AMD are rumoured to have only 16,000 RX Vega GPUs available at launch

Well according to Don Woligroski from AMD in an AMA session on Toms Hardware -

Vega performance compared to the Geforce GTX 1080 Ti and the Titan Xp looks really nice.

So either he's vastly over exaggerating, He means for the price point it looks nice compared to those 2 or he's telling it like it is.

We already covered this, man.

It's incredibly vague. That is the hype train...

BTW I have been thinking it over and basically it either went down one of two ways.

1. Polaris launched, and with the technical data (cores, die size, clocks etc) it was established that Vega (at least the one AMD had shown) would be around as fast as a 1070.

2. AMD may have released a rough outlay of the core details at some point.

I am heavily leaning toward *1*. The guys that do all of these calculations really know their crap. It's kinda like when Nvidia released the spec of the Titan XP (not Xp) people did all of the calculations compared to the 1080 and said it would be around 10-20% faster.

Which of course was absolutely spot on.
 
From day one that is what it was. Unless there are more than just that one then that is what it is. It's hype that has slowly built it up to being something more than just a 1070 competitor.

However, on paper as penned out by AMD that is what Vega basically equates to.

Only time will tell on this one.

Only releasing a 1070 competing GPU makes zero sense. I'm sorry, but that makes no sense for a company to do that, and I don't remember ever hearing or seeing evidence to point to it. The GTX 1080 is market is huge. The 1080Ti market, no, but the Fury X was faster than a 980. All they have to do is keep the price down and to beat 1080 and the cards will sell.

The main issues with the Fury X:
Poor DX11 support. Vega fixes this as DX12 and Vulkan are actually here now.
Low amounts of memory. Vega fixes this with 8GB and HBC.
Poor pricing. If Polaris and Ryzen are to be a trend for AMD, Vega should fix this.
Too few options. If Vega follows Polaris, this issue should be resolved.
 
Only releasing a 1070 competing GPU makes zero sense. I'm sorry, but that makes no sense for a company to do that, and I don't remember ever hearing or seeing evidence to point to it. The GTX 1080 is market is huge. The 1080Ti market, no, but the Fury X was faster than a 980. All they have to do is keep the price down and to beat 1080 and the cards will sell.

The main issues with the Fury X:
Poor DX11 support. Vega fixes this as DX12 and Vulkan are actually here now.
Low amounts of memory. Vega fixes this with 8GB and HBC.
Poor pricing. If Polaris and Ryzen are to be a trend for AMD, Vega should fix this.
Too few options. If Vega follows Polaris, this issue should be resolved.

What else can they do fella? magic up a card that's twice as fast as Pascal out of thin air?

They will get out what they have put in. And I would guess that would be a fraction of what Nvidia has invested into Maxwell/Pascal/Paxwell.

Making GPUs is not luck. You get out what you have to put in.

The Fury X made no logical sense when it launched. Absolutely none. You would have been barking to actually buy it over the 980Ti for the same price. Only true die hards bought them, or, people like me who got them cheap.

Whilst the temps were pretty good the rad is an anchor and a real PITA if you have the wrong type of case (which of course I absolutely did).

My Fury X is better now than it was at launch. Which is completely bloody silly.
 
What else can they do fella? magic up a card that's twice as fast as Pascal out of thin air?

They will get out what they have put in. And I would guess that would be a fraction of what Nvidia has invested into Maxwell/Pascal/Paxwell.

Making GPUs is not luck. You get out what you have to put in.

The Fury X made no logical sense when it launched. Absolutely none. You would have been barking to actually buy it over the 980Ti for the same price. Only true die hards bought them, or, people like me who got them cheap.

Whilst the temps were pretty good the rad is an anchor and a real PITA if you have the wrong type of case (which of course I absolutely did).

My Fury X is better now than it was at launch. Which is completely bloody silly.

I'm not talking about luck or magic pixie dust. I'm talking about a Fury X already being within 10% of a GTX 1070 in performance. How on earth after potentially two years of development and a die shrink can AMD only increase its performance by 10% or so from a card we both agreed underperformed? The Fury X was held back by poor clock speeds. 14nm fixes this as AMD's 28nm was always a challenge in that area. The Fury X was held back by low amounts of VRAM. HBM2 fixes this. That's just the stuff we know about.

By doing the math, if the clock speeds of a Fury X were 1400Mhz, a very conservative number for Vega, the Fury X would be incredibly closeto a GTX 1070 and matching or beating it in a lot of modern titles. Which means you're essentially saying the only thing that AMD have been able to do is increase its clock speeds. There were no improvements to GCN, no improvements to compression, no memory improvements (the leaked clock speed alone of 700Mhz is a big increase over the Fury X's 500Mhz), and nothing else. Is that what you're saying?

Again, I'm not talking about luck or magic unicorn excrement. It is completely illogical MATHEMATICALLY to produce a GPU that's only 10% faster than their previous flagship. And I still haven't seen any provable evidence to say that Vega is a 1070 competitor only. For goodness sake, we saw it beating a GTX 1080 almost six months ago! I don't care whether it's using Vulkan. Vulkan is a real thing. Just because it's not as widespread as DX11 does not mean it does exist or is irrelevant.
 
Vega will either compete with a 1080ti or be close to it. Either way, it's a little to late unless they can get Navi out the door Q1 and latest early Q2 next year.
 
I'm not talking about luck or magic pixie dust. I'm talking about a Fury X already being within 10% of a GTX 1070 in performance. How on earth after potentially two years of development and a die shrink can AMD only increase its performance by 10% or so from a card we both agreed underperformed? The Fury X was held back by poor clock speeds. 14nm fixes this as AMD's 28nm was always a challenge in that area. The Fury X was held back by low amounts of VRAM. HBM2 fixes this. That's just the stuff we know about.

By doing the math, if the clock speeds of a Fury X were 1400Mhz, a very conservative number for Vega, the Fury X would be incredibly closeto a GTX 1070 and matching or beating it in a lot of modern titles. Which means you're essentially saying the only thing that AMD have been able to do is increase its clock speeds. There were no improvements to GCN, no improvements to compression, no memory improvements (the leaked clock speed alone of 700Mhz is a big increase over the Fury X's 500Mhz), and nothing else. Is that what you're saying?

Again, I'm not talking about luck or magic unicorn excrement. It is completely illogical MATHEMATICALLY to produce a GPU that's only 10% faster than their previous flagship. And I still haven't seen any provable evidence to say that Vega is a 1070 competitor only. For goodness sake, we saw it beating a GTX 1080 almost six months ago! I don't care whether it's using Vulkan. Vulkan is a real thing. Just because it's not as widespread as DX11 does not mean it does exist or is irrelevant.

I completely agree. However, that is what the rumours are. Time will tell I guess. Only five weeks or so :)
 
Vega will either compete with a 1080ti or be close to it. Either way, it's a little to late unless they can get Navi out the door Q1 and latest early Q2 next year.

2018 was the initial strategy for Navi.

I completely agree. However, that is what the rumours are. Time will tell I guess. Only five weeks or so :)

Yeah, five weeks and we'll know. I liked the rumour Dicehunter posted in the Quick News thread. 400$ for a 1070 competitor, $500 for a 1080 competitor, and $600 for a 1080ti competitor with the reveal at the beginning of June. That would be very nice indeed.
 
$400 is still too much in this day and age for 1070 performance IMO. I bought my Titan XM about two months before the 1070 came out for sale (just after the announcement of the impending 1070 and 1080) for £440. Here I am a year later, really wanting to upgrade and I can pretty much get the same performance for the same money.

Two years ago (t'was June '15) I bought my Fury X. I also paid £440 for that (he wanted £420 but mentioned a charity so I stuck £20 into that). And here I am, almost two years to the dot and what can I get for £420? the same s**t.

That's the part that has put me off big time. That is why I still have a Fury X and TXM. Because nothing for around that price even two years later offers me anything more than a bit more VRAM which TBH these days I really couldn't give a toss about. When I installed RE7 I just left it to its own devices, that's how bothered I am about graphics these days....

Nvidia said that the 1080 would drop to $499 IIRC. It came down to about £450 for a few days, then soon rocketed back to over 500 notes. For a mid range GPU.

Yeah, screw them apples. Knowing my luck I would drop £500 on a card and they'll announce Volta in June (Nvidia hate other people getting any thunder so they always do SOMETHING to mess it up. Even Ryzen launch they were making fart noises).

That's the problem for me. I have been priced out of my own passion, my own hobby. So now it's less of a hobby for me. Nice one /roll eyes.
 
A 1080 is not as mid range GPU. In the lineup of overpriced Nvidia cards yes, but in the market it is not. Mid range is less than $300 or whatever currency you use.
 
A 1080 is not as mid range GPU. In the lineup of overpriced Nvidia cards yes, but in the market it is not. Mid range is less than $300 or whatever currency you use.

Everything but the price is mid range.

$0 - nothing.

$500+ 1080

$699 - 1080ti

$1300 - Titan Xp.

In fact no, even the price is mid range, if you consider Nvidia's stupid pricing.

But for the performance it is still way too much. Things are totally stagnant now like they were in CPU land before Ryzen launched.
 
To me:

GTX 1030 = budget
GTX 1050 = low-end
GTX 1050Ti = low to midrange
GTX 1060 = midrange
GTX 1070 = mid to high-end
GTX 1080 = high-end
GTX 1080Ti & Titan = flagship

Yes, the 1080 is not a huge chip, but its performance and price suggest high-end. It reigned as the fasted single GPU for quite a few months and was price very, very high. I can't call that a midrange GPU.
 
To me:

GTX 1030 = budget
GTX 1050 = low-end
GTX 1050Ti = low to midrange
GTX 1060 = midrange
GTX 1070 = mid to high-end
GTX 1080 = high-end
GTX 1080Ti & Titan = flagship

Yes, the 1080 is not a huge chip, but its performance and price suggest high-end. It reigned as the fasted single GPU for quite a few months and was price very, very high. I can't call that a midrange GPU.

Remove the word "Flagship" and you get to where I was. I don't consider the 1080Ti and TXp as flagship GPUs, just high end ones. Nvidia have added the BS belmet moniker to crank up the price.

I also did not consider the 1080 a flagship before the 1080Ti launched. I just considered it for what it really was, mid range Pascal with the higher end cards to launch at some point.

It's mostly the same reason why I do not have a 1080, because the price difference between it and the 980 was a joke. And I knew it wasn't the true big Pascal so I didn't fancy dropping several hundred notes into it only to be slapped in the face by Nvidia a few months later.

And those people who did were slapped hard. First came Titan XP, then 1080Ti, and then just as TXP owners were trying to convince themselves they hadn't been screwed Nvidia make sure by launching the TXp.
 
I agree with Goldfish most of the way. I don't feel Titan should be in the equation.

I my eyes Titans are not for gaming. It's more a budget production card.
In the gaming world, it's just to showoff money.

@Alien.
If they aren't flagships, then what is?
 
Yeah, you're right about the Titan. I just put it in there 'cause it performs the same as a 1080Ti. It's not necessary to include it. But I definitely consider the Ti to be the 'flagship'. It is the best Nvidia can currently offer for 'reasonable' gamers and is therefore the flagship.

Nvidia are not the only company to have that title granted them. In the guitar world it's a description used all the time by the manufacturer themselves. Amplifier builders will use their original or favourite design as their foundation, their flagship that all other designs will be based on. It is usually the best they can offer for the widest audience. The Titan could be considered the flagship, but its high price puts it in a different league altogether.

The GTX 1080 I didn't consider to be the flagship because I knew it wasn't the best Nvidia could do with Pascal. It was one once the most powerful, but it was never their flagship in my eyes. The 480, 580, 680, they were the flagships of their respective architectures, but once Nvidia released the 780Ti, things changed. Instead of pushing the 780 back, it added another GPU on the end.
 
@Alien.
If they aren't flagships, then what is?

Just high end silicon?

All of this nonsense has been created by Nvidia. Who would pay £1300 for a high end GPU?

Who would pay for a TITAN OMG FLAGSHIP INNIT GPU?

What Nvidia have done is realise that 99% of hardware enthusiasts are just braggarts and have capitalised on it very nicely indeed.

When I look around me I see how many "true" enthusiasts that I know. Know what? none of them from 2009 are here apart from the staff.

Most come along, show off and then get bored.

I just call a spade a spade is all really.
 
I think you are putting too much into the term Flagship. It just mean the leader of the pack ;) i.e. the top of the highend.
And i agree, 1300£ is waaaay to much. I'm a gamer, and i like nice graphics.
I bought a 2560x1440 monitor, and the 1070 wasn't always enough for Ultra settings 60fps. So i went for a 1080ti. So in my mind, there is a need for better graphics cards.

And when the competitor is nowhere to be seen, then Nvidia can name the price.

I don't see myself throwing money down the drain. I always try to make "smart investments" :D
I went 560ti - 670- 780ti (2,5 years) - 1070 (75days) - 1080ti
But "only" 2500k - 4670k - 4790k
And the i7 is bought used, only because i found one cheap (70£ upgrade cost) otherwise i would just have kept the i5, it's still more than capable in games.
 
Back
Top