What do you think the top end Vega model needs to do to be successul?

As for Vega, I showed you info. You're assuming...


What info, I didn't see a link, the only people that ever thought Vega 11 was a bigger chip was forum talk from what I remember, and that talk was taken by a couple of websites and postulated as big Vega, just hearsay.
 
What info, I didn't see a link, the only people that ever thought Vega 11 was a bigger chip was forum talk from what I remember, and that talk was taken by a couple of websites and postulated as big Vega, just hearsay.

Check again. I edited my post a couple of seconds after I submitted it.
 
yeah Guru3d read their source material wrong, Vega 11 is small vega according to where they sourced their info from......

Unless they are saying Vega 11 rumor is from somewhere else.

Vega11 - lets call it "Big Vega' which is to replace the FIJI / Fury (X) parts. So as suggested, where Vega10 would get 4096 stream processors and could compete with the GeForce GTX 1080 and GTX 1070 Vega11 on its end would feature 6144 stream processors (=rumor)

They even say rumor.

Now for the info from Videocardz (from the link from the Guru3d article)

http://videocardz.com/63700/exclusive-first-details-about-amd-vega10-and-vega20

We have also learned that AMD is planning to replace Polaris 10 with Vega 11 next year. The specs were not released and we can only expect Vega 11 to share the same fabrication node with Vega 10 (14nm).

Smaller Vega.
 
Vega could come with a supermodel who gives out free blowjobs and Nvidia would still out-sell them.
Wait. Let's not get carried away here...
It could be 15-30% faster as AMD has been in past, in multiple segments at times over last 10-15 years, but lets face it - a majority of you brand self-concious e-peen fanboys, still wouldn't dare jump ship. You're the apple zombies of the PC world. You'll shit on AMD at any slight chance, glorify Intelvida, then say 'but muh cumpatitionz gud' whilst never doing shit about it when they do pull a clear winner out.
If AMD offered the best I would buy AMD. Ryzen is on my radar now, so I'm eagerly awaiting reviews. I am still pessimistic when it comes to their video cards, though. AMD needs to stop playing catch up with NVIDIA and come out on top, for once.
 
Vega 11 is the high end, I don't need to document it, it's been known for a good while now.



So, care to post a photo of your 7970?

I'm using two 7970 GHZ Editions in Crossfire, watercooled with Krakens and I'm ready to upgrade this generation. Willing to give AMD a chance on Vega as I've gotten tremendous value from my last two AMD gpus. Came from Nvidia before these cards. I want the best bang for my buck. If that means waiting I'll wait.
 
I dont care about competition if its a year late. And even then not competitive. If you want to play the #waitforever game your life is over and you never get to enjoy.

Now try again to think on who is annoying. Those that keep telling you to wait and buy something slower in the meantime, or those that just live with the times in the world of reality so you can enjoy it while you live?

Remember to cheer on VIA and Matrox next time too, even better, buy them. You know...for competitions sake.


Well I don't really see a lot of people recommending others to wait forever until the new stuff comes out... usually if something is out in the distant future people would just recommend getting the best bang for your buck at the moment and enjoy it. But if something new is released soon like in a month, it would be smart as a consumer to wait because the new stuff might be better and the old might drop in price. It's a WIN-WIN for us, the consumers.

But you can go ahead and bash any brand as long as they are not Intel/Nvidia. I've seen some like you over the years on [H] and they usually aren't here anymore when I come back; seems like they can only wake up with an agenda to bash certain computer hardware brands for so long.
 
Vega is a card that should have been out 3 months ago. Not in 6 months from now.

Where are you seeing that Vega isn't coming out until Summer 2017? Everything that I have seen points to early 2017 (January or February).

Empirically, they showed that Vega demo at the event the other day so I highly doubt they will wait 6 months until they release it.

This is another one of your "spew some speculated information" and treat it as fact. You repeatedly do this in many threads and it's getting annoying. Why don't you just wait and see what the performance is like on release? If it is shown to be a dog by 3rd party review sites then go ahead and bash it. At the moment we don't really know how it will perform though.
 
Where are you seeing that Vega isn't coming out until Summer 2017? Everything that I have seen points to early 2017 (January or February).

Empirically, they showed that Vega demo at the event the other day so I highly doubt they will wait 6 months until they release it.

This is another one of your "spew some speculated information" and treat it as fact. You repeatedly do this in many threads and it's getting annoying. Why don't you just wait and see what the performance is like on release? If it is shown to be a dog by 3rd party review sites then go ahead and bash it. At the moment we don't really know how it will perform though.

AMD writes H1 for Vega in their financial information. Q1 for Zen. So its not going to be Q1 to begin with.

Polaris was demoed long before launch.
 
Where are you seeing that Vega isn't coming out until Summer 2017? Everything that I have seen points to early 2017 (January or February).

Empirically, they showed that Vega demo at the event the other day so I highly doubt they will wait 6 months until they release it.

This is another one of your "spew some speculated information" and treat it as fact. You repeatedly do this in many threads and it's getting annoying. Why don't you just wait and see what the performance is like on release? If it is shown to be a dog by 3rd party review sites then go ahead and bash it. At the moment we don't really know how it will perform though.

1st half of 2017 is what AMD has stated repeatedly. And they have stated this many times in different Q calls and other presentations. That can be 1st Q or 2nd Q most likely AMD doesn't have an idea yet that is why they are giving a wide net for the date. IF they don't have an idea, better to pick Q2.
 
There is no official Vega, the RX490 is pretty much unknown at this point, it may be a hybrid Vega/Polaris with HBM slapped with a Fury X 2.0 type feel. And that is perfectly fine given that Vega is more Q1 2017 ETA probably towards the back. The RX490 should be closer to the 1080 to be relevent. I am also hoping for a RX485 flavour to push the 1070
 
Where are you seeing that Vega isn't coming out until Summer 2017? Everything that I have seen points to early 2017 (January or February).

Empirically, they showed that Vega demo at the event the other day so I highly doubt they will wait 6 months until they release it.

This is another one of your "spew some speculated information" and treat it as fact. You repeatedly do this in many threads and it's getting annoying. Why don't you just wait and see what the performance is like on release? If it is shown to be a dog by 3rd party review sites then go ahead and bash it. At the moment we don't really know how it will perform though.

Zen is coming. Late this month or Jan of 2017. AMD has always stated 1st half of 2017 for Vega. That can mean as late as June.
 
If companies start buying a bunch of 3 petaflop racks I'd imagine that could put off the consumer release a bit. Especially if HBM is in short supply or any other constraint. While Naples is 2H that could potentially eat into the supply as well depending on timing.

What would be interesting to see for a top end Vega model would be SSG style technology. Maybe SODIMMs or a built in secondary memory pool of DDR3/4. Get the card reporting 32/64/128GB of VRAM and completely remove DX12/Vulkan resource management from the picture. Should take a significant load off the CPU as well. Let games load all the resources with lower latency and minimal CPU involvement. M.2 Raid-0 would still be fun for some enthusiasts. I'm sure someone would be ballsy enough to create a 4TB graphics card.

I'll also add my requisite dual Vega design with 16GB HBM and 4TB of overflow.
 
I want better than Titan Xp performance with HBM2 RAM in a Nano-sized package…!!!
 
That would be called a GTX 1080ti with a water block on it. You probably won't miss the HBM2.

I would water-cool this theoretical Vega BeastGPU, but I still want it in an ITX-sized package for the card/full cover waterblock portion of the package…
 
I would water-cool this theoretical Vega BeastGPU, but I still want it in an ITX-sized package for the card/full cover waterblock portion of the package…

My RX 480 is tiny without it's reference cooler. I would expect VEGA to be smaller because of HBM2.
 
Depends on the price. If they can make a card that performs between a GTX 1070 and 1080 for $400 then it will be a good buy. I know a lot of people here only care about Titan X and the like, but practically no one buys the $900 graphic cards. They need to hit the key market good; the $200-$400 range. Power draw within reason does not bother me. I have a 650 watt Earthwatts PSU and I will like to keep it as long as I can, but an extra 50-60 won't break my PSU. Far from it. I can't imagine such a difference even being noticeable on a power bill. It depends on how much you run your GPU at full speed.
 
It depends how long this card is intended to be on the market before a higher tier model comes out.
At a minimum 1080 level of performance is fine, but at some point they will also need a higher performing card by 6- 8 months; may be that they have to go dual GPU or if like I am thinking use the full core count eventually for gamers that I expect to be initially only on the HPC cards (launches a bit later than the gaming model), 1st consumer gaming card has cut cores IMO and following similar strategy to what Nvidia did with GP102 where only more recently did they released the full core version and that was for HPC as the Tesla P40, Titan was a cut down core model that was available a few months before the Tesla P40.
But the delay launching Vega has hurt AMD a fair bit IMO (sales and top performance perception), and has ramifications for when to expect next generation relative to Nvidia Volta.

Cheers
 
Last edited:
To be successful in my opinion
  • It will need to beat each price segment, as in 1070 class, $399 and 10%+ faster.
  • Likewise 1080 class, $599 and 10%+ faster.
  • This would blur the whole Nvidia lineup -
    • Why buy a 1070 when you can get the XXX from AMD that is 10% faster or why buy the 1080 when the same XXX is only 10% slower or buy YYY and it is 10% faster then Nvidia etc.
    • Or why buy the 1080Ti when YYY is only 10% slower and $300 cheaper.
Really AMD would mainly need to win big in the 1070 price point and have a lower version like at the $349 price point. The top halo cards sell much less then these. Still having the top performing card does give benefits so if AMD can just wipe the floor would be ideal.

I guess it will depend on clock speed and power limits in the end if Vega will be a smash or another Fiji.

As for speeding up Pascal for an update, I think that would be weak - my 1070's already run over 1950mhz on their own in games - going to like 2100mhz on an update would at best a 7% increase in performance meaning almost nothing in the end for gaming experience.

Some areas that AMD could automatically win will be form factors - getting some cheaper versions of the Nano type cards also a Halo Nano type card can open up some doors that Nvidia can't get into. Particularly with high end OEM gaming rigs.

I think Shintai will have an AMD card sooner then he thinks :D
 
It needs to perform around the 1080 or better (assuming we're taking that as the top tier rather than the semi prosumer Titan) and have better price/performance.

Even if it comes out next month it'll be 8 months behind the 1080, it has to differentiate itself by either price or performance or both, otherwise what was the point? They can't get away with a FuryX style release where they put out a slower card for the same price, and that was only a few weeks late.
 
For AMD to win they would also have to counter anything that Nvidia could do. For example Nvidia could enable more cuda cores for the 1070 to up the performance like they did with the GTX 260 and reduce the price of the 1080. Also much flexibility in GP 102 configurations from the Titan X to the 1080Ti. Nvidia could enable more cuda cores there as well and go with DDR5(x) with the 1080Ti. Nvidia just has a ton of options including major price reductions.

AMD will have to play it smart and if they can blow it out of the park would be their best chance of a successful launch.
 
I'm happy with what they've been doing for years as long as the launch also fires the entire marketing team. For me it's about price performance ratios and it's hard to beat any AMD card even factoring in the extra power consumption. Just because I can afford SLI Titan XPs doesn't mean I like throwing away hard earned money. I'll gladly take less performance at a substantial price cut and that's what I buy. Vega just needs to be faster than the fury X so there is some high end card to sell and cost an amount that keeps AMD alive and an amount that makes it sellable. I'm tired of having the fastest computer at whatever cost, anyone can build a 6950k with Titan XP computer and brag but be a copy paste of everyone else. I can take pride at having the fastest computer at a price bracket like 1k$ 2k$ etc.
 
It needs to be able to play games at 8k @ 5000 fps and be able to make my bed, take out the trash, cook dinner, etc

In all reality it needs to be akin to whatever the 1080ti is going to perform
 
I think the GTX1080 is a great performing graphics card. But Vega is a bit late to the party and we still don't know exactly which month it will launch. IMO, it needs to perform at least between the 1080 and the new Titan. The closer to Titan the better. Just equaling the 1080 is not quite good enough.
 
In all honesty I think the key for a successful launch is performance relative to expectations. Vega needs to exceed the expectations set by AMD's marketing team, and given what we've seen already the flagship vega needs to smoke a GTX1080 to exceed expectations...
 
Why does everyone forget about the Fury X ?
AMD already has a card at 1070 levels.

It's last year's generation, it sometimes competes with the 980Ti (which means it sometimes competes with the 1070). It never competed with last year's Titan, let alone this year's.

Thread is asking what Vega needs to be to be successful. If the Fury X was all they needed to do, then it's game over; that ship already sailed.
 
In all honesty I think the key for a successful launch is performance relative to expectations. Vega needs to exceed the expectations set by AMD's marketing team, and given what we've seen already the flagship vega needs to smoke a GTX1080 to exceed expectations...
More than that. If any of the Volta rumors are true then Vega will be competing against NVIDIA's next gen with current gen performance.
 
I would like to see AMD with their own "Titan" for a change. Just blow everything out of the water and create some buzz for having the fastest thing on the planet for gaming.


that would be nice, the 4k 5k 6k series were fun watching nvidia and AMD go back and forth. especially when Nvidia finally thought they had AMD beat on that series AMD would then release the X2 cards and slap nvidia back down to reality. it's sad to see that ended after the 6k series only for Nvidia to do their version of that with the single gpu titan cards.
 
Why does everyone forget about the Fury X ?
AMD already has a card at 1070 levels.
The Fury X is seriously suffering with quite a few modern games )mix of complexity with driver support/in some way the games/etc), I listed 4 modern games on another thread that either show the Fury X close performance to the 480, only slightly ahead, and in one or two behind, this is also at 1440p resolution - I was not even including ones such as Shadow Warriors 2.
And this performance is not necessarily fixed to the right levels in these games.

Post #34 https://hardforum.com/threads/im-th...-but-i-have-concerns.1920005/#post-1042707632

Cheers
 
Last edited:
I would like to see AMD with their own "Titan" for a change. Just blow everything out of the water and create some buzz for having the fastest thing on the planet for gaming.

That would be nice but they would have a GPU primarily used in super computers that Nvidia does in order to justify making such a beast and I don't see AMD doing that especially with their lower R&D.

My issue with AMD is, when they have a winner of a product they also have a huge issue getting product on the shelves. I'd like to say that I would of owned more AMD GPU's in the past but I got tired of waiting for stock or refreshing a page hoping for some to come in and just went with Nvidia.

My favorite was the X800XT, had cash burning a hole in my pocket and couldn't find one to save my life except for the horrendously over priced stores and ebay listings. Ended up going with a 6800GT after I fried my X800pro.
 
I remember also trying to find an X800XT. I never could find one instock, and got the 6800GTX instead.
 
The Fury X is seriously suffering with quite a few modern games )mix of complexity with driver support/in some way the games/etc), I listed 4 modern games on another thread that either show the Fury X close performance to the 480, only slightly ahead, and in one or two behind, this is also at 1440p resolution - I was not even including ones such as Shadow Warriors 2.
And this performance is not necessarily fixed to the right levels in these games.

Post #34 https://hardforum.com/threads/im-th...-but-i-have-concerns.1920005/#post-1042707632

Cheers

You're right, i do own a Fury X and it's like a surprise with every new games, you never know how it's gonna perform!
I was lucky that mine overclocked by over 10% (i'm at 1105 core) but i really hope AMD comes up with a 1080 killer so that i can upgrade! (I have a freesync 3440 monitor so i'm kinda stuck with AMD)
 
It's going to be hard for Vega to pull out any kind of win. Stay afloat is probably their best bet. Even if it matches 1080, Nvidia has a lot of leeway on price now. Their process is mature and yields are higher. They could easily drop prices and cut AMD off at the knees. If it outperforms 1080 or even Titan XP, Nvidia still has an ace up its sleeve. We've yet to see the full GP102, Titan XP is a cut 102. Nvidia could release the 1080ti as the full GP102 like they did with 780ti at a price close to the plain version. Nvidia has several options to counter anything AMD comes out with, but AMD doesn't.
 
Last edited:
even if this card is better than the 1080 or 1080ti at a lower price point, people are going to wait for the inevitable nvidia price drop and buy a 1080 or 1080ti.

quote me, i bet i'm right.

If Nvidia chooses to drop prices, which they might not do because people are buying their stuff anyway.

Quite frankly, the way I see it, if Vega outperforms the 1080 at a lower price point, I'm totally buying it and a Freesync monitor, because as it stands I would need to spend an extra $200-$300 CAD in order to buy Gsync over Freesync. The overall package is significantly less expensive when I factor that in.

AMD has been consistently disappointing, so I'm hoping Ryzen and Vega break that trend for the sake of the consumer.
 
It could be 15-30% faster as AMD has been in past, in multiple segments at times over last 10-15 years, but lets face it - a majority of you brand self-concious e-peen fanboys, still wouldn't dare jump ship. You're the apple zombies of the PC world. You'll shit on AMD at any slight chance, glorify Intelvida, then say 'but muh cumpatitionz gud' whilst never doing shit about it when they do pull a clear winner out.

Quoted for truth. I don't understand the fanboyism. Unless you're heavily invested as a shareholder, if you're not making buying decisions based on price/performance, you're an idiot.
 
It's going to be hard for Vega to pull out any kind of win. Stay afloat is probably their best bet. Even if it matches 1080, Nvidia has a lot of leeway on price now. Their process is mature and yields are higher. They could easily drop prices and cut AMD off at the knees. If it outperforms 1080 or even Titan XP, Nvidia still has an ace up its sleeve. We've yet to see the full GP102, Titan XP is a cut 102. Nvidia could release the 1080ti as the full GP102 like they did with 780ti at a price close to the plain version. Nvidia has several options to counter anything AMD comes out with, but AMD doesn't.
The full GP102 has been released but looks to be reserved for Tesla HPC model; check out the Tesla P40.
Unfortunately I doubt (would be nice though) they will release the 1080ti with more cores than the Titan XP.

From NextPlatform:
nvidia-pascal-table.jpg


https://www.nextplatform.com/2016/09/13/nvidia-pushes-deep-learning-inference-new-pascal-gpus/
Also worth noting the reason the P100 has more SMMs is due to it being a ratio of 64-cores per SMM rather than the normal 128-cores.
HP is FP16 (only on Tegra and GP100), SP is FP32, and DP is FP64.
Cheers
 
If Vega is HBM2 then I think nVidia will have nothing to worry about. A lot of us here called it with the Fury line. We predicted additional cost, yield issues and supply issues for no significant gain. That's what ended up happening. Also I never saw why people saw on die VRAM as a good thing... Let's concentrate the heat more. That should be especially awesome with the new node where the heat density already went up.

AMD won't be able to make enough to dent the market.
 
Back
Top