Amd Fury X, Will 4Gb Of Hbm Bottleneck Ultra 4K Without Aa?
#1
Posted 21 June 2015 - 11:39 PM
its creepcamper again! incase i havent know me i like to spamchat in game with random facts and statements.
anyways right now i am rocking a clevo with 3.4 ghz max boost i7 and a labtop equalivent of HD 7870. I recently got a 4k 32 inch ips benq monitor and i am not happy about the lag from scaling up 1080p ( i tried all low for native 4k, looks so clear but framerate is 20-30 ish in combat and thats not cool.)
im thinking of getting a Fury X because lets face it, i dont want a " soon to be kepler" maxwell card that is gonna get nerfed when new gen comes out. and fury X has so much potential at the same pricepoint. maybe it is just me but i think fury x is more futureproof
question is, if i want 1 card 4k gaming ( turning aa off cause i dont need that in 4k) would the 4 gb limit framerate for mwo in 4k?)
(my mind says wait for skylake and pascal/ HBM2 AMD card, my soul says 4k now.......)
cheers mechwarriors, please dont use the inexpensivecampervan joke when you run in to me in pug games!
#2
Posted 22 June 2015 - 02:08 AM
In actual games this behavior doesn't seem to show up. A single Fury X was demoed playing both Dirt Rally and Sniper Elite III not just at 4k, but on triple monitor 4k, at 60fps. So put simply, the card does not bottleneck at high resolutions. By the time anything is capable of bottlenecking the memory system, it's already hilariously beyond the ability of the card to render in the first place.
I know that I'm certainly putting down for one the instant they go on sale.
#3
Posted 22 June 2015 - 02:53 AM
Gibbo states that the performance at stock at 4K is exceptional considering it's a single card.
at standard resolutions tho, it's slower than the G1 980TI - Which is odd, perhaps because it's not leveraging all of that HBM bandwidth.
It's voltages are locked, so very very limited overclocking initially but they should release these controls in time.
No overclocking the HBM though - which really shouldn't mean much.
Early suggestion is that 4Gb of HBM is equivalent of 6GB of GDDR5 because of the much higher bandwidth. So it should be fine.
However personally this is first gen, i would wait for the die shrink of pascal and arctic islands and HBM2.
Catamount, on 22 June 2015 - 02:08 AM, said:
I wouldn't.
Initial supply is very very limited, and prices will be inflated. Give it a month or two to avoid paying a premium just to be first.
Edited by DV McKenna, 22 June 2015 - 03:04 AM.
#4
Posted 22 June 2015 - 01:15 PM
Come Wednesday there's one price I intend to pay, and that's $649. If the stores have it early in the morning, then I'll have it shipping before most people are even awake. I do agree about the refreshes; it's entirely possible I'll be selling this card in a year for an upgrade, but I also have the budgets these days to do something like that. Computer parts are a relatively trivial expense. Would I jump on the Fury X bandwagon if the card had to be a long term purchase? You know, honestly I think I would, retail price holding. As long as the use is really there for an upgrade (I have a 7970 and I want an Asus MG279Q ASAP, maybe this week, for 90hz freesync, so for me it is), it really is a good deal and a solid jump up in performance.
The locked voltage does give me pause. The whole reason to buy a ~275W card with a 500W liquid AIO is to overclock the living **** out of the card. I don't care about the HBM, because the bandwidth is already absurd (OCing would be superfluous), but I want that core overclock :/ OTOH, that, too, is little more than a rumor coming from Gibbo at OCUK running his mouth with wild assumptions, which he likes to do (center of attention thing with him, and nothing new). There are, for the moment, no supplied drives that enable overvolting, at least in CCC, but that's not evidence that the VRMs don't support it, or that the retail product won't support it, even immediately upon release. Most of what we're getting is just wild speculation by Gibbo, in the form of "well if AMD is going to unlock them, why don't they now and get more impressive reviews?". Funny how no one's reviewing them yet.
In short, there is, at present, no information that really indicates anything about voltage, though if they're really locked I fully intend to just buy from whatever board partner fixes it first. That said, I can't imagine AMD is voltage locking a card that they've overbuilt both the cooling and power delivery to massively. That just makes no sense, whatsoever.
We'll see how things turn out.
Edited by Catamount, 22 June 2015 - 01:39 PM.
#5
Posted 22 June 2015 - 01:22 PM
Without voltage control you can't really overclock much, when reviews hit there will be no way of knowhing how much overclocking headroom is there. Too bad.
Edited by Flapdrol, 22 June 2015 - 01:23 PM.
#6
Posted 22 June 2015 - 01:42 PM
Catamount, on 22 June 2015 - 01:15 PM, said:
Come Wednesday there's one price I intend to pay, and that's $649. If the stores have it early in the morning, then I'll have it shipping before most people are even awake. I do agree about the refreshes; it's entirely possible I'll be selling this card in a year for an upgrade, but I also have the budgets these days to do something like that. Computer parts are a relatively trivial expense. Would I jump on the Fury X bandwagon if the card had to be a long term purchase? You know, honestly I think I would, retail price holding. As long as the use is really there for an upgrade (I have a 7970 and I want an Asus MG279Q ASAP, maybe this week, for 90hz freesync, so for me it is), it really is a good deal and a solid jump up in performance.
The locked voltage does give me pause. The whole reason to buy a ~275W card with a 500W liquid AIO is to overclock the living **** out of the card. I don't care about the HBM, because the bandwidth is already absurd (OCing would be superfluous), but I want that core overclock :/ OTOH, that, too, is little more than a rumor coming from Gibbo at OCUK running his mouth with wild assumptions, which he likes to do (center of attention thing with him, and nothing new). There are, for the moment, no supplied drives that enable overvolting, at least in CCC, but that's not evidence that the VRMs don't support it, or that the retail product won't support it, even immediately upon release. Most of what we're getting is just wild speculation ny Gibbo, in the form of "well if AMD is going to unlock them, why don't they now and get more impressive reviews?". Funny how no one's reviewing them yet.
In short, there is, at present, no information that really indicates anything about voltage, though if they're really locked I fully intend to just buy from whatever board partner fixes it first. That said, I can't imagine AMD is voltage locking a card that they've overbuilt both the cooling and power delivery to massively. That just makes no sense, whatsoever.
We'll see how things turn out.
Thing is WCCFTech isn't reliable. At all.
OCUK have already stated they are only being given 100 Fury X's in their initial batch despite ordering more. AMD can't supply more.
And this is a company (OCUK ) that is selling 500 980TIs at £499 purely for marketing; actually making a retail loss of £3 per card.
I sincerely doubt that the small supply of Fury X is a rumor. AMD have always had quantity supply issues out of the gate; just the same as Nvidia. Gibbo's story is leant further weight as his suggestion limited supply would see prices skyrocket by NCIXs price.
http://m.ncixus.com/...cts/sku/110441/
Way above the RRP.at $849
Edited by DV McKenna, 22 June 2015 - 01:43 PM.
#7
Posted 22 June 2015 - 03:16 PM
I know what OCUK is claiming, but frankly, I put no stock in anything Gibbo says. The entire message has consistently been "hey guys, the Fury X totally sucks, but oh hey, lucky for you, we just happen to have a really special 980 TI deal that you should totally buy right now instead. I promise these two statements have nothing to do with each other!". That's not to say he's wrong, just lacking, in my opinion, in any credibility or hint of objectivity, whatsoever, and wildly prone to mixing fact with outlandish speculation.
I guess we really will know in two days.
#8
Posted 22 June 2015 - 03:47 PM
#9
Posted 22 June 2015 - 10:09 PM
Catamount, on 22 June 2015 - 03:16 PM, said:
I know what OCUK is claiming, but frankly, I put no stock in anything Gibbo says. The entire message has consistently been "hey guys, the Fury X totally sucks, but oh hey, lucky for you, we just happen to have a really special 980 TI deal that you should totally buy right now instead. I promise these two statements have nothing to do with each other!". That's not to say he's wrong, just lacking, in my opinion, in any credibility or hint of objectivity, whatsoever, and wildly prone to mixing fact with outlandish speculation.
I guess we really will know in two days.
Well we will see but I don't think that's a typo. Not even close.
And you might want to catch up on the OCUK thread because the message last I looked yesterday was most defiantly not "oh hey guys the Fury X sucks " quite the opposite.
#10
Posted 22 June 2015 - 11:24 PM
DV McKenna, on 22 June 2015 - 10:09 PM, said:
So Canada is getting the Fury X for 78% of the US price? How does that make sense? I guess I can always just buy one there.
Quote
Of course it is. No, no one denies the card is powerful because no one can. We have too many numbers. Gibbo is just giving every attempt at negative speculation possible, bending overbackwards in every way possible to do so. His comments about the voltage locking are a classic example. Again, I suppose this has nothing at all with the company's 980TI promotion.
Edited by Catamount, 22 June 2015 - 11:27 PM.
#11
Posted 22 June 2015 - 11:27 PM
I love the title mishap, but while this doesn't give any solid information about performance, it does suggest that, again, the memory is more than adequate.
#12
Posted 23 June 2015 - 01:10 AM
Catamount, on 22 June 2015 - 11:24 PM, said:
So Canada is getting the Fury X for 78% of the US price? How does that make sense? I guess I can always just buy one there.
Of course it is. No, no one denies the card is powerful because no one can. We have too many numbers. Gibbo is just giving every attempt at negative speculation possible, bending overbackwards in every way possible to do so. His comments about the voltage locking are a classic example. Again, I suppose this has nothing at all with the company's 980TI promotion.
He also said that will change as drivers are revised and come out. Just the same as 8 pack (who will do OCUKs Overclocked Fury Xs)
#13
Posted 23 June 2015 - 11:16 AM
DV McKenna, on 23 June 2015 - 01:10 AM, said:
Well, again, we all know that. He just made it sound like that would take as long as he could possibly assume without flying right in the face of facts
I do take the stock rumor seriously, of course. I'm not entirely convinced that there's any shortage. The evidence seems scant. Does that stop me from having an alarm set for 12am PST (3 my time)? Nope, not at all. I'll check midnight our time, too, but most agree midnight Pacific time is the likely release point.
Of course, all of this depends on my damn bank cooperating and not continuing to hold work checks for no reason :/
Edited by Catamount, 23 June 2015 - 12:33 PM.
#14
Posted 24 June 2015 - 10:08 AM
It's a good card, but NVidia's chips aren't exactly new
Would've wished for a landslide, not on par
Hope AMD has a bigger margin on those, God knows they need it
The 980ti can be overclocked quite nicely compared to fury as it seems
At that high price point I don't care any more about a few bucks more for a 980ti if it overclocks so much nicer
DX12 might change things a bit again
Ehh
In any case I would say here the same thing I said to a team mate
If you're happy with you're card now, then wait until you ACTUALLY need a new one
Occulus is still quite some time away
And until Fallout 4 is here streetprices will have settled in
As to 4GB of VRAM
AMD seems to have found a way to compress data quite some
So it doesn't seem to matter
However I think AMD and NVidia are bringing HBM with 8GB next year
So again
If you can wait, then do
As a small reminder
AMD still seems to have problems with stutters, that can be adresses by newer drivers, but the last WHQL is 6 months old
They need to step up they're game in the driver section too
#15
Posted 24 June 2015 - 10:22 AM
#16
Posted 24 June 2015 - 10:24 AM
xWiredx, on 24 June 2015 - 10:22 AM, said:
Should see the AMD <3 club on the Star Citizen forums.
HardOCCP's review has pretty much been declared invalid because its out of line with other reviews...........................
#17
Posted 24 June 2015 - 11:51 AM
DV McKenna, on 24 June 2015 - 10:24 AM, said:
Should see the AMD <3 club on the Star Citizen forums.
HardOCCP's review has pretty much been declared invalid because its out of line with other reviews...........................
PS Perspective, Hot Hardware, Tech Report, and Legit Reviews seem to have pretty similar results to HardOCP. Not that I really frequent any of those, though. I don't see results up on Anandtech yet, but I'm willing to bet they'll be similar, too.
The bottom line here is that the new Titan is the only satisfactory card for hardcore 4K gaming, and the Radeon Fury X only keeps up or slightly lags behind the 980 Ti when it comes to 1440p. I have no doubt the Fury X is probably a monster as soon as you remove the VRAM limitation, but review sites are hitting on what AMD touted the card for in the first place: 4K gaming. If you are looking for a top-tier card and you can get one for less than a 980 Ti, do it. If it's the same price or more expensive, might as well get the 980 Ti.
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users