AMD Radeon Fury X Series | HBM, Small Form Factor And Water Cooling | June 16th

Oh, of course. I was just pointing out that memory stacking isn't an AMD feature but a DX12 feature.


SFR mode is a feature of mantle. Civilization does this under mantle in which for CrossFire the memory pools are not replicated. In DX mode it uses AFR.


Be very careful about the vram across gpus becoming one.

Its only on DX12 and so far we havent saw any real world examples of this.

Pick your salt rock.

Google mantle, SFR, Civilization
 
SFR mode is a feature of mantle.
It's not a "Mantle feature", and it's not a "DX12 feature" either. It's certainly not an AMD feature or an Nvidia feature for that matter.

If a developer manually supports multi-GPU (which they have to in low-level APIs) then they can decide how to distribute their data, and they may decide, depending on their rendering setup, not to replicate everything. But it's also certain that you won't get a straightforward additive combination of available memory. Some assets/buffers will always need to be replicated.

People really misunderstand this a lot.

Low-level APIs don't feature SFR. They allow developers to implement SFR if they choose to.
Also, such custom solutions may or may not replicate all or part of the GPU memory. It's not a simply X+X = 2X calculation.
 
People really misunderstand this a lot.

Low-level APIs don't feature SFR. They allow developers to implement SFR if they choose to.
Also, such custom solutions may or may not replicate all or part of the GPU memory. It's not a simply X+X = 2X calculation.


Thanks for that however I know how this stuff works in terms of APIs. I call it a feature because that's how the wording in the press releases and media refer to it as when Mantle was being talked about. Same with DX 12.

I understand where you're going with this.
 
I will wait for next week just to confirm, but I fail to see how they could not include HDMI 2.0 in a card made specifically for "4k gaming"? Seems totally counterintuitive. People are going to hook it up to big screen 4k TV's and are going to need that HDMI 2.0 support.

DP to HDMI 2.0 converters don't currently exist, and when they do they will probably cost >$100.
 
Can't a DP to HDMI 2.0 cable work?

No. DisplayPort didn't gain the ability to output a HDMI 2.0 compatible signal until 1.3. There are no DP 1.3 capable video cards. A simple cable will not work.

edit: As far as active converters go, I have no idea but those active converter boxes are not cheap.
 
No, active converter cables won't work. You need a specific DP to HDMI 2.0 converter cable and they don't exist. One company claims to be working on one but who knows?
 
New ASIC design though, surely it could have included HDMI 2.0 if they wanted. Still using GCN yeah but it's not the same GCN as in Hawaii.
 
So someone on overclockers.uk extrapolated from the FarCry 4 Ultra demo they showed at the PC Gamer show a 28% increase over Titan X performance at that 4k resolution.

Yeah it's just one game but as far as benchmarks go, it's a start.
 
Does the 980ti come with HDMI 2.0?

Edit: Google says yes. Not good AMD...

That's why if true AMD could potentially lose a lot of sales from the 4k gaming crowd. I will wait until next week to confirm just to be sure, if so I will pick up a 980ti.
 
The release date for the Fury X is the 24th this month, so we'll probably see reviews by then, unless there are 0 cards available worldwide.

Didn't expect this to release so soon.

When is the regular Fury releasing? Going to buy one when a 3rd party puts their own cooler on them.
 
That's why if true AMD could potentially lose a lot of sales from the 4k gaming crowd. I will wait until next week to confirm just to be sure, if so I will pick up a 980ti.

Not saying you are wrong, but that only applies to 4K TVs. The overwhelming majority of 4K monitors are DisplayPort 1.2a.

When is that nano expected to be released?

All we know is "summer". So before the end of August.
 
So someone on overclockers.uk extrapolated from the FarCry 4 Ultra demo they showed at the PC Gamer show a 28% increase over Titan X performance at that 4k resolution.

Yeah it's just one game but as far as benchmarks go, it's a start.

Didn't that game favor AMD GPUs?

http--www.gamegpu.ru-images-stories-Test_GPU-Action-Far_Cry_4-nv-test-FarCry4_3840.jpg


This benchmark had the 290X nearly matching the 980, so I wouldn't be too surprised if the Fury X did better than the Titan X / 980 ti.
 
So someone on overclockers.uk extrapolated from the FarCry 4 Ultra demo they showed at the PC Gamer show a 28% increase over Titan X performance at that 4k resolution.

Yeah it's just one game but as far as benchmarks go, it's a start.

That was posted here somewhere I think but I don't think the extrapolation means too much because it was comparing a slide from their event to a completely different benchmark from a review. It was the same game at the same resolution but under completely different circumstances. 30% improvement over the TitanX would be far beyond just about anyones expectations.
 
If someone told me that AMD was going to release a 596mm² GPU that would match (or even beat) Nvidia's newest top chip a year ago or so, I would've called bullshit. I know there was that leaked Synopsys slide stating >550mm², but still. I'm actually impressed.
 
I read on PC Gamer that Fury X is just Fury with a water cooler. But I think I recall people saying Fury X has more cores than the Fury card.

Anyone know? Has it been clarified?
 
LIstening to that AMD Roy interview. I realized he just said the GPU have 4096 bit memory bus and 4096 GPU cores. I wonder is this on purpose. Will the bus width always be the core count?
 
LIstening to that AMD Roy interview. I realized he just said the GPU have 4096 bit memory bus and 4096 GPU cores. I wonder is this on purpose. Will the bus width always be the core count?

Coincidence. As process technology scales, core count will again overtake bus width.
 
I read on PC Gamer that Fury X is just Fury with a water cooler. But I think I recall people saying Fury X has more cores than the Fury card.

Anyone know? Has it been clarified?
No official specs out for Fury (Fiji Pro).

Even the usual rumor sites left the specs blank this week.
 
That, not showing it and delaying it after the xt for 3 weeks really rubs me the wrong way.
Nvidia not talking about 980Ti after Titan X should rub the same way too.

Then again I'm not mad only because I'm interested in the $650 segment, if the tables are turned between XT and Pro I could be miffed too hehe.
 
SFR mode is a feature of mantle. Civilization does this under mantle in which for CrossFire the memory pools are not replicated. In DX mode it uses AFR.

Didn't know about Mantle supporting it, but my point was mainly that this isn't a hardware feature that will just work with AMD cards. It's going to take specific coding. I really wouldn't buy a card from either side with the expectation that it will be available for all or even very many games.
 

That Sapphire model for $330 should be a good upgrade from my Radeon 7850 eh? The Witcher is making my poor card cry so I'm in the market for a new gpu by the time Windows 10 is released. The 300 price point is my sweet spot but the card is out of stock. I'm hoping for more deals and maybe a free game for this or a similar card.
 
If the Fury Nano isn´t much faster and cheaper then the GTX 970 Mini I will buy the Nvidia Card.

Are there any big advantages for the Nano over a GTX 970?

Maybe there will be a pricedrop soon?
 
If the Fury Nano isn´t much faster and cheaper then the GTX 970 Mini I will buy the Nvidia Card.

Are there any big advantes für the Nano over a GTX 970?

Maybe there will be a pricedrop soon?

Performance?

I'm wondering if I'm the only one in these threads who are still waiting for benchmarks before jumping to conclusions based on cards sizes, TDP, prices, etc.
 
If the Fury Nano isn´t much faster and cheaper then the GTX 970 Mini I will buy the Nvidia Card.

Are there any big advantages for the Nano over a GTX 970?

Maybe there will be a pricedrop soon?

Some info on the Nano here:

http://wccftech.com/amd-radeon-r9-n...ngle-8pin-connector-sff-design-faster-hawaii/

Apparently it's faster than a 290X, which is impressive from AMD's engineers as it has a 175W TDP. This card is looking like a really smart move as I reckon they are going to release it for $499 or under with performance ahead of a 980 as it's basically a cut-down Fiji GPU.
 
That was posted here somewhere I think but I don't think the extrapolation means too much because it was comparing a slide from their event to a completely different benchmark from a review. It was the same game at the same resolution but under completely different circumstances. 30% improvement over the TitanX would be far beyond just about anyones expectations.

Didn't that game favor AMD GPUs?

http--www.gamegpu.ru-images-stories-Test_GPU-Action-Far_Cry_4-nv-test-FarCry4_3840.jpg


This benchmark had the 290X nearly matching the 980, so I wouldn't be too surprised if the Fury X did better than the Titan X / 980 ti.

Oh absolutely. FarCry 4 works well on AMD hardware AND it is only 1 benchmark. I'm quite sure AMD picked this one to show off specifically because it favors their position.

I fully expect the Titan X and the 980Ti to beat the Fury X in some games, at certain resolutions and when the Nvidia specific features are turned on.

With all of that said if the Fury X is neck and neck with the Nvidia offerings in most benchmarks while being nearly silent then it will most likely be my next card.
 
Top Bottom