AMD Official Statement - Polaris is NOT delayed

[Armsdealer]

Even though the strategy of pursuing high margin products works for NVIDIA, it makes sense that AMD wants to sell as many GPU's as they can (with the <$300 GPU's) to regain market share. If AMD is at 20% market share give or take for the DGPU market right now, falling any further would create a situation where it becomes less and less of a priority for devs to bother optimizing for both vendors. Grabbing more market share also makes them more relevant and gives them more brand recognition. Lets hope they can get Polaris 10 out soon.

Consequently their product should offer an *extremely* compelling value proposition. Regaining dGPU share in PC is more important to them right now than margin. Devs are going to optimize for their hardware anyway given their monopoly on console.

 

[antihelten]

AMD AIBs need to eat $150 in profit per card by selling a card worth $449+ for only $299? No wonder why they are "frustrated".

OEM won't eat anywhere near $150 in opportunity cost by AMD pricing a Polaris 10 card with performance equal to a 1070 at $300.

First off the 1070 is only $450 for the founders edition. The custom editions have MSRP of $380. Secondly AMD generally won't be capable of competing with Nvidia at equivalent prices and equivalent products, given the perceived differences in brand value. As such the highest they could probably price Polaris 10 at and still stay competitive would be $350.

As such the loss per card is only about $50, but this is of course compensated by the fact that OEMs will obviously sell a higher volume of a given card if it's priced at $300 instead of $350.

At the end of the day, it's impossible to say whether or not this is a good deal for OEMs, unless we could somehow get ahold of the cost price of Polaris 10 cards. But as Raghu mentioned, if Nvidia can sell a ~320mm2 die at $380, then AMD shouldn't have any problem selling a ~230mm2 die at $300 and see roughly comparable profits.

 

[tential]

On track to what? Also duh that AMD will be at Computex but this says nothing about its AIB partners showing off Polaris GPUs for gamers.

Great PR spin from AMD, but it doesn't actually directly address the rumors.

The fact amd is able to do pr spin like this is a step up. Nvidia would have done far better but amd is learning. It's like a toddler trying to imitate daddy. They'll get it soon.

So you're kind of blueballing me here. Should I be ready for the incoming amd launch disappointment or should I act like a battered wife and say this time it'll be different?

If only I was railvens gf then this wouldn't be a decision for me. :sigh:

 

[Ajay]

Of course AMD will be at Computex, they would be stupid to miss that...

Why aren't they on the E3 exhibitor list yet, though?

Maybe they are hosting an invite only event at a near by hotel?
I think they've done this b/4 - much cheaper.

 

[Slaughterem]

OEM won't eat anywhere near $150 in opportunity cost by AMD pricing a Polaris 10 card with performance equal to a 1070 at $300.

First off the 1070 is only $450 for the founders edition. The custom editions have MSRP of $380. Secondly AMD generally won't be capable of competing with Nvidia at equivalent prices and equivalent products, given the perceived differences in brand value. As such the highest they could probably price Polaris 10 at and still stay competitive would be $350.

As such the loss per card is only about $50, but this is of course compensated by the fact that OEMs will obviously sell a higher volume of a given card if it's priced at $300 instead of $350.

At the end of the day, it's impossible to say whether or not this is a good deal for OEMs, unless we could somehow get ahold of the cost price of Polaris 10 cards. But as Raghu mentioned, if Nvidia can sell a ~320mm2 die at $380, then AMD shouldn't have any problem selling a ~230mm2 die at $300 and see roughly comparable profits.

You are not even close to a good estimate of profit because you are basing things on same size chip. AMD at 232 mm2 will have almost 2X the amount of good dies as compared to 317mm2 Nvidia GP104

 

[96Firebird]

Maybe they are hosting an invite only event at a near by hotel?
I think they've done this b/4 - much cheaper.

I hope they haven't done that in the past, because that is one of the dumbest things I've heard... E3 is arguably the best way to get their product in front of gamers, you know, the main audience of AMD's graphics cards.

 

[maddie]

You are not even close to a good estimate of profit because you are basing things on same size chip. AMD at 232 mm2 will have almost 2X the amount of good dies as compared to 317mm2 Nvidia GP104

Yep. Many more dies per wafer, probably much lower power consumption with all the production cost benefits that go with it.

They can sell for much less and have similar margins relative to 1070, and I'm fairly certain that right now, AMD will settle for lower margins than Nvidia, to regain market share.

 

[Azix]

I hope they haven't done that in the past, because that is one of the dumbest things I've heard... E3 is arguably the best way to get their product in front of gamers, you know, the main audience of AMD's graphics cards.

The PC gaming show from last year is coming back and its very likely AMD will be there at minimum.

 

[thesmokingman]

This means our report suggesting AIB partners were unhappy with AMD is completely false and based on incorrect sources. Therefore, we were wrong in disclosing this information and wish to clarify the real truth after receiving an official word from AMD. So what does this all mean? Well, AMD isn&#8217;t struggling to produce quick driver updates, the Polaris architecture or failing to attend Computex. Also, their Q1 financial results were actually better than most analysts predicted. Therefore, any claims predicting their future demise are completely inaccurate. Of course, the Q1 financials are going to be poor when they&#8217;re preparing product launches later in the year for both the graphics and CPU segments. Hopefully, this clarification helped to quash the rumours being published right now, and we openly admit our original articles were wrong.

Now that is some crow.

The fact amd is able to do pr spin like this is a step up. Nvidia would have done far better but amd is learning.

Who do you think is behind the negativity?

 

[.vodka]

Good to see AMD shooting down this possible FUD before it did some real damage. They seem to be getting better at the game.

Just a week or two until we get some actual information.

 

[C@mM!]

Who do you think is behind the negativity?

Clickbait shit like this gets you ad revenue when people are focusing on new cards coming, and a better chance of getting review samples of cards off Nvidia if you look sympathetic to them.

 

[thesmokingman]

Clickbait shit like this gets you ad revenue when people are focusing on new cards coming, and a better chance of getting review samples of cards off Nvidia if you look sympathetic to them.

But nothing like this right?


https://consumerist.com/2006/02/06/did-nvidia-hire-online-actors-to-promote-their-products/

 

[JDG1980]

The rumor was specifically about Polaris 10, not Polaris in general, being delayed and having low clocks. The rumor was not that AMD wouldn't be able to launch any Polaris GPU at any speed for any market.

For arguments sake, if the Polaris 10 rumor were true, AMD could still launch Polaris 11 on schedule and still claim Polaris is on schedule to launch this summer and technically not be lying.

Or if Polaris 10 fell far short of its frequency target, AMD could just launch those as mobile GPUs, or perhaps desktop Tonga replacements, rather than a 390X replacement.

Why do you think an anonymous message board post should be given any credibility at all?

AMD has done public demos of Polaris. They've said, from the start, that they were targeting mid-2016 as the release timeframe. And they've now said explicitly that these latest rumors are BS.

 

[JDG1980]

Who do you think is behind the negativity?

Short sellers?

Stock market manipulation by social media (message boards) has been going on for over 15 years.

Alternatively, it could just be fanboyism or trolling. There are some other, less likely, possibilities as well. Maybe one particular AIB was disappointed for some reason (poor communication? limited supply? lack of a direct GP104 competitor?) and framed their own complaints as if they were coming from the industry as a whole. Maybe this was even an intentional test using false information to find out who was responsible for leaking.

 

[JDG1980]

For sure. It's enough reassurance for me to know that Polaris 10 is not an October product as some tried to proclaim on other forums. That means more mining rigs incoming for me in June/July. Good stuff, about to order another 1000W Platinum PSU for 3-4 Polaris cards. Looking forward to more info on BW-E as well.

Me, I'm hoping for a WWDC announcement of a new Mac Pro with Broadwell-E and Polaris 10. That would definitely be my next system (assuming it can still run Windows 7).

 

[eRacer]

dude do you know the wafer costs at TSMC and GF. Yield rates. Actual die size of GP104 looks to be 315-330 sq mm. Polaris 10 is rumoured to be 230-240 sq mm.

Doesn't matter if Polaris 10 was only 10mm^2 and cost $1 to manufacture. If Polaris 10 were better than GTX 1070 it shouldn't be selling for $150 less. And in reality it almost certainly wouldn't, because GTX 1070 prices would drop in order to be more competitive.

 

[eRacer]

Why do you think an anonymous message board post should be given any credibility at all?

I'll give the guy a little credit because he supposedly works in retail distribution (including video cards), and didn't just show up on the internet for the first time to make these rather bold claims. Based on some of his recent posts at OCN he doesn't seem to be an AMD hater either.

I can't give him a lot of credibility, though, because:

1) I don't know much about his history other than what he and others have stated about himself.

2) Even people with normally trustworthy sources can get fed bad info either intentionally (malice, find sources of leaks, etc.) or unintentionally (source had bad info but didn't know it).

3) What's true today may not be true tomorrow. Perhaps there was a severe glitch but AMD finds a fairly simple workaround and a crisis is averted.

AMD has done public demos of Polaris.

At what frequency? If the frequency was around 850MHz then running public demos does absolutely nothing to dispel the rumor.

Is this final silicon? Is Polaris 10 in mass production already? Are Polaris 10 cards already in AIB hands? If you can't answer yes to all of these then it can't be proven how well retail Polaris 10 would clock at this point.

They've said, from the start, that they were targeting mid-2016 as the release timeframe. And they've now said explicitly that these latest rumors are BS.

AMD didn't explicitly state anything about the rumors being BS in that story. Here is the quote: "there are no production issues whatsoever with Polaris and it&#8217;s on schedule"

"No production issues" is not the same as "no design issues". No production issues means GloFo hasn't screwed up, or it means Polaris 10 mass production hasn't even started yet so GloFo hasn't had the chance to screw up. It is quite possible to make a very disappointing chip with terrific yields.

Do you know what the production schedule for Polaris 10 is? Has mass production started yet? Has AMD made it public? If production of Polaris 10 hasn't started yet AMD is free to change the schedule and quantities produced at any time in the near future, and still technically not be lying when saying its on schedule today.

 

[kraatus77]

Doesn't matter if Polaris 10 was only 10mm^2 and cost $1 to manufacture. If Polaris 10 were better than GTX 1070 it shouldn't be selling for $150 less. And in reality it almost certainly wouldn't, because GTX 1070 prices would drop in order to be more competitive.

290x was better than titan and sold for almost half. 980 was better than 780ti and sold for 100 less, heck even 1070 is supposed to be faster than tx and selling for 550$ less. that's how you get market share.

 

[JDG1980]

AMD didn't explicitly state anything about the rumors being BS in that story. Here is the quote: "there are no production issues whatsoever with Polaris and it’s on schedule"

"No production issues" is not the same as "no design issues". No production issues means GloFo hasn't screwed up, or it means Polaris 10 mass production hasn't even started yet so GloFo hasn't had the chance to screw up. It is quite possible to make a very disappointing chip with terrific yields.

Do you know what the production schedule for Polaris 10 is? Has mass production started yet? Has AMD made it public? If production of Polaris 10 hasn't started yet AMD is free to change the schedule and quantities produced at any time in the near future, and still technically not be lying when saying its on schedule today.

It wasn't a direct quote, so all this detailed parsing is beside the point. According to article author John Williamson (in the comment section):

It’s pretty much been paraphrased and said in a colloquial manner and quoting it wouldn’t really add insightful information as it’s from excerpts. If you wish, I can do this. The basic argument is the Computex comments were rubbish, same with Polaris delays

Polaris won't be delayed. It will be coming out mid-2016 as planned.

 

[positivedoppler]

There were a hundo million shares short going into the earnings which is high for a stock that trades ~10MM shares / day. It's not unusual for negative rumors to show up after someone's caught with their pants down.

This. I suspect there's a few people on this forum with a financial stake in this. One day they will trip and hopefully this will go back to being a hardware enthusiast forum.

 

[eRacer]

290x was better than titan and sold for almost half. 980 was better than 780ti and sold for 100 less, heck even 1070 is supposed to be faster than tx and selling for 550$ less. that's how you get market share.

Titan and TitanX examples are somewhat meaningless as those are low volume halo products with luxury prices that don't reflect their performance. Change Titan to GTX 780 and Titan X to GeForce 980 Ti and the performance/$ comparisons become much closer.

Yes, AMD tends to offer better performance/$. Yes, products can be launched with surprisingly attractive performance and performance/$ to gain market share (GTX 970 for example). If AMD launches at $150 below GTX 1070 with higher performance, I'll be very pleasantly surprised (from a consumer standpoint). We'll see in the coming months what AMD actually delivers for consumers (and investors).

 

[IEC]

This. I suspect there's a few people on this forum with a financial stake in this. One day they will trip and hopefully this will go back to being a hardware enthusiast forum.

Would be nice if all the shorties lose their shirts and Polaris surprises us on the upside.

Full disclosure: I own shares of NVDA. I do not own shares of AMD. Despite this, I own AMD graphics cards and a Freesync monitor. Because I'm not paying a 50% premium to get equivalent performance... except for my laptop, where there really isn't any competition from AMD (yet).

 

[Armsdealer]

This. I suspect there's a few people on this forum with a financial stake in this. One day they will trip and hopefully this will go back to being a hardware enthusiast forum.

I am a hardware enthusiast for the record.

That said, being aware of the fact that there are people sitting on $100MM of paper losses over the last month in AMD's stock that are not in a good situation to exit that position given the lack of daily liquidity might help one understand the likelihood of negative rumors appearing. I don't know for certain that that's where it's coming from, but given I have a little bit of a background in stocks I know it's something that occurs often.

If spending $20-30k to move the stock with some bad news helps you exit a few percentage points better and no one is the wiser, people will definitely do it, particularly short sellers.

 

[Despoiler]

I don't think AMD was ever in danger of being behind schedule as they publicly stated they were targeting a back to school time frame for Polaris. The earliest and best indicator of where a company is in relation to launch is when they start clearing the channels of current inventory.

 

[railven]

Good to see someone at AMD squashing these rumors. They catch on quickly and spread like wild fire.

No comment on the Vega rumor, though? Maybe they want that one to perpetuate. haha.