icecold1983
Member
Not a great showing for vega in yet another title. Tho it was clearly faster than the 1080. It was only behind in 1 scene in the entire video
-
Hey Guest. Check out your NeoGAF Wrapped 2025 results here!
Radeon RX Vega thread
- Thread starter PFD
- Start date
Not a great showing for vega in yet another title. Tho it was clearly faster than the 1080. It was only behind in 1 scene in the entire video
Is there ANY rumors or speculation about price yet?
Mindshare and PR have absolutely nothing to do with how well a videocard works inside my computer to produce graphics on my screen. If you have Nvidia stock or something you should be forthcoming about that, because you're talking about business garbage when I'm trying to discuss the performance of a card with respect to price.
If you want to a discussion about how some idiots automatically buy green because of PR regardless of how good a card is at it's pricepoint, PLEASE have that discussion with someone else. I don't have time for bullshit, so you're typing into the ether. Thanks.
Mindshare and PR have absolutely nothing to do with how well a videocard works inside my computer to produce graphics on my screen. If you have Nvidia stock or something you should be forthcoming about that, because you're talking about business garbage when I'm trying to discuss the performance of a card with respect to price.
If you want to a discussion about how some idiots automatically buy green because of PR regardless of how good a card is at it's pricepoint, PLEASE have that discussion with someone else. I don't have time for bullshit, so you're typing into the ether. Thanks.
llien
Banned
Given the R&D budget disparity, it's surprising AMD even exists.
Bit's & Chips "leaked" it to be very competitively priced.
Linus has no proof, but he expects it to land around $600 for the model that is faster than gtx1080.
Hey, man you're the one who responded with "but but price and performance", which the default answer of every AMD evangelist to anything negative about AMD. If you don't have time to deal with my bullshit, maybe try to follow the thread instead of getting angry.
My original response was to napata, who originally argued "what does Vega being potentially bad have to do with the PC gaming industry?"
Which is pretty self evident why a non-competitive AMD would be bad. Mindshare and PR absolutely have a lot to do with whether or not a GPU is competitive or not on the market. If you want to deflect everything with price and performance, AMD isn't going to get anywhere. You might buy AMD if they're $50 cheaper but evidently the majority of the market won't.
RenegadeXV8
Banned
Yeah, currently on a quarterly basis NV outspends AMD by about $150 million in R&D. Now consider that AMD has to split their budget between CPUs and GPUs.
There is no chance they will ever catch up unless Volta is another FX/Fermi blunder (highly unlikely) and this is exactly why AMD can't "just compete better".
I doubt a Fermi blunder would impact GPU sales a whole lot. NV pretty much started clawing whatever marketshare AMD gained with the GTX460 and GTX500 series. Individual bad products only harm the brand temporarily.
Its going to take a blunder the size and duration of Intel's recent offerings for AMD to start seriously clawing back positive PR and mindshare. Recent Intel products are more or less fine but the sheer amount of negative PR seems to be impacting their reception with reviewers and the public.
Almost no one really cares that Intel produced the fastest processor in the world. Everyone is fixated on the negatives: super hot VRMs on the motherboards (which probably isn't actually a real issue but shouldn't even be an issue in the first place), stupid high power consumption (looks terrible but task energy is still very competitive), thermal paste on $1000+ processors (total garbage move fostering years of frustration with enthusiasts), paid software RAID (cash grab), Kaby Lake-X disabling half the physical functions of a X299 board (weakass move). Intel's just fostered years of community frustration that AMD's Zen is harnessing extremely well by offering the opposite of all of these complaints.
Unknown Soldier
Banned
To be fair to AMD, all they had to do was temporarily hire Jim Keller back to design a new CPU microarchitecture for them and at the same time ask Intel to be nice and slow down for about 6 years so they could catch up. Now we have Ryzen.
Unfortunately it doesn't look like Nvidia's planning on slowing down anytime soon nor is there a superstar GPU designer like Jim Keller available for hire at the moment. I don't think this will work out well for AMD.
StrongBlackVine
Banned
I saw a rumor that to end version Vega Nova that competes with the 1080 will be $499. $399 for Vega Eclipse which will compete with the 1070. No idea of that is any where near accurate though.
Wait for navi
In contrast to Intel, Nvidia never ever "slowed down" and thats what makes the difference here. Intel was never asked to slow down though, what a dumb statement.
Also, Zen was designed by a large team thats still at AMD, Keller was a part of it. On the CPU side, AMD is doing great, and thats were the money probably went to.
llien
Banned
Don't be ridiculous.
Gap between nvidia and AMD is mostly imaginary (unlike insane the gap with Intel, which was mostly addressed by Ryzen, which was better than anyone has expected). AMD knows how to make good GPUs, their x80,x70 series are very competitive with minor (30-50W is not really a big deal) downside on power consumption side of things. They don't have to beat $700+ GPU of competitor to "catch up", merely have xx80/xx70 line products out. They have bet on single mem type and that didn't quite work out. (I'm pretty sure nVidia also had dev team working on it, it's just, they could also afford to work on GDDR in parallel)
Yes, but nVidia loves big margins too much.
Which years is the "slow down" period for Intel?
On top of it, unlike intel, nvidia has never was as far ahead.
I dont think its that easy. Nvidia is at least one generation ahead and it looks like it will stay that way. Sure AMD can create a midrange chip, at the size of a highend chip, that competes with Nvidias 60s, 70s and 80s lineup in terms of performance...but their effeciency does not look so great atm. And while energy effeciency isn´t everything, if the disparity is as big as it looks to be right now, it will factor into reviews and the image those chips have in the public.
Vega should be their Titan-esque halo chip, but it looks like it will be around the 1080´s levels of performance, maybe a bit more but at a much higher die size, with exotic and expensive memory and a high power draw. Maaaaybe they will gain a lot once games optimize for it, but ~1080-levels of performance in the vast majority of games with a few outlier highend games is not that attractive.
Part of what made Zen possible was that Bulldozer sucked so hard, that they made a clean break more or less. GCN is still doing okayish compared to what happened with Bulldozer back in the day, so they are not forced to go back to the drawing board. They probably can´t afford it, too.
Please keep in mind that i replied to a statement from Unknown Soldier. While i dont know what he meant exactly (if his statement was well thought out to beginn with), from my POV Intel definitely slowed down in terms of core counts. While IPC increased steadily at a slow place, they were never forced to give consumers that much more because there was no pressure.Which years is the "slow down" period for Intel?
On top of it, unlike intel, nvidia has never was as far ahead.
Aztechnology
Member
I'm not sure why that's necessarily true. It'll all depend on the price. AMD despite what many said was stupid, went first for the lower/mid market with the 470/480 and really improved their market share. The amount of people looking to buy a Titan level card is insignificant comparative to those in the 1060/1070 range. If the Vega is priced somewhere between the 1070 and 1080 that could work out fine. But obviously then Nvidia will just adjust price too. Still, many people, myself included are looking to get away from Nvidia. I want to leverage Freesync on my display and hopefully future TV's.
neurosisxeno
Member
For 99% of users 2-4 cores are perfectly fine. Intel offered higher core counts in the Xeon line which they intended for workstations/servers--the stuff that actually benefits from more cores.
Not even reading your reply. Please stop talking to me.
Lagspike_exe
Member
This. The RX460/470/480 are the key markets. Titan/1080/1080Ti are borderline insignificant when it comes to volume (although margins are nice). Halo effect exists but it is hugely overblown.
Also, "going back to drawing board" as some have suggested here doesn't really work that way. It takes years to make an architecture from scratch and AMD's position isn't as nearly dire as some here believe.
While that is true, that does not change the fact that they could have offered more, yet did not because there was no pressure.
I agree that they will have to fight with low prices. But then again, i see no reason at all to go for a Vega chip other than "no Nvidia". Nvidia offers the same performance with lower power draw, and they can adjust their prices accordingly as you wrote it already. Sure they will sell a good amount of them, partly because mining drained the stocks and Vega (from what i remember) is not so great at it.
But if we look at the wider picture, for AMD and RTG this thing sucks atm from a gaming perspective. Its a large chip with exotic, expensive memory. At the end of the day, AMD can´t operate on goodwill and flowers, they need to make money. And Vega does not look like something that can deliver in this area on the long run.
600 sounds like why not save for a 1080ti...
500 sounds good if it's > 1080. 400 for something comparable to 1070 sounds like a pretty bad deal
Buildzoid, the guy who does PCB analysis for Gamers Nexus, claimed on his livestream today that his Vega was power throttled at 300w, let alone 450w, which justifies the huge 12 phase stock VRM configuration. Basically, if you're probably going to get nice performance if you lift the power limit but you better have the cooling and PSU ready for it.
I can't see how they can get the price low on this card without taking huge hits on their profit margin. You've got a huge die combined with HBM combined with a mandatory power delivery design that hasn't been seen on stock GPUs because no one has really made anything that can potentially draw so much power. There's no way they can cut costs anywhere without decreasing the performance/performance ceiling/reliability of the card.
Tacklemacain
Member
If they have the ressources, AMD can definetely catch up again. The last year, they focus way more on CPUs which will make them more money. Navi will also be the first architecture mainly developed by Raja Kudori, it will also use infinty fabric and can hopefully duplicate waht ryzen is doing in the CPU space right now. The biggest problem for AMD is to keep things going, after the developed something good.
Sir Abacus
Member
I realize the AMD playbook for the last decade has been developing disappointments and then selling them at low cost to go for "VALUE!" hype but how exactly are they going to sell a 500-something mm² die as a value proposition next to the 1080? Are they just going to make zero money on their flagship product? AMD bean counters were up there on stage talking about how they're going after the higher margin markets but they seem to have no strategy to have a margin. Maybe they should start developing new cryptocurrency fads because it's the only thing keeping the RTG alive right now.
As a consumer of the high end market segment hardware I'm so pissed at AMD because Nvidia now have me over a barrel and my options are either eat whatever price they set for Volta or... nothing. I can't even buy an AMD card and switch over to the Freesync ecosystem because nothing they make now or in the near future even comes close to my current 1080 Ti. So pissed. If I had a Freesync monitor right now I'd be absolutely livid at these clowns.
opticalmace
Member
Link?
That's not cheap enough.
US market 1070s and 1080s can be had for less than that.
Unless Amd somehow has something up their sleeve as far as the actual Vega gaming chips they are fucked unless a 1070 competitor is $300 and 1080 $400
I have a freesync monitor and I'm not livid at AMD. Presumably they are doing the best they can. I think if Vega is not up to snuff it's because they spent time optimizing more for the professional market, which honestly is more financially lucrative if they can pull it off. And if they do, having more resources towards R&D will allow them to better compete in the future. I think that's what the strategy is atm.
If Vega falls significantly short of 1080 Ti it will be their first GPU product in awhile that didn't compete on the high end. Fury was still pretty close to 980 Ti, and 290X traded blows with the 7xx series.
Anyways I've pretty much given up chasing top performance due to Nvidia's vendor lock in. I find some FreeSync monitors are superior to their GSync counterparts (actually, the one I have - 38UC99 has no GSync equivalent currently). So Nvidia GPUs just aren't an option to me anymore. There's also a good chance of FreeSync coming to TVs in the future. I'm at the point where I'd rather sacrifice performance for flexibility - but it is thanks to Nvidia, not AMD that I need to make this compromise.
Latest WAN show, I'm on mobile atm tho :'(
opticalmace
Member
Np, thanks, I'll find it. I haven't watched any LTT material in a while but wouldn't mind hearing what they have to say.
1080 vs Vega FE vs 1080 Ti on Witcher 3 @4k
https://www.youtube.com/watch?v=eALjhkpWTFk
https://www.youtube.com/watch?v=eALjhkpWTFk
Honestly a big question mark is how well this thing can OC with an aftermarket or hybrid set up.
I'm assuming this is the stock cooler and it's staying pretty cool. Wonder if they could crank out a 200mhz OC on there and see what happens. If it can get close to a ti ocd for less money than a 1080 might sell OK.
Buildzoid sort of gets into this. This thing power throttles really goddamn hard so that's your main roadblock when it comes to performance.
If we're going by raw performance, I suspect you can probably get past the GTX1080ti with sufficient cooling and aftermarket BIOS/power delivery changes so the power limit is higher. But to get there, you're going to need a significantly more powerful PSU than whatever you'd need with your average GTX1080ti. Its basically AMD's version of Fermi.
Alright so yeah it needs to be cheaper to offset increased power draw, but I feel like pretty much anyone in this market segment probably already has or is looking at a 600w-800w at least bronze psu which should be fine.
But a cheap price on this kills margins on such a big chip, so ultimately looks like a failure for amd.
I do feel like amd driver game is still weak as hell as far as optimization for game performance so I could see this card having great legs like Hawaii. Even Fury had pretty good legs but was limited by that 4gb. Could push well past a Ti down the line, but people don't buy graphics cards for 1-2 years from now as there's no guarantee it will keep improving like Hawaii did.
JohnnyFootball
GerAlt-Right. Ciriously.
Considering the pre-release hype for Vega is the complete opposite than for the 480/470 and Ryzen mnaybe its part of the game to have expectations low and then deliver slightly better than 1080 performance at $399
StrongBlackVine
Banned
You really think that this is possible? I don't.
LelouchZero
Member
AMD has been so weird with Vega, they've talked a little about it, teased it a bit and then after that we've just have silence.
They're being so hush hush about it that I've now begun to expect mediocre performance from it. Perhaps they're having trouble with it and don't want to say too much about it, as it could create hype and then disappointment if it doesn't impress?
The only thing I can hope for now is that it's competitively priced against the GTX 1080 and offers similar or slightly better performance at a price-point in-between the GTX 1070 and GTX 1080.
Earlier this year there were leaked AMD slides claiming Vega would only draw 225 Watts.
Since Frontier Edition guzzles 300 Watts while begging for more, either those leaks where fake or there's a big gap between what Vega is currently capable of and what AMD expected it to be capable of.
The faint hope that Vega might turn out ok hangs on the fact that AMD first showed off a working Vega card running Doom at the AMD Tech Summit in December last year. That makes it at least eight months from having working silicon to launch. Even in the chip business that's time enough to do some tinkering.
One hypothetical scenario goes like this: In December last year AMD had a working Vega chip, but it had serious flaws preventing it from working as designed. AMD did a respin they thought would fix the problem and announced a H1 '17 release window.
At some time in Q1 they get the respun chip back from Global Foundries and... Oops! That didn't actually fix the problem. At this point AMD has two choices:
a) Release the flawed chip in Q2 and accept that they would not gain market share this generation.
or
b) Delay the launch, spend lots of money on another respin, and hope that this time they could actually fix it.
Of course there is also a slim possibility that instead they did:
c) "Why don't we have both". AMD for some reason launches the flawed chip at the end of Q2 as the Frontier Edition while also sending Vega back for a second respin. If said respin was successful the RX Vega they have promised to launch at SIGGRAPH may turn out to be an ok chip after all.
Personally I wouldn't bet any money on that.
Since Frontier Edition guzzles 300 Watts while begging for more, either those leaks where fake or there's a big gap between what Vega is currently capable of and what AMD expected it to be capable of.
The faint hope that Vega might turn out ok hangs on the fact that AMD first showed off a working Vega card running Doom at the AMD Tech Summit in December last year. That makes it at least eight months from having working silicon to launch. Even in the chip business that's time enough to do some tinkering.
One hypothetical scenario goes like this: In December last year AMD had a working Vega chip, but it had serious flaws preventing it from working as designed. AMD did a respin they thought would fix the problem and announced a H1 '17 release window.
At some time in Q1 they get the respun chip back from Global Foundries and... Oops! That didn't actually fix the problem. At this point AMD has two choices:
a) Release the flawed chip in Q2 and accept that they would not gain market share this generation.
or
b) Delay the launch, spend lots of money on another respin, and hope that this time they could actually fix it.
Of course there is also a slim possibility that instead they did:
c) "Why don't we have both". AMD for some reason launches the flawed chip at the end of Q2 as the Frontier Edition while also sending Vega back for a second respin. If said respin was successful the RX Vega they have promised to launch at SIGGRAPH may turn out to be an ok chip after all.
Personally I wouldn't bet any money on that.
StrongBlackVine
Banned
I'm new to this era of PC gaming so maybe I don't understand the dynamics, but why does it matter if Vega doesn't compete with the 1080 Ti? It's not like that card is a best seller or anything. If Vega can give slightly better performance than 1080 at a cheaper price it will be a good value.
Unknown Soldier
Banned
The issue is that Nvidia has had the top end to themselves for over a year now. Anyone in this market segment already owns a 1080 or 1080 Ti. The only people who are still waiting for Vega are the most dedicated of dedicated fans and I don't think there's really all that many of those people out there.
The problem with always assuming the card will be much better 1-2 years from now is that in 1-2 years we'll all be using Volta anyways so it's irrelevant. AMD is far too late with Vega now to play the "Fine Wine" card, everyone who would have bought a Vega save the small number of dedicated fans already have Pascal and most of them are thinking about what they'll do when consumer Volta drops. AMD doesn't even exist at this point at the high end because they have been away from it so long, and producing a product which is barely competitive with the more than 1 year old 1080 while drawing double the power will move the needle not even the tiniest iota.
Fury X had the benefit of launching around when 980 Ti did. This time, Vega will be launched probably 5-6 months after 1080 Ti did. Vega has literally no chance at this point, AMD needs to just launch it, accept it's failure, and move quickly to ship Navi as soon as possible so they can try to keep up with Volta instead of trailing by more than a year again. Honestly if AMD can't have Navi out the same year as Volta, they might as well just spin off or dissolve the Radeon Technology Group because it's completely dead at this point with zero market share, zero mind share, and zero hope for the future.
Because it's more than a year later than the 1080 (Pascal) at this point. PC gaming moves quickly, a new GPU generation comes along every 18 months or so. AMD is going to be shipping Vega to compete with Pascal just about when Nvidia is going to be launching their next generation Volta.
If it launches with better performance than 1080 at lower price too it's good value for gamers yes(tho its releasing more than a year later and drawing more power), not so much for amd when they are selling a more expensive card to produce for a cheaper price
Linus on Linus Tech Tips detailed the huge problem AMD has with regards to actually selling Vega to the end consumer. As cool as HBM sounds, it really wrecked price flexibility.
NV already have working samples of Volta for the professional market...which means the consumer grade stuff shouldn't be that far behind. If Vega is amazing, they drip feed leaks about consumer Volta and then launch a few months later to make Vega look silly. If it isn't amazing, they just price cut Pascal and see if AMD wants to engage in a price war that they can't win with such a card.
NV already have working samples of Volta for the professional market...which means the consumer grade stuff shouldn't be that far behind. If Vega is amazing, they drip feed leaks about consumer Volta and then launch a few months later to make Vega look silly. If it isn't amazing, they just price cut Pascal and see if AMD wants to engage in a price war that they can't win with such a card.
StrongBlackVine
Banned
So what happens with a new generation like Volta? A new GPU with 1080 Ti performance for the price of 1080 or even 1070?
Pretty much.
Most likely. See: The 1070 versus the 980 Ti and the 970 versus the 780 Ti.
yup, no one is buying this
There is no node shrink this time around, so don't expect miracles.
Hope that AMD keeps pressure on Nvidia, so that both companies remain offering affordable and competitive products.
llien
Banned
The issue with only beating 1080 (if even that, given the benchmarks) is that Vega is a 1.5 times bigger chip with much more expensive memory.
It doesn't mean they can't roll it out with competitive perf/$ with current 1070/1080 prices, but if they are able to produce Vega's en mass, nVidia can easily counter even with Pascal cards.
StrongBlackVine
Banned
So when is Volta expected to launch? Sometime later this year hopefully. Seems like I need wait for both companies to reveal their hands before I make I decision.