Vega Rumors

[Gideon]

In some titles yes, in the majority of titles Ryzen 7 is considerably behind 7700K, especially games that are single threaded bound, and there alot of those around. AMD knows that very well, that's why they used 7700K when benchmarking Vega to put it in the best light possible. So yeah Ryzen 7 is still not the best choice for gamers.

At what point did I mention the 7700K, that you felt the need to run in here and mention it? All I said that Ryzen was improved since launch and likely the cause of the hope that Vega might receive the same since it has some new features that would have to be added by developers. At 1440p and above there is little to no difference between a Ryzen and 7700K and in fact sometimes Ryzen has slightly better frame rates in 4K, does it mater... nope. Unless your a benchmark queen no one gives a shit that your only getting 105 fps instead of 115 fps in a game, it's well beyond what you need anyway. If Ryzen sucked at 4K then I could see the point about saying Ryzen is no good at gaming but that is not true, also I use my computer for more then just gaming and like the extra cores I got for the price I paid. So stop stomping around acting like everything AMD does is a fail, both sides on the gpu front have had their fails and so has Intel, no one bats 1.000.

 

[exdriver]

It's in AMDs marketing slides. They provided numbers for ten top titles with DSBR enabled. There's one Specview benchmark where that feature provided over 100% gains. These are conservative estimates so I wouldn't call them incredible. The only difference is that AMD hasn't committed to when all the features and performance will be there.

I'm betting there's a good reason for that. No need to hide good results.

Could be in a week, month, or year. I tend to think more in the months range. In another year there will be a refresh of products likely with improved titles to play. The people with a Vega wouldn't be stepping back either as they would already have hardware to run the titles.

I doubt anyone's buying today's Vega in 12 months. A refresh would be wise but let's continue as the next part comes around to that.

It doesn't seem unreasonable to see Vega ahead if Titan at that time, nor is it unreasonable to expect a hardware refresh to compete with Volta. Just look at what the 580 did. Faster HBM, core clocks, etc just from process improvements. No reason there isn't a bigger Vega to compete with Volta. AMD has 10, 11, 12, and 20 on the books already.

What Vega product would be faster than the Titan or TI? When and how would this occur?

I don't find the Titan to be an attractive product at $1200.

Since the previous question covers 10, what product from 11, 12 or 20 would be faster than current NV offerings at the same power and price point... or just simply faster at all?

"Concrete" is the current problem. You're looking at a product that with a refresh in six months and zero hardware changes could claim significant performance increases. The problem is there is no time frame for those changes, or guarantee, but we've seen nothing from AMD indicating the hardware is seriously flawed. Just that they need time to improve drivers.

I'm not buying promises from AMD on GCN. Concrete is always the problem for AMD. H has shown the track record of 5% from driver improvement. If you can provide better data do so.

As for power, I'd wait for reviews of RX with the voltages tuned and power saving features fully enabled. So far all anyone has done is benchmark the worst case scenario with max voltages.

There comes a point that even when you support a product you should acknowledge its weaknesses. I understand that you are committed to showing Vega in a positive light, but this aspect of the product does not shine.

 

[Ieldra]

It's in AMDs marketing slides. They provided numbers for ten top titles with DSBR enabled. There's one Specview benchmark where that feature provided over 100% gains. These are conservative estimates so I wouldn't call them incredible. The only difference is that AMD hasn't committed to when all the features and performance will be there. Could be in a week, month, or year. I tend to think more in the months range.

DSBR was already active in the FE for pro applications, and it still didn't compare favorably. DSBR is also active in the RX Vega benchmarks performed by AMD whose data was presented in their slides.

Also, AMD marketing statements regarding performance are conservative? Who are you kidding? 45 % perf gains from async... Dual RX480 being more powerful and more efficient than a 1080... The list goes on.

I don't know who you're kidding but nobody with any common sense will believe what you're saying

 

[FrameBuffer]

DSBR was already active in the FE for pro applications, and it still didn't compare favorably. DSBR is also active in the RX Vega benchmarks performed by AMD whose data was presented in their slides.

Also, AMD marketing statements regarding performance are conservative? Who are you kidding? 45 % perf gains from async... Dual RX480 being more powerful and more efficient than a 1080... The list goes on.

I don't know who you're kidding but nobody with any common sense will believe what you're saying

From what I was told (by person with vega in hand) that the numbers AMD presented (using Vega 64 Air) did NOT have DSBR present in the drivers, ie the numbers presented do NOT contain DSBR enhancements, (outside of the Pro suite, running s different fork). The only numbers with DSBR are the SpecView, showing (at the time) the best outlook for DSBR.

 

[Stryke1983]

Poster makes outrageous performance claims about upcoming product based solely on hope.

Rumors go against the outrageous claims, poster continues to insist they're correct.

Product's manufacturer announces product with performance that approximately matches rumors and directly contradicts poster's claims.

Poster continues to insist that everyone else, including the actual manufacturer, is wrong and that their outrageous claims will be proven correct at some unsubstantiated point in the future via various unsubstantiated methods that will change as they are disproven one by one.

Looks like just another day in the video card forum.

 

[razor1]

You have over 200+ posts, (just in this thread alone) claiming just the opposite. And that Vega is a flop, etc.

You really are a special kind of person.

Still talking about posts, guess it blows your mind that you can't understand what the reality is.......

And ya I am special always have been

BTW if you ever meet John Stewart in person, he will blow your mind away, he is one of the smartest and wittiest guys I have ever met.

Did you see my post on fixing the latency. Don't want to talk about that? Then why do you bring up conversations you don't want to talk about (you stated you wanted to talk about how they can do it lol)? For what? For a magic unicorn that shits rainbows?

Shit AMD even said they aren't looking at transparancy for mpgu, how do you go from that to "lets talk about this rumor and believe its real" If you want to discuss, analyze the problem and propose how to fix it.

You are 50 years old, you don't know how to problem solve. First step is to identify the problem. Thensplit the problem up based on processes. Then propose changes on those processes.

You seem to want to skip all that, and think infinity fabric will solve it all. No it won't lol, do you know what people believe stuff like that? People that don't know anything about GPU's and how they function. Creating GPU's and programming them is all math, everything is known before work is done, nothing magically happens that is out side of what is planned for. Yeah you can have some variations up or down, but for the most part the targets, features and what not are well know well before the chips even go into manufacturing.

 

[Factum]

#waitforvega has now turned into #waitforFP16patches...

 

[razor1]

#waitforvega has now turned into #waitforFP16patches...

then #waitforVega20
then #waitforVega20drivers
then #waitforNavi

AMD only buyers should just not buy any graphics ever from now on.

 

[Captain Newmackwa]

Wouldn't this mean that NVIDIA has confirmed that they are no longer worried about competition anymore (at least to a degree that it doesn't factor into their product planning)? I.E. they used to react to AMD but now they don't have to.

That said, unlike Intel, NVIDIA has been reasonable keeping up a decent development pace despite a lack of competition. NVIDIA seems to be in good shape compared to Intel who are now being forced to hastily react to Ryzen and Threadripper which resulted in a bit of a rushed rollout of x299.

I can see that Nvidia is saying that it's competing more with itself than AMD. They have to offer us new products that compel us to upgrade in order to sustain their revenue growth. As for pricing without competition though, that's a whole other story.

 

[razor1]

I can see that Nvidia is saying that it's competing more with itself than AMD. They have to offer us new products that compel us to upgrade in order to sustain their revenue growth. As for pricing without competition though, that's a whole other story.

Pricing still has to stay reasonable, not saying they won't increase, they might and probably will, but lets say 50 bucks a bracket? Yeah I can see them doing that. Just like how Intel increased prices for their CPU's. A little for every generation, but after time, it really adds up.

 

[exdriver]

I can see that Nvidia is saying that it's competing more with itself than AMD.

Maybe they say that, but I believe they are just committed to the path they are on and there really isn't much room for them to change course at this stage.

 

[Captain Newmackwa]

Maybe they say that, but I believe they are just committed to the path they are on and there really isn't much room for them to change course at this stage.

Agreed. One thing is for sure, it did sound like Nvidia isn't threatened at all by AMD-RTG.

 

[NKD]

#waitforvega has now turned into #waitforFP16patches...

What is wrong with talking about fp16? 2 games are already supporting it. No one is giving false statement here. It's a fact. And no vega is not great and no one should buy Vega on future promises. But fp16 is not a myth I was surprised developers are actually using it already.

 

[Chimpee]

What is wrong with talking about fp16? 2 games are already supporting it. No one is giving false statement here. It's a fact. And no vega is not great and no one should buy Vega on future promises. But fp16 is not a myth I was surprised developers are actually using it already.

Isn't it just a report from Golem.de? Don't get me wrong, I would like to use how FP16 will be use but since it isn't confirm by the developers, shouldn't we treat that as rumor for the time being?

 

[RPGWiZaRD]

I was really disapppointed in seeing the prices, was hoping for the last decent option Vega could have, to create a performance vs bang-for-buck buzz but then they seem to price in-line according to Nvidia products, no price wars are happening, Nvidia is thanking AMD and are happy they can continue milking their customers like before and we customers are the one losing for having to pay just more and more every generation.

 

[griff30]

then #waitforVega20
then #waitforVega20drivers
then #waitforNavi

AMD only buyers should just not buy any graphics ever from now on.

Shill for Nvidia much?

 

[TaintedSquirrel]

#waitforvega has now turned into #waitforFP16patches...

I'll pass on all of this after the async hysteria 2 years ago. At this point it sounds like the community grasps at straws when AMD launches a failed product (async = Fury X, FP16 = Vega). Damage control.
Vega is so far behind, even if every game started using FP16 from here on out it would still be a flop.

Shill for Nvidia much?

He's not wrong. "AMD FineWine" is a meme they created themselves and it's centered around waiting for shit to improve. They like to celebrate when it finally materializes but they hate when it doesn't, and they hate to acknowledge the fact that it took years to get there.

 

[RPGWiZaRD]

Why would I want to wait around for improvements coming in an unforseeable future when I'm an enthusiast who would buy new stuff as soon as something more interesting comes around. When I buy something new I buy it based on today, what good is there to pay extra for for example 16 core CPU if 75%+ of cases the reasonable utilization ends at like 6~8 cores, in 2-3 years when hopefully core utilization at least sliiiiightly bit has improved there's faster 16 core options out there for much less.

 

[Factum]

Shill for Nvidia much?

When the truth is flamebait....you know facts have left the debate...

 

[SixFootDuo]

Pricing just seems terrible to me.

I think it would be funnier than hell if nVidia dropped the 1080 ti down to $599 LOL ............ god I hope they do this. Hell even $649 would be funnier than hell.

 

[Factum]

What is wrong with talking about fp16? 2 games are already supporting it. No one is giving false statement here. It's a fact. And no vega is not great and no one should buy Vega on future promises. But fp16 is not a myth I was surprised developers are actually using it already.

Nothing is wrong persay.
It's just the modus operandi in forums the last copuple of years.
Wait for X.
How many failed wait does it take before facts become apparent?

/Que #WaitForNavi

 

[noko]

then #waitforVega20
then #waitforVega20drivers
then #waitforNavi

AMD only buyers should just not buy any graphics ever from now on.

Na, I'll just wait for Nano Vega

The only reason this thread is 88 pages long depending upon settings is because AMD has not release the damn cards!

Well, I have to say that AMD extra DX 12 features, more then anyone else I do believe, will at least help developers explore those new features. That should be a plus for gamers regardless of what graphics card you have now - future cards, all of them should have the extra DX 12 levels at some point.

Fast recap:

• some here, no names argued and argued Vega would be no faster then 1200mhz - hehe looks like 1700mhz is the upper limit -
• some here, no names argued that performance will be slower or at 1070 levels, now we are talking 1080 or greater - I think it will be faster after several months
• some here, no names argued that HBM2 will be no faster then 400mb/s - reality it is around 484gb/s which is slower then expected 512gb/s
• Some here thought it would be out in May - I was one of those - boy was I WRONG! Now AUGUST!
• Some speculated it would be $800 or it could not be manufactured or would take a lost - starts at $399

Remember this is a rumor speculation thread and it should have been long ago closed/ignored etc. because a owner/official thread was started. At this point Vega is getting rathe boring - except for the Nano Vega.

 

[NKD]

Nothing is wrong persay.
It's just the modus operandi in forums the last copuple of years.
Wait for X.
How many failed wait does it take before facts become apparent?

/Que #WaitForNavi

Thats the point. I don't think anyone really said wait for fp16. Few people made posts and commented about it. There is little doubt it will boost peformance. Nothing wrong with that conversation.

 

[chenw]

• Some speculated it would be $800 or it could not be manufactured or would take a lost - starts at $399

Guilty as charged, I was one of those who mathed out the $800 cost to be feasible, though I didn't firmly believe it would be that high, it wasn't mathematically impossible given the information we had (about AMD's claims of price difference between their two blind-test setups), and after some google-fu, I came up with not at all pleasant numbers, as the gap between the monitors appeared to be LARGER than the price difference they were quoting.

Still, wrong in the better way though, at least it's not ridiculous levels.

 

[Boil]

AMD only buyers should just not buy any graphics ever from now on.

That is right, the forthcoming Ryzen/Vega APU will be more powerful than a 10-core Skylake-X & GTX 1080Ti combined...! ;^p

 

[Captain Newmackwa]

I was really disapppointed in seeing the prices, was hoping for the last decent option Vega could have, to create a performance vs bang-for-buck buzz but then they seem to price in-line according to Nvidia products, no price wars are happening, Nvidia is thanking AMD and are happy they can continue milking their customers like before and we customers are the one losing for having to pay just more and more every generation.

Neither AMD nor Nvidia would want to enter into a price war and would like to avoid it as much as possible unless it's intentionally a part of a strategy for market penetration and long-term profitability. Racing to the bottom may be good for us consumers but it's not good for business and these to companies are businesses with costs, expenses, employees, investors and competitive survival to think about.

 

[NKD]

I'll pass on all of this after the async hysteria 2 years ago. At this point it sounds like the community grasps at straws when AMD launches a failed product (async = Fury X, FP16 = Vega). Damage control.
Vega is so far behind, even if every game started using FP16 from here on out it would still be a flop.


He's not wrong. "AMD FineWine" is a meme they created themselves and it's centered around waiting for shit to improve. They like to celebrate when it finally materializes but they hate when it doesn't, and they hate to acknowledge the fact that it took years to get there.

look vega sucks. But asynch and fp16 both increase performance. Why would you act just like amd fanboys and try to knock down facts. Bullshit is bullshit, facts are facts. Stick to those and you won't sound like the same people you are trying to criticize.

 

[NKD]

Neither AMD nor Nvidia would want to enter into a price war and would like to avoid it as much as possible unless it's intentionally a part of a strategy for market penetration and long-term profitability. Racing to the bottom may be good for us consumers but it's not good for business and these to companies are businesses with costs, expenses, employees, investors and competitive survival to think about.

Spot on. AMD probably didn't price vega 50 bucks cheaper simply because they don't want to start a price war. They know they can't afford to make less profit on vega. So yea they may not sell more than 1080 but they would rather make that extra 50 bucks when they do sell them. Thats the whole point here.

 

[noko]

Fine Wine was more like the 7970 which did see some rather good improvements, next best I would say was the Rx480 which was around 10% if I am not mistaken. Fiji I don't think saw that much in the end. Really it was more in certain games where there were big improvements where others was little to none. Meaning even if 5% it could be huge in big titles and zero in a number of titles that may already did well in anyways. Nvidia had some fine wine as well in certain titles, such as BF4 and even Doom

 

[noko]

Spot on. AMD probably didn't price vega 50 bucks cheaper simply because they don't want to start a price war. They know they can't afford to make less profit on vega. So yea they may not sell more than 1080 but they would rather make that extra 50 bucks when they do sell them. Thats the whole point here.

AMD will most likely sell everything they have and can make first two months so no need to stir anything up. Prices can be reduced later if needed. Still $399 for a 8gb HBM2 card is actually good - just might not give you much over the DDR5 card is the sad part. As for Nvidia, they may want to clear inventory towards the end of the year, good time for it in preparation for Volta. Still think Volta will be 2nd quarter launch next year ~ April availability earliest, unless they release the lower DDR5x skews first then the DDR6 cards later.

 

[harmattan]

AMD will most likely sell everything they have and can make first two months so no need to stir anything up. Prices can be reduced later if needed. Still $399 for a 8gb HBM2 card is actually good - just might not give you much over the DDR5 card is the sad part. As for Nvidia, they may want to clear inventory towards the end of the year, good time for it in preparation for Volta. Still think Volta will be 2nd quarter launch next year ~ April availability earliest, unless they release the lower DDR5x skews first then the DDR6 cards later.

That will likely be the case. As NKD pointed out, it's neither in nV nor AMD's interest to start a price war (AMD because they can't afford it, and nV because they don't want to cut margins). AMD also doesn't want to set a precedent for lower pricing on the next gen, and lowering price would be contrary to the inflationary trend in the GPU market over the past 5-6 years.

That said, once the process and sourcing ramps up, costs go down so they may be able to shave off some on the price to better compete. Retail pricing is an elegant dance of multiple actors including Nash equilibrium, competition, market forces, manufacturing dynamics, time, price trending etc and you can bet your bottom dollar AMD considers everything.

 

[CSI_PC]

Anarchist4000 haha!

Yes. Let me know when Vega is outperforming Titan Xp in something that isn't an amd tech demo. I'm still waiting for confirmation of your claimed 40% performance improvements from async.

You have a long list of shit you've yet to prove

Yep remember this wonderful demo on 'Async Compute':

The only place I see FP16 making any notable contribution will be consoles, in the same way 'Async Compute-Shaders' has mostly contributed to consoles rather than PC where at absolute best we have seen with a game is 10% on PC and that is ignoring it is never sustained and that ideal situation does not happen frequently on PC while also influenced by resolution.
Most games are around 4% to 8% regarding Async Compute on PC where the option can be tested.
AoTS manages between 8% to 10% (taking the ideal situation), probably the heaviest game to use it, based upon analysis reviews that managed to measure this better with independent 3rd party tools.

But I do think the FP16 may be notable with consoles just like async compute can be, minimal for the rest of us and the context and focus of all posting here, especially considering the demograph of PC users with access to accelerated FP16 will be like 0.2% and it could mean some fudged rendering engines again just like we saw with those that fudged from DX11 to DX12 in early days as the resource commitment will just not be there yet and will not be until there is a sizable demograph meaning all AMD Polaris and Nvidia users replacing their GPUs.
Cheers

 

[FrameBuffer]

#waitforvega has now turned into #waitforFP16patches...

followed by #WaitForVolta

 

[Factum]

followed by #WaitForVolta

Why would you say that...since Pascal beats Vega already...logic fail much?

 

[RPGWiZaRD]

Neither AMD nor Nvidia would want to enter into a price war and would like to avoid it as much as possible unless it's intentionally a part of a strategy for market penetration and long-term profitability. Racing to the bottom may be good for us consumers but it's not good for business and these to companies are businesses with costs, expenses, employees, investors and competitive survival to think about.

The problem with the current pricing if we assume it would be roughly GTX 1080 performance and significantly higher power consumption (which should also be a factor of the price) and likely have worse thermals and noise levels is that I don't expect them to sell that well, my guess comparing GTX 1080Ti/1080/1070 vs Vega RX 64/65 then it would continue to be in Nvidias favor despite GTX 1000 series is very old by now, probably by a factor of 60~65 vs 35~40% in Nvidias favor. Pricing the RX 64 (cheapest SKU) $400 and 56 at $329 you'd probably even skew that sales figures maybe potentially even to like 60 vs 40% in AMDs favor. Of course this is obviously difficult to predict exactly how much of an impact but I'm not so sure they will get that much more revenue pricing that high if they cannot move enough products to begin with.

Nvidia have lately had an upper hand in "image", even if both cards are equal in everything, the market share typically slightly favors Nvidia, steam and various sources support this. Pricing themselves too close to Nvidia while potentially having a lot higher power consumption/heat/noise can become a problem to get sufficient number of sales, especially the response would be pretty "meh" in this case among reviews and consumers alike. Starting a price war is the only "working" one when you come like 14 months late, with higher power consumption etc, it's the only way to get a positive spin on the situation (AMD isn't ripping ppl off, yay!). I think you're downplaying the importance of this scenario.

No matter how expensive HBM2 or the architecture is and now matter how much they desperately need revenue, the argument is still valid.

 

[Archaea]

Where can I preorder?

I'm selling off my dual 1080ti because freesync is just that good. (Came from Crossfire Fury X, two weeks back, sold them when I bought the 1080ti - That was a mistake)

 

[Araxie]

Pricing just seems terrible to me.

I think it would be funnier than hell if nVidia dropped the 1080 ti down to $599 LOL ............ god I hope they do this. Hell even $649 would be funnier than hell.

that would be actually terribly and would solidify VEGA as 1080Ti/ Titan X contender at 599$, what worse AMD Fanboys will claim everywhere that AMD Forced Nvidia prices down.. it would be a terrible marketing material for Nvidia. lol.

 

[Ieldra]

that would be actually terribly and would solidify VEGA as 1080Ti/ Titan X contender at 599$, what worse AMD Fanboys will claim everywhere that AMD Forced Nvidia prices down.. it would be a terrible marketing material for Nvidia. lol.

AMD fanboys are already claiming Pascal is obsolete and that Titan Xp won't just get beaten by Vega in a year's time, but it won't even be capable of running games at high settings. I don't think it's physically possible for them to get any more enthusiastic about Vega

 

[geok1ng]

vega's price is not based on gaming performance alone, but rather combined mining + gaming value. Freesync was already cheaper , now Vega's bundle make it even more attractive. It is too soon to call the variable refresh rate format war over, but Freesync has a dominant market position and vega will make it even stronger.

 

[razor1]

Shill for Nvidia much?

When its reality, its not shilling

How many people here said Vega was going to be the next best thing since sliced bread, Still have a couple of people in this very forum still thinking that way. Can't get much more disillusioned then that. I can understand people saying wait for reviews, but when ya got people saying features will give 100% improvements when activated. LOL the only way to believe in something like is if they got a pile of crack!

There are always excuses from the Red side man for Ryzen too, CCX problems can be fixed, sorry can't be, reduced but not fixable. Vega is in the same boat, nothing to fix, its just NO WHERE near what the competition can do in performance and perf/watt or power consumption. Its obvious AMD is behind by a full 1 generation and soon to 2 generations in tech.

Don't remember wait for Fiji, its going to put Maxwell to shame? Polaris too shit they took a mid range chip and made it into a gtx 1070/80 killer. And now with Vega, its a Titan X/1080ti killer, now expectations are down a bit because AMD is comparing it to a 1080, but still have some crazy ideas out there of it going to go up against Volta once drivers are ready.