase of Polaris 10 running DX12 1440p. However, I cannot confirm more than that at this time.

http://www.redgamingtech.com/interv...ock-polaris-graphics-technology-vr-dx12-more/

 

[Abwx]

This is how they quoted him in their latest interview (18th April 2016)

http://www.redgamingtech.com/interv...ock-polaris-graphics-technology-vr-dx12-more/

And..?..

The card was draining less than 30W to produce thoses 60FPS, you think that a 232mm2 chip is limited to 30W.?..

 

[Hazy24]

Just pointing out that he didn't confirm the max settings...

 

[airfathaaaaa]

And..?..

The card was draining less than 30W to produce thoses 60FPS, you think that a 232mm2 chip is limited to 30W.?..

you are talking about the first live demo of polaris 11 with the power meter

Just pointing out that he didn't confirm the max settings...

you are talking about the second live demo of polaris 10

here

 

[Abwx]

Just pointing out that he didn't confirm the max settings...

He said that it was at 1440p and that he wasnt allowed, apparently, to give more precisions, we can confidently assume that both Polaris and the 950 were running at the same settings, otherwise the comparison wouldnt make sense at all.

Besides, if Polaris was in the vicinity of 30W then the 950 was at about 84W, wich is close to its 90W specification, you dont see a pattern and a logic here..?..

you are talking about the first live demo of polaris 11 with the power meter

Whatever, they couldnt get at a higher power for Polaris as the GTX950 was already close to its max TDP.

 

[R0H1T]

That's true for CPUs as well. What happened with Intel CPUs before 22nm? The biggest hyped process became the lowest gain in years.

Remember Fury's HBM? We've seen that it was a necessity just to be able to compete with Nvidia, rather than jump forward.

I think the companies are becoming better and better at hyping technologies. I don't think many of us expected both Nvidia and AMD were planning a staggered launch.

Isn't it clear by now that 2.5x perf/watt gain is over Tonga? And its 2x compared to current generation? We don't even know that will apply uniformly to all Polaris products. Also don't forget 14nm only applies to the computing part of the video card. It certainly doesn't apply to memory, and even I/Os on the chip gain far less. That's marketing for ya. I can't believe that top Polaris will be faster if at all than Fury X if they justified on releasing $1599 Radeon Pro Duo. A 20-25% gain would be too close to one that would cost 3-4x and get 80% gain situationally. Unless they make not-the-smartest decision.

$1599 Radeon Pro Duo
$400 Polaris 10

If Polaris 10 performs better than Fury X, what goes in between $1599 and $400?

That's because it doesn't apply uniformly across the same architecture, with different chips, & is the case with CPU's as well. The 380x is the same GCN3 as Fury(x) & Nano yet the most efficient product in that stack is the Nano, but not only because it's artificially TDP/power constrained & very likely is a higher bin as compared to the Fury X's D:

In case of Maxwell the GTX 980 is the most efficient chip across many resolutions, while the cut down 970 (with higher clocks) is less efficient across the board. So, as is the case everywhere else, the number 2.5x floating around is probably for the best case scenario they've come across but that doesn't mean it's the limit of Polaris architecture cause they could just as well cherry pick some cards, slap HBM2 & with the right clocks proclaim it to be 3x as efficient as Nano. I'm not saying that will happen but there's no hard limit to performance & thus efficiency, it should be taken as an average number with lots of *qualifiers

 

[Silverforce11]

Just pointing out that he didn't confirm the max settings...

Watch the video I linked. The guy claims he talked to AMD Robert H and it was confirmed to him it was maxed 1440p. It may not be in the interview with published on the record stuff though.

 

[hawtdawg]

That's true for CPUs as well. What happened with Intel CPUs before 22nm? The biggest hyped process became the lowest gain in years.

Remember Fury's HBM? We've seen that it was a necessity just to be able to compete with Nvidia, rather than jump forward.

I think the companies are becoming better and better at hyping technologies. I don't think many of us expected both Nvidia and AMD were planning a staggered launch.

Isn't it clear by now that 2.5x perf/watt gain is over Tonga? And its 2x compared to current generation? We don't even know that will apply uniformly to all Polaris products. Also don't forget 14nm only applies to the computing part of the video card. It certainly doesn't apply to memory, and even I/Os on the chip gain far less. That's marketing for ya. I can't believe that top Polaris will be faster if at all than Fury X if they justified on releasing $1599 Radeon Pro Duo. A 20-25% gain would be too close to one that would cost 3-4x and get 80% gain situationally. Unless they make not-the-smartest decision.

$1599 Radeon Pro Duo
$400 Polaris 10

If Polaris 10 performs better than Fury X, what goes in between $1599 and $400?

Comparing this to the CPU space is not apples to apples. CPU performance relies on single threaded improvements, which at this point can really only be attained with higher clock speeds.

Fury X didn't need HBM to compete, hell, Hawaii is in the GM200 range in DX12 without even using their color compression tech with regular DDR5. If anything, AMD is the one whos had the better hardware this entire time. On top of that, they have not even done a major overhaul yet like Nvidia did with Maxwell.

Also, we don't know what AMD is cooking up. They may go the NVidia route with Polaris, and strip a lot of the compute abilities out of it, which would still necessitate a need for the Pro Duo.


The fact of the matter is that history and rumored specs have this thing in Fury X range, and the only thing you have really to counter it is "nuh uh". We are on a new process node. look at every new GPU generation ever released on a new process node, and you'll realize that pegging Polaris 10 in Fury X performance range is actually being conservative.

 

[renderstate]

Some people will dismiss AMD's inn...olaris, motivating the new marketing moniker.

 

[guskline]

I'll wait till next year when the Big Vega appears.

 

[maddie]

Comparing this to the CPU space is not apples to apples. CPU performance relies on single threaded improvements, which at this point can really only be attained with higher clock speeds.

Fury X didn't need HBM to compete, hell, Hawaii is in the GM200 range in DX12 without even using their color compression tech with regular DDR5. If anything, AMD is the one whos had the better hardware this entire time. On top of that, they have not even done a major overhaul yet like Nvidia did with Maxwell.

Also, we don't know what AMD is cooking up. They may go the NVidia route with Polaris, and strip a lot of the compute abilities out of it, which would still necessitate a need for the Pro Duo.


The fact of the matter is that history and rumored specs have this thing in Fury X range, and the only thing you have really to counter it is "nuh uh". We are on a new process node. look at every new GPU generation ever released on a new process node, and you'll realize that pegging Polaris 10 in Fury X performance range is actually being conservative.

The Pro Duo is not a fp64 machine, unlike Hawaii based 390/390X.

 

[Slaughterem]

Pro Duo isn't even going to be released to consumers. It's a total non-factor to gamers, and probably means little even to AMD at this point. Although, I think they got a supercomputer contract using them, so there's that at least.

Besides, who needs a completely balanced stack? What went between the $3000 Titan Z and the $550 980?

You are partially correct, Pro duo is not intended for consumer but is being released. The rest of your statement is short sided from what Raja claimed is the importance of the card.
AMD new arch is Polaris which is defined in wiki as a foundation star, hence the first 14nm GCN 4 designs. Polaris 10 and 11 will compete with the entire stack of current cards from mobile to enthusiast which have all be relegated to mainstream under their new road map.
Vega is defined as a double - double star under wiki descriptions. Hence if you consider ram and processor doubling you can see double of shaders and bandwidth with HBM 2. There will be a Vega 10 and 11 which will be a double -double of Polaris 10 and 11. My guesstimate will be something in the 320 mm2 range like GP104 and in 2017 when the 14nm process should have improved yields.
Navi is defined as a variable star both intrinsic and extrinsic. Here is where Pro Duo is important and laying the foundation for multi gpu development under the new API's. Using Interposers you can take all of the foundation and double archs (Intrinsic) and mix and match them with next and current gen memory (Extrinsic).
With Pro Duo the point is Devs need time to learn and implement multi gpu over the next 18 - 24 months so that their game engines can use the forward thinking tech of Navi to be truely successful. Sort of the same process with ACE's but better adoption rate.The fact is that ACE's were a very advanced tech but since devs in the beginning did not make use of them nobody really cared. Now since AMD has the console effect happening and is laying the foundation for NAVI, Devs will adopt to this more rapidly, same as what we are seeing with DX12 adoption rate.

 

[el etro]

My point of view is that all due to economics. A GTX680 card was very likely cheaper to manufacture than a hugely complex(many PCB layers, more VRMs), 250W GTX580 card. In the end they could price the 680 lower than the 7970 while still making good or even better profit with it.

That's where GM206 cards make sense, too.

 

[railven]

Robert Hallock is no Joe Macri.

But you are right, it's better to be careful, wait for the info and real benchmarks to come out instead of hyping things up and get disapointed afterward. Been like that for 3 years already. Hope Zen, Polaris and Vega will be up to the task.

How I'm feeling right now. Would appreciate it if AMD gives me the performance at least in my first month of ownership, not almost a year after I sold off my hardware.

Keeping an eye on Polaris 10, but not buying into the hype this time around.

 

[Mopetar]

How I'm feeling right now. Would appreciate it if AMD gives me the performance at least in my first month of ownership, not almost a year after I sold off my hardware.

Keeping an eye on Polaris 10, but not buying into the hype this time around.

The hype level is getting somewhat out of control, but at least AMD has shown off the hardware in live demonstrations. However, they could certainly choose something like the Hitman demo to showcase the best-case scenario where the new card performs exceptionally well due to architecture improvements or additions in order make it appear better than it actually is in most cases.

Polaris 10 will be a good card, but the ultimate question comes down to what they price it at. Even if it doesn't equal Fury X in all cases, at ~$300 it's still worth the price of admission. But if Polaris 10 is good, Vega should be as well, so I can't see a lot of reason not to wait until both AMD and NVidia have their biggest dies available before making a decision, unless you'll just flip whatever you buy until those come out.

 

[railven]

The hype level is getting somewhat out of control, but at least AMD has shown off the hardware in live demonstrations. However, they could certainly choose something like the Hitman demo to showcase the best-case scenario where the new card performs exceptionally well due to architecture improvements or additions in order make it appear better than it actually is in most cases.

Polaris 10 will be a good card, but the ultimate question comes down to what they price it at. Even if it doesn't equal Fury X in all cases, at ~$300 it's still worth the price of admission. But if Polaris 10 is good, Vega should be as well, so I can't see a lot of reason not to wait until both AMD and NVidia have their biggest dies available before making a decision, unless you'll just flip whatever you buy until those come out.

It ultimately depends where they land. If both cards are 20-30% over my stock 980 Ti, I'd buy one (which would matter on a few things). If either is only 10% or less over my card, I'll skip em.

If I were to buy, I'd pass em on when something faster comes along. With the CPU stagnation, I'm not in a rush to upgrade my whole rig like I use to be.