• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Vega/Navi Rumors (Updated)

Page 197 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.
So it will barley beat a Vega FE at gaming and is priced high. What other reason could there be to even bring up the "total cost" debate. Only reason is your product (GPU) doesn't offer great performance/$ and performance/watt compared to competitor.

If it has trouble even matching a GTX 1080 (at twice the TDP for the watercooled SKU) and is going to be more expensive at retail due to manufacturing costs, then why release it at all? Who would buy something like that? Are the handful of sales to die-hard AMD fanboys really worth the scathing reviews that will inevitably result?

This is what I don't understand. Isn't there anyone at AMD/RTG willing to say "No, we're not going to release this because it's a pile of crap and will make us look like laughingstocks". The WX 9100 and Instinct MI25 can remain, they're fine for their market niches, but it's clear that Vega 10 as it currently exists is a complete flop for gaming. The biggest questions at this point are: (1) can this be fixed with a respin or new tapeout, and (2) is it due to the unbalanced nature of Vega 10's resources or does the entire architecture just suck for gaming?
 
If it has trouble even matching a GTX 1080 (at twice the TDP for the watercooled SKU) and is going to be more expensive at retail due to manufacturing costs, then why release it at all? Who would buy something like that? Are the handful of sales to die-hard AMD fanboys really worth the scathing reviews that will inevitably result?

This is what I don't understand. Isn't there anyone at AMD/RTG willing to say "No, we're not going to release this because it's a pile of crap and will make us look like laughingstocks". The WX 9100 and Instinct MI25 can remain, they're fine for their market niches, but it's clear that Vega 10 as it currently exists is a complete flop for gaming. The biggest questions at this point are: (1) can this be fixed with a respin or new tapeout, and (2) is it due to the unbalanced nature of Vega 10's resources or does the entire architecture just suck for gaming?
Do you have a job?

Let's say you work at Burger King. Do you stop selling Whoppers because Five Guys makes a better burger? Why not? Whopper is pretty sad in comparison, and you guys at Burger King have been practicing yours for 30 years longer.

Or at higher end, if you're an attorney, going to hang it up because another lawyer in town kicks your ass every time?

If you haven't heard, these parts are in development over years and usually more than one design in the works. They bring to market the best of what they have, when they can. They do not care if fans are frustrated, they only care if they can make enough profit on their parts to keep the doors open and fund development of the next part.

Position in performance hierarchy is largely irrelevant as pricing can make it attractive.

The ONLY way AMD would not bring Vega to market is if it can't be sold at a profit.
 
The ONLY way AMD would not bring Vega to market is if it can't be sold at a profit.

And that could actually be the problem. HBM2 is just expensive with the interposer and additional lost dies due to problems mounting the HBM2 and GPU on the interposer. That is easily $40 more per chip. But you also want your margin on that chip so you need to charge $60 more. at the sub $600 price point that already requires you price your product 10% more expensive than competitor or have a 10% lower margin. That is just huge. The almost certainly have to price it (full card) above $400 to make any profit at all. Add to that all the different Ryzen versions and Ryzen actually delivering, it can fill all of AMD 14nm capacity alone especially if Epyc takes off. And Ryzen dies have a much higher margin than this Vega will ever have.
So I guess they will price it high and will produce at low volume so price can stay high and it doesn't eat too much fab capacity from ryzen dies. Lol yeah. If Epyc really takes of, AMD doesn't need to sell a single Vega as GF fabs will be running at full capacity anyway.
 
So I guess they will price it high and will produce at low volume so price can stay high and it doesn't eat too much fab capacity from ryzen dies. Lol yeah. If Epyc really takes of, AMD doesn't need to sell a single Vega as GF fabs will be running at full capacity anyway.
Yes, GloFlo has a limited capacity, however, AMD did make the changes to be able to use other fabs, so if it really does take off, then they could use Samsung's fabs, assuming they have room.
I also think that is wrong, AMD definitely needs to sell Vega as well, going one year is bad enough, but two years without any answer to nvidia's line would be very, very bad news.
 
I also think that is wrong, AMD definitely needs to sell Vega as well, going one year is bad enough, but two years without any answer to nvidia's line would be very, very bad news.

If there isn't a miracle coming compared to FE like a respin and/or new driver which results in at least +30% gain vs FE or -50% power, then they already lack an answer. Who will buy a 1080 competitor that uses roughly double the power at same or higher price? Not many. Needs to be <$500 for sure but even then it would be a tough sell.
 
Yes, a very tough sell for certain markets, the gamble is, if those other markets can make up for all the loss revenue the hard-core gaming people would have made up.

It wouldn't be shocking if this (Vega RX) was a very limited run as well, just something to get out there.
 
If there isn't a miracle coming compared to FE like a respin and/or new driver which results in at least +30% gain vs FE or -50% power, then they already lack an answer. Who will buy a 1080 competitor that uses roughly double the power at same or higher price? Not many. Needs to be <$500 for sure but even then it would be a tough sell.
If performance is what it looks like, then 450$ is the tops they can charge.
Also the gtx 1080 likely overclocked better as well as the power consumption, Vega will be faster in the long run though as well as having hbcc so effectively 11-12gb ram vs 8.

The real nightmare is if nvidia drops the prices of GP 104 to below 450$ and 350$ respectively, which they can easily do.
Nvidia would likely lose a lot of money doing that though just to spite AMD- who won't sell many of these at any rate.
 
Yes, a very tough sell for certain markets, the gamble is, if those other markets can make up for all the loss revenue the hard-core gaming people would have made up.

It wouldn't be shocking if this (Vega RX) was a very limited run as well, just something to get out there.

I guess at the very worst they've got that Apple contract to soak a bit of inventory. Then there'll be some spare chips that don't make it for whatever reasons. Like with Tonga I suppose.
 
If there isn't a miracle coming compared to FE like a respin and/or new driver which results in at least +30% gain vs FE or -50% power, then they already lack an answer. Who will buy a 1080 competitor that uses roughly double the power at same or higher price? Not many. Needs to be <$500 for sure but even then it would be a tough sell.

Everyone considering a new monitor would seriously consider a Vega at the same price as 1080 if performance is comparable and that is their price point. I would think for most people saving $200+ on the monitor trumps "Oh noes this thing will cost me $1.83 more a month to game on!".

The people loyal to AMD brand will buy it.

Miners might depending on hash count.

People tired of their 1070/1080 might buy it for something different.

With a few free games thrown in people might go AMDs way.

People who just don't care about power consumption might buy it.

There are a lot of potential sales avenues for AMD, and power consumption is not the most important facet of a part for a lot of people.
 
https://videocardz.com/71090/amd-radeon-rx-vega-3dmark-fire-strike-performance

Benchmarker is Thegametechnician, Jason Evangelho who works for Radeon in their marketing department.

Radeon-RX-Vega-3DMark-1.jpg


2h4ly8i.jpg
 
And that could actually be the problem. HBM2 is just expensive with the interposer and additional lost dies due to problems mounting the HBM2 and GPU on the interposer. That is easily $40 more per chip. But you also want your margin on that chip so you need to charge $60 more. at the sub $600 price point that already requires you price your product 10% more expensive than competitor or have a 10% lower margin. That is just huge. The almost certainly have to price it (full card) above $400 to make any profit at all. Add to that all the different Ryzen versions and Ryzen actually delivering, it can fill all of AMD 14nm capacity alone especially if Epyc takes off. And Ryzen dies have a much higher margin than this Vega will ever have.
So I guess they will price it high and will produce at low volume so price can stay high and it doesn't eat too much fab capacity from ryzen dies. Lol yeah. If Epyc really takes of, AMD doesn't need to sell a single Vega as GF fabs will be running at full capacity anyway.

n0Ipjc7.jpg


IFTLE 279 2016 European 3D Summit: Cost Modeling Memory Stacks; Needed Tech for Next Gen 3DIC
 
If performance is what it looks like, then 450$ is the tops they can charge.
Also the gtx 1080 likely overclocked better as well as the power consumption, Vega will be faster in the long run though as well as having hbcc so effectively 11-12gb ram vs 8.

The real nightmare is if nvidia drops the prices of GP 104 to below 450$ and 350$ respectively, which they can easily do.
Nvidia would likely lose a lot of money doing that though just to spite AMD- who won't sell many of these at any rate.
Now when RX 580 and GTX 1060 are being sold at €300-350 price range, and GTX 1070 goes for €500, I think Vega priced at ~ €500-500 and with 250W TDP is the best thing gamers could get at this moment
 
Now when RX 580 and GTX 1060 are being sold at €300-350 price range, and GTX 1070 goes for €500, I think Vega priced at ~ €500-500 and with 250W TDP is the best thing gamers could get at this moment

MSRP for GTX 1080 is $600, so I would expect $500 MSRP for Vega right now.

Which... for many people, might not be enough Price/Performance.
 
Do you have a job?

Let's say you work at Burger King. Do you stop selling Whoppers because Five Guys makes a better burger? Why not? Whopper is pretty sad in comparison, and you guys at Burger King have been practicing yours for 30 years longer.

Or at higher end, if you're an attorney, going to hang it up because another lawyer in town kicks your ass every time?

Yes, if you have an inferior product, you can still move it by selling it at a lower price. That's what Burger King does, and it's what those lawyers with "$100 Divorce" signs in their windows do. And if your own costs are lower, you can still make a profit that way.

The problem is that it's looking increasingly likely that AMD can't sell RX Vega at a price that consumers will pay, and still make a profit. AMD was clearly targeting the $500-$750 price point with this card - it needs that price point because it uses a big die, large interposer, expensive HBM2, and a beefy power stage capable of 300-375W. But for gaming, the card is inferior to the GTX 1080 with a MSRP of $499. Full RX Vega could be successful at $399, but AMD almost certainly can't make a profit selling it at that price.

There's also opportunity cost to consider. Every wafer used for Vega is a wafer that can't be used for much more successful products like Ryzen, Threadripper, and EPYC, or even Polaris (which is currently selling like hotcakes).

Again, all this applies just to RX Vega. The WX and Instinct cards may be competitive enough in their niches, and there is more room to move with pricing on these. And of course they still have to fulfill the Apple contract. But gaming RX Vega looks to be DOA unless they pull out a miracle with a new driver, BIOS, or stepping.
 
MSRP for GTX 1080 is $600, so I would expect $500 MSRP for Vega right now.

Which... for many people, might not be enough Price/Performance.
1080 is $500 and I don't see AMD pricing below that when they get equal performance. They need as much profit on these as possible at the start.
 
1080 is $500 and I don't see AMD pricing below that when they get equal performance. They need as much profit on these as possible at the start.

Where is that an official MSRP?

In March this year Nvidia dropped 1080 MSRP to $600, though it is available (but out of stock) on their own website at $550
 
it still cost around 650USD here in eu.We all can only thanks miners.
Btw miners will buy out vega anyway.They buy out everything that is not at 2-3x MSRP.
 
Status
Not open for further replies.
Back
Top