AMD VEGA confirmed for 2017 H1

[D.P.]

No one is saying AMD will only offer a single $600-700 GPU, quite the opposite. Of course AMF will offer a range of Vegas based cards at different price points. 1070,1080 and 1080ti competitor with similar prices and performances.

 

[Seanspeed]

It is certainly possible to increase pricing and change your products to be premium centric with appropriate marketing. Stella beer is a classic example, a truly terrible beer that had declining sales but they increased the price and threw a load of money at marketing it as a premium beer. Sales sky rocketed and profits ballooned.

The difficultly for graphics cards is they are reviewed and benchmarked so people have objective reasons for not believing in the premium marketing. If AMD had released a 1080 competitor at the 1080 launch they certainly could have and should have released it at the same price point. A year later that isn't such an obvious move.

Also, if the rumors are true and there is a very tight supply of Vega cards and HBM2 prices are high then I don't see AMD charging less than Nvidia pricing. They may not be able to keep up with demand anyway so might as well maximize profits.


AMD don't need to set lower prices. If you look at performance per Dollar:
https://www.techpowerup.com/reviews/Sapphire/RX_580_Nitro_Plus/32.html

The RX480 and 10760 have basically the same ratio, the RX580 has worse performance per dollar. Supposedly RX480 is selling OK. Given the HBM2 and interposer costs, die size etc I imagine the vega 1080 competitor to be very similarly priced but AMD will make a lot less profit.

If HBM is killing them on cost and availability, I dont expect that to stop them from going a bit more aggressive on price, even if it means a dent in profit margins.

AMD has spent the last year already losing market share in the $300+ enthusiast market, which Vega is expected to *finally* come in and compete with. They cant come in with pricing similar to Nvidia at the same performance. They just cant. People called Fiji a disaster, this would be well beyond anything like that. It would be a catastrophe. If AMD have made a bad bet on HBM2, they're gonna have to suck up the margins on it in order to keep themselves in the game. This is how competition works. For AMD, marketshare is potentially even more valuable than short-term profitability, though of course that's always difficult to convince shareholders. But it's an obvious situation.

 

[Panos]

Gibbo when we pre-order?


LOL

To those who cry about Vega availability and that going to be only 20,000 units at launch day.
Do do forget we are 2 months later, pre-orders are done for many AIB GTX1080Tis and still have unknown delivery date yes?

Pay a visit to the other threads in here like the EVGA GTX 1080 Ti FTW3/SC2 iCX......

And thats a mature tech that everyone was expecting to come out since last year...........
Does that mean that there is GDDR5X availability issues, or NV cannot produce GTX1080Tis?

 

[AlamoX]

If HBM is killing them on cost and availability, I dont expect that to stop them from going a bit more aggressive on price, even if it means a dent in profit margins.

AMD has spent the last year already losing market share in the $300+ enthusiast market, which Vega is expected to *finally* come in and compete with. They cant come in with pricing similar to Nvidia at the same performance. They just cant. People called Fiji a disaster, this would be well beyond anything like that. It would be a catastrophe. If AMD have made a bad bet on HBM2, they're gonna have to suck up the margins on it in order to keep themselves in the game. This is how competition works. For AMD, marketshare is potentially even more valuable than short-term profitability, though of course that's always difficult to convince shareholders. But it's an obvious situation.

+1
companies are impacted not just by their own revenue, but also by the revenue of their competitors, AMD would be better off with lower margins if they can manage to stay within the same revenue while grabing money off their competitor.
AMD 1B$ vs Nvidia 1.5B$ is much better than AMD1B$ vs Nvidia 2B$ per quarter, there is a lot more going on than just margins that ppl seem to be focused on, analyst and investors are more likely to tank your value through your market share, or how far behind your competitor you are, than how much margins you make out of your products.

 

[LoadsaMoney]

If HBM is killing them on cost and availability, I dont expect that to stop them from going a bit more aggressive on price, even if it means a dent in profit margins.

AMD has spent the last year already losing market share in the $300+ enthusiast market, which Vega is expected to *finally* come in and compete with. They cant come in with pricing similar to Nvidia at the same performance. They just cant. People called Fiji a disaster, this would be well beyond anything like that. It would be a catastrophe. If AMD have made a bad bet on HBM2, they're gonna have to suck up the margins on it in order to keep themselves in the game. This is how competition works. For AMD, marketshare is potentially even more valuable than short-term profitability, though of course that's always difficult to convince shareholders. But it's an obvious situation.

Its a compute card (like Nvidias Tesla), hence the HBM2.

 

[TNA]

They havn't made a bad bet on HBM2, its a compute card.

Don't for get VR

lol

 

[D.P.]

+1
companies are impacted not just by their own revenue, but also by the revenue of their competitors, AMD would be better off with lower margins if they can manage to stay within the same revenue while grabing money off their competitor.
AMD 1B$ vs Nvidia 1.5B$ is much better than AMD1B$ vs Nvidia 2B$ per quarter, there is a lot more going on than just margins that ppl seem to be focused on, analyst and investors are more likely to tank your value through your market share, or how far behind your competitor you are, than how much margins you make out of your products.

Investors lowered share value because net profit was much lower than revenue stream should indicate. And lowering profit margins doesn't necessarily increase revenue. If AMd sells a 1080ti competitor at $600 vs $650 what would eb the total sales difference , revenue and profit? One could easily see revenue actually decline because sales volume is relatively inelastic at that price point.

 

[spoffle]

Its a compute card (like Nvidias Tesla), hence the HBM2.

People say that about Titans when it suits them.

 

[D.P.]

Its a compute card (like Nvidias Tesla), hence the HBM2.

Its not a compute card, it only has 1/16th FP64 support. It will find itself in soem HPC uses like deep learning but its not comparable to tesla offerings.

 

[D.P.]

People say that about Titans when it suits them.

For the original Titan it was true. Several colleagues got them, insanely cheap Tesla basically.

 

[Phixsator]

For the original Titan it was true. Several colleagues got them, insanely cheap Tesla basically.

The only time a Titan card had a reason to exist. nVidia then took care of that and now there are no reason for it to exist which mean they get to increase the price and it sells like hot bread. The world is a weird place with no sense in it.

 

[Freddie1980]

Thought wcc was a load of rubbish

Thought the minimum VR spec was an RX480 at 199 USD!

 

[spoffle]

For the original Titan it was true. Several colleagues got them, insanely cheap Tesla basically.

It wasn't a compute card. It was a gaming card. nVidia Marketed it as a gaming card and nothing else.

It of course excelled at compute, but that wasn't intentional which is why they made sure any further Titan cards on new architectures were cut down in compute performance.

 

[N19h7m4r3]

I hope the air cooled Vega is as sexy was the 2900 series was!

Everyone knows it goes faster with flames.

 

[AlamoX]

Investors lowered share value because net profit was much lower than revenue stream should indicate. And lowering profit margins doesn't necessarily increase revenue. If AMd sells a 1080ti competitor at $600 vs $650 what would eb the total sales difference , revenue and profit? One could easily see revenue actually decline because sales volume is relatively inelastic at that price point.

nothing in Q1's report was surprise, everyone knew that from AMD's projections, investors didn't all of a sudden discover that, there is a lot of money to be made...
yes, not necessarily, but a well executed margin cut can increase revenue and even profit, all you need is to identify the right product for it, and how big of a cut.
you said it yourself, 600 vs 650$, the difference is negligable, 50$ doesn't make ppl stop to ponder over the value of a 600$ product, they usualy just spend the extra 50$ for the peace of mind, i know i would.
so what's the point ? they might as well put it at the same price, they would roughly sale the same units as the -50$ ones, but put a 499$ price tag on it, and you would grab the attention of 90% of the customers, if your product is good enough you might end up with a good portion of that, so not only you get customers from your competitor's, you also decrease his margins since he would need to lower his price to stay relevant.
all of this depands on how important market share is to you, or if the public perception just doesn't allow you to claw back market share, AMD ave been stuck between 23% and 27% going up and down without ever breaking the 30%, let alone the 40% to become relevant again in gpu market.

 

[LoadsaMoney]

Its not a compute card, it only has 1/16th FP64 support. It will find itself in soem HPC uses like deep learning but its not comparable to tesla offerings.

Yeah deep learning, if they did a gaming card, it would no doubt be using GDDR5/5X, like Nvidias gaming cards do

 

[N19h7m4r3]

Yeah deep learning, if they did a gaming card, it would no doubt be using GDDR5/5X, like Nvidias gaming cards do

That involves a new design compared to just reusing the same interposer with HBM2 next to the GPU.

AMD being cash strapped that's most likely something they just couldn't afford.

 

[D.P.]

It wasn't a compute card. It was a gaming card. nVidia Marketed it as a gaming card and nothing else.

It of course excelled at compute, but that wasn't intentional which is why they made sure any further Titan cards on new architectures were cut down in compute performance.

Pretty sure they marketed as a both a compute an gaming card in one.

EDIT: 1 min of google:
http://www.anandtech.com/show/6760/nvidias-geforce-gtx-titan-part-1/4

Titan is most certainly geared in part as a gaming video card (and that’s largely how we’ll be looking at it), that’s not the only role it serves. Titan is also going to be NVIDIA’s entry-level compute card

Titan, as we briefly mentioned before, is not just a consumer graphics card. It is also a compute card and will essentially serve as NVIDIA’s entry-level compute product for both the consumer and pro-sumer markets.

Nvidia could have easily disabled the Fp64 support as they did with the 780ti etc. but they purpsoely left it enabled as a prosumer compute card without the professional support.


Maxwell had to drop Fp64 support due to transistor constraints. By the time we got to Pascal Nvidia has diverged HPC and gaming lines.

 

[LoadsaMoney]

That involves a new design compared to just reusing the same interposer with HBM2 next to the GPU.

AMD being cash strapped that's most likely something they just couldn't afford.

Yeah, can't do both like Nvidia, plus for gaming, they are years later, and at a performance level thats years old, and has already been surpassed.

 

[D.P.]

That involves a new design compared to just reusing the same interposer with HBM2 next to the GPU.

AMD being cash strapped that's most likely something they just couldn't afford.

no, not a complete design but reuse the memory controller form Polaris etc.