AMD's Polaris graphics cards rumored to launch in late May

[Scorpus]

All the attention in the graphics card space has been focused on Nvidia recently, with the company announcing the GeForce GTX 1080 and GTX 1070 a few days ago. However it won't be too long before Nvidia has some competition, as AMD is expected to launch their new Polaris-based graphics cards at the end of May.

According to a report from Zolkorn, AMD is gearing up for a Polaris launch event in Macau sometime in the last few days of May. The event would occur before Computex 2016 begins on May 31st in Taipei, where board partners will be exhibiting their custom versions of AMD's new graphics cards.

Both Polaris 10 and Polaris 11 products are rumored to be shown off at AMD's launch event, although actual availability of the cards won't come until a few weeks later. This is a fairly typical launch pattern for a new graphics card, as AMD will want to drum up as much public interest in their products ahead of their in-store launch.

Judging by various leaks, AMD will be focusing on mid-range graphics cards with their Polaris 10 GPU, and energy efficient notebook graphics with Polaris 11. The top-end Polaris 10 graphics cards are expected to be successors to the Radeon R9 390 and R9 390X, bringing better efficiency and performance similar to an R9 Fury X (or GeForce GTX 980 Ti) down to lower price points.

Considering Nvidia's GTX 1070 launched at $379 with better performance than their last-generation flagship card, the Titan X, AMD's Polaris 10 cards will have to undercut this price by a decent margin if performance doesn't stack up. Some rumors have suggested Polaris 10 will launch around $300, which could certainly be competitive.

Gamers wanting a true AMD competitor to the GTX 1080 may have to wait until early 2017, when AMD is expected to launch a minor update to Polaris called 'Vega', bringing HBM2 and high-performance GPUs.

Permalink to story.

https://www.techspot.com/news/64757-amd-polaris-graphics-cards-rumored-launch-late-may.html

 

[Evernessince]

"Considering Nvidia's GTX 1070 launched at $379"

will launch or supposed to launch at. This hasn't happened yet so it can't be past tense.

WCCFTech has the die size of Polaris 10 at 1/3 smaller than the 1070. It would be very easy for AMD to undercut Nvidia and still make a very good profit.

 

[Squid Surprise]

"Considering Nvidia's GTX 1070 launched at $379"

will launch or supposed to launch at. This hasn't happened yet so it can't be past tense.

WCCFTech has the die size of Polaris 10 at 1/3 smaller than the 1070. It would be very easy for AMD to undercut Nvidia and still make a very good profit.

Well, they announced the price, so I guess they figure that counts as "launching".... If the new AMD cards only match Fury, then they are in some trouble... Nvidia has found a way to take their $1000 card and beat it for almost a third the price... Fury was already behind the Titan, so it doesn't look good...

 

[cliffordcooley]

Fury was already behind the Titan, so it doesn't look good...

That's what I'm thinking. I definitely hope not though.

 

[veLa]

"Considering Nvidia's GTX 1070 launched at $379"

will launch or supposed to launch at. This hasn't happened yet so it can't be past tense.

WCCFTech has the die size of Polaris 10 at 1/3 smaller than the 1070. It would be very easy for AMD to undercut Nvidia and still make a very good profit.

Well, they announced the price, so I guess they figure that counts as "launching".... If the new AMD cards only match Fury, then they are in some trouble... Nvidia has found a way to take their $1000 card and beat it for almost a third the price... Fury was already behind the Titan, so it doesn't look good...

That's what I'm concerned with too. AMD really needs to kick its butt into gear with this launch.

 

[darkzelda]

If this is true, they'll leave the high end market to Nvidia completely. I have a GTX 970 right now, but I expect my next upgrade to be AMD if they are up to it.

 

[Evernessince]

Well, they announced the price, so I guess they figure that counts as "launching".... If the new AMD cards only match Fury, then they are in some trouble... Nvidia has found a way to take their $1000 card and beat it for almost a third the price... Fury was already behind the Titan, so it doesn't look good...

I think we should wait for benches before taking Nvidia's press numbers into consideration. I was looking at the doom demo Nvidia was running off a 1080 and wasn't too impressed by the 166 or so frame average at 1080p.

Source

Heck even an 280x can get a solid 60 fps in that game on ultra. Given that, I'd call total bullshit on Nvidia's claim that the 1080 is faster than sli 980s. This is most likely only going to hold true in games with poor sli scaling (and probably why Nvidia used that comparison). Given the only game we've seen played at the Nvidia event, I'd put the 1080 above the 980 Ti but it's only going to be a 20% boost in most games. Of course, it could very well be that Nvidia is still having issues with DX 12 as vulcan is being used in the game.

 

[Squid Surprise]

I think we should wait for benches before taking Nvidia's press numbers into consideration. I was looking at the doom demo Nvidia was running off a 1080 and wasn't too impressed by the 166 or so frame average at 1080p.

Source

Heck even an 280x can get a solid 60 fps in that game on ultra. Given that, I'd call total bullshit on Nvidia's claim that the 1080 is faster than sli 980s. This is most likely only going to hold true in games with poor sli scaling (and probably why Nvidia used that comparison). Given the only game we've seen played at the Nvidia event, I'd put the 1080 above the 980 Ti but it's only going to be a 20% boost in most games. Of course, it could very well be that Nvidia is still having issues with DX 12 as vulcan is being used in the game.

You want us to wait for benches before we laud Nvidia - I agree... but I also think we need to wait for benches before we denigrate them as well...

Over the past few generations of video cards, Nvidia have fairly consistently beaten AMD - until proven otherwise, I see no reason to believe they won't continue to beat them...

 

[Evernessince]

You want us to wait for benches before we laud Nvidia - I agree... but I also think we need to wait for benches before we denigrate them as well...

Over the past few generations of video cards, Nvidia have fairly consistently beaten AMD - until proven otherwise, I see no reason to believe they won't continue to beat them...

I'm not saying their performance won't be better than AMD's, just that Nvidia's "faster than SLI 980s" claim is totally false.

Another thing to note, AMD's polaris 10 die is 33% smaller than Nvidia's 1070. The yields are going to be much better and it's going to be cheaper to produce. AMD will be able to sell them at a lower price and still make very good margins.

 

[Flebbert]

AMD is not trying to beat the Pascal cards in the launch, their cards are aiming to deliver two things:

1. Driving down the performance per watt required in the Polaris 11 range to bring "console" gaming to ultra books
2. Bring a VR capable card to the market at a reasonable price for everyday users with true DX12 and Vulkan support

This will only be a mid range release of cards and I don't think we will see anything amazing from the release.

I am a little disappointed but guess we have to wait for Vega to see some heat.

 

[theBest11778]

"Considering Nvidia's GTX 1070 launched at $379"

will launch or supposed to launch at. This hasn't happened yet so it can't be past tense.

WCCFTech has the die size of Polaris 10 at 1/3 smaller than the 1070. It would be very easy for AMD to undercut Nvidia and still make a very good profit.

Well, they announced the price, so I guess they figure that counts as "launching".... If the new AMD cards only match Fury, then they are in some trouble... Nvidia has found a way to take their $1000 card and beat it for almost a third the price... Fury was already behind the Titan, so it doesn't look good...

It should be alright. AMD's targeting the more lucrative segment the $200-$300 range. They did this with the Radeon 4850. Nvidia was bleeding money on high cost, but more powerful chips when AMD made a cheaper and powerful enough solution for 90+% of the market. People in tech forums like this expect to pay $350-$600 every 18 months on video cards. Most other people expect to pay $200 every 2-3 years. Both companies are going after different consumers. Nvidia won't be able to keep 1080's or 1070's in stock. My bet is AMD will have a much easier time keeping supply up with demand, and that's more important than having the faster card as more people have your chips the more word of mouth you get.

 

[cliffordcooley]

I am a little disappointed but guess we have to wait for Vega to see some heat.

Better yet the lack thereof. lol

 

[Squid Surprise]

I'm not saying their performance won't be better than AMD's, just that Nvidia's "faster than SLI 980s" claim is totally false.

So what evidence do you actually have to say that Nvidia's claim is "totally false"?!?!?!

Again - let's take your own advice and wait for the benchmarks before making any assumptions... cause when you assume, you make an @ss out of me and you...

 

[Squid Surprise]

It should be alright. AMD's targeting the more lucrative segment the $200-$300 range. They did this with the Radeon 4850. Nvidia was bleeding money on high cost, but more powerful chips when AMD made a cheaper and powerful enough solution for 90+% of the market. People in tech forums like this expect to pay $350-$600 every 18 months on video cards. Most other people expect to pay $200 every 2-3 years. Both companies are going after different consumers. Nvidia won't be able to keep 1080's or 1070's in stock. My bet is AMD will have a much easier time keeping supply up with demand, and that's more important than having the faster card as more people have your chips the more word of mouth you get.

Well, Nvidia did market the 960 and 950.... they've actually been pretty good in the mid-low end as well... check out the various GPU buying guides from Techspot and you'll see that they tend to have some pretty good cards at every price point.

In general, they are fairly even in the mid-low levels with the winner being decided by which products happen to be on sale at the time. The thing is, PROFITS actually don't come from low end cards - they sell for only a little bit over cost, thus netting little... it's the mid-high (not uber high) cards that reap profits... the 970 and 960 have been making Nvidia a lot of money - I suspect the 1070 and 1060 will do the same

 

[Evernessince]

So what evidence do you actually have to say that Nvidia's claim is "totally false"?!?!?!

Again - let's take your own advice and wait for the benchmarks before making any assumptions... cause when you assume, you make an @ss out of me and you...

I put a link in my video above showing the 1080 playing doom uncapped. At that point you simply compare the 1080's fps to the 980 Ti's fps.

Once again though, it's running on vulcan and doom, two things AMD seems to run very well on. I really don't know why they picked that game, it currently runs much better on AMD hardware.

 

[dividebyzero]

I'm not saying their performance won't be better than AMD's, just that Nvidia's "faster than SLI 980s" claim is totally false.

Their claim if you had bothered to follow any of the presentations, was aimed at a few well selected scenarios. For example, SLI scaling in the Witcher 3 is about 70%

While the presentation noted that the GTX 1080 was doing at least that well against the GTX 980 SLI at 1920x1080

Will the GTX 1080 beat SLI'd GTX 980's in all, or even a majority, of gaming scenarios across all resolutions. No.
Will the GTX 1080 beat SLI'd GTX 980's in some gaming scenarios. Undoubtedly - and will be an absolute certainty in VR gaming, and with game engines that don't support multi-GPU.

You seem intent on trying to apply a universal result to what is obviously marketing driven narrow subset (with the exception of VR, Async heavy settings, and any code that optimized for preemption). I think I can find this type of behaviour all thoughout the industry - some of it extremely hyperbolic, and coincidentally (?) you don't seem to have a problem with.

Another thing to note, AMD's polaris 10 die is 33% smaller than Nvidia's 1070. The yields are going to be much better and it's going to be cheaper to produce. AMD will be able to sell them at a lower price and still make very good margins.

That's a lot of assumptions considering you don't know what wafer agreement criteria AMD has with Globalfoundries, whether the latter's yields* (especially of the higher ASP defect free GPUs) are conducive to large OEM orders, and whether Pascal's launch might cause AMD to revise pricing to margins that fall short of "very good". AMD's average GPU selling price has fallen to around $28.90 even with blanket marketing of Fury, consoles, and VR.

*When was the last time Globalfoundries claims of new process yield success matched anything approaching the truth? There's a reason that AMD are in court because of this very matter. Both 32nm and 28nm had major ramping delays and yield issues, 20nm was silently killed, 14nmXM went from being a supposedly world leading process (60% jump in battery life! hows that for hyperbole) to "yeah, we didn't want to do it anyway"

 

[Puiu]

"Considering Nvidia's GTX 1070 launched at $379"

will launch or supposed to launch at. This hasn't happened yet so it can't be past tense.

WCCFTech has the die size of Polaris 10 at 1/3 smaller than the 1070. It would be very easy for AMD to undercut Nvidia and still make a very good profit.

Well, they announced the price, so I guess they figure that counts as "launching".... If the new AMD cards only match Fury, then they are in some trouble... Nvidia has found a way to take their $1000 card and beat it for almost a third the price... Fury was already behind the Titan, so it doesn't look good...

it doesn't matter, $380 is still a high price for a video card. it's not what sells and makes money for AMD/Nvidia.
if AMD plays their cards right they could seriously eat into Nvidia's market share.

 

[Evernessince]

Their claim if you had bothered to follow any of the presentations, was aimed at a few well selected scenarios. For example, SLI scaling in the Witcher 3 is about 70%

While the presentation noted that the GTX 1080 was doing at least that well against the GTX 980 SLI at 1920x1080

Will the GTX 1080 beat SLI'd GTX 980's in all, or even a majority, of gaming scenarios across all resolutions. No.
Will the GTX 1080 beat SLI'd GTX 980's in some gaming scenarios. Undoubtedly - and will be an absolute certainty in VR gaming, and with game engines that don't support multi-GPU.

You seem intent on trying to apply a universal result to what is obviously marketing driven narrow subset (with the exception of VR, Async heavy settings, and any code that optimized for preemption). I think I can find this type of behaviour all thoughout the industry - some of it extremely hyperbolic, and coincidentally (?) you don't seem to have a problem with.

That's a lot of assumptions considering you don't know what wafer agreement criteria AMD has with Globalfoundries, whether the latter's yields* (especially of the higher ASP defect free GPUs) are conducive to large OEM orders, and whether Pascal's launch might cause AMD to revise pricing to margins that fall short of "very good". AMD's average GPU selling price has fallen to around $28.90 even with blanket marketing of Fury, consoles, and VR.

*When was the last time Globalfoundries claims of new process yield success matched anything approaching the truth? There's a reason that AMD are in court because of this very matter. Both 32nm and 28nm had major ramping delays and yield issues, 20nm was silently killed, 14nmXM went from being a supposedly world leading process (60% jump in battery life! hows that for hyperbole) to "yeah, we didn't want to do it anyway"

My statement on performance wasn't universal. I posted specifically that one benchmark in a vulcan game was very odd for Nvidia and can flavor the results. If you read elsewhere "Card A is better than Card B" and are taking offense to that, you need to realize that these statement are made with overall results in mind. You can nitpick definitive statements all day but it's well understood that over-reaching statements like the forementioned are not all inclusive.

AMD can afford to have lower margins than Nvidia, Polaris was obviously designed to be a smaller more manufacturable GPU and even AMD's marketing suggests this. It's a good thing AMD has samsung as well. We don't know for sure if Globofoundires is even going to produce polaris GPUs at this point. AMD could very well fill its required quota through other means. There's too much assumption going on right now so let's just go on what little bit we have. We know for a fact that the 1080 is bigger than Polaris 10. This always translates to getting more product on the shelf and now that AMD has Samsung and Globo a supply chain issue is very unlikely.

 

[misor]

If NVidia and amd will remain true to the projected price range of incoming gpus, I am almost 100% certain to buy a gpu upgrade once these gpus become available at local ASUS retail channel (nearby city from my mountain hideout). goodbye asus gtx 960, goodbye evga gtx 960: the only gpu upgrades I had since I had my solo NVidia 9800gt...

 

[WangDangDoodle]

I'm guessing we'll see the 1080 Ti arrive early 2017 then. Nvidia is sure to sit on that gem until they see what manner of beast (or gimp) AMD delivers. I'm getting a bit worried about the prices, though. A 980 goes for ~670 USD in Norway, and the 980 Ti goes for ~790 USD. The 1080 will be priced similar to the 980 Ti, so I expect the 1080 Ti will cost me an arm and a leg. :s

 

[cliffordcooley]

I'm guessing we'll see the 980 Ti

I'm guessing you meant 1080 Ti?

 

[meric]

One thing here, why do we all assume that Nvidia's performance claims are all true? I mean we didn't see much benchmarks yet but we all are jumping conclusions. Some forum folks sound as if they know exactly how these cards perform and base further assumptions on assumptions Many are ready to pre-order these cards. For us consumers, it is always the safest bet to be patient and wait for detailed analysis made by respected review sites such as this one

 

[WangDangDoodle]

I'm guessing you meant 1080 Ti?

Ah yes, of course. Thanks. Best correct that.

 

[dividebyzero]

My statement on performance wasn't universal.

It sounds pretty definitive to me:

I'm not saying their performance won't be better than AMD's, just that Nvidia's "faster than SLI 980s" claim is totally false.

AMD can afford to have lower margins than Nvidia

Well, no they can't. The reason AMD are in the mess they are in is because their principle GPU competitor have gross margins over 100% higher than AMD (as does Intel). If you are content with AMD continuing to pare away R&D to allocate funds to debt servicing ahead of their 2019 fiscal armageddon, then fine...but without higher margins, there is no profit, and without profit there is no R&D, without R&D the product cadence slips and the holes need to be plugged with rebrands - and while rebrands are perfectly serviceable they don't drive revenue and profit like new architecture.

It's a good thing AMD has samsung as well.

A good thing indeed, since Globalfoundries owners have basically stopped investing in their own company and seem intent on get rid of it as soon as they got disillusioned with Glofo's execution (or lack of). Having Samsung on board can't hurt but you'd have to wonder why Apple made the decision to dump Samsung for the A10. That is one hell of a large contract.

One other point to consider: When was the last time Nvidia beat AMD to new product on a new process node?* There are Pascal products in abundance by the looks of it with reviewers already having samples - many having multiple samples. Haven't seen too much of Polaris...

* Twelve years ago- when Nvidia just beat out ATI with 6000 series cards built on IBM's 130nm FSG process, which beat ATI's 130nm TSMC Lo-K GPUs in real availability. Since 2004, ATI/AMD have been first to the punch in seven consecutive new processes...and now they have slipped for the first time in a dozen years even using much smaller silicon . For a company that has an overwhelming advantage in process technology thanks to its experience with its own fabs, this speaks volumes that they feel the need to start small - especially since their previous strategy has been to lead with the flagship GPU in every instance except 40nm (RV740).

You may feel that all these factors are not worthy of taking into consideration - the gross margins, the resultant decreased R&D, the slippage in process leadership (and Vega slipping a quarter in introduction resulting in Nvidia having the HPC and server markets to themselves for almost a year), and TSMC's 16nmFF+, a process supposedly immature and behind Samsung's is pumping out product ahead of its rival including the largest non-Intel GPU ever built......but I've seen enough of the semicon industry over the years to pay them some heed. I sincerely doubt the rosy picture you are trying to paint will end up the fairy tale landscape you are trying to convince me will result.

One thing here, why do we all assume that Nvidia's performance claims are all true?

Not all. Some people are claiming they are all false - whether they be benchmarks, or that the cards claimed (and demonstrated) overclocking on air is in fact some elaborate ruse, and in fact the cards require liquid nitrogen.
I suspect most people would fall into the "wait and see" category - especially since sales don't begin until ten days after the reviews go live. Basic performance based solely on architectural features would say that the 1080 should perform on par with a vendor custom 980Ti (I'm already on record as stating such). A quick perusal of that thread and others should show that the "insta-buy" crowd and the "It's Nvidia, so it will suck" crowd both represent minority viewpoints - or should do. This is supposedly an enthusiast site, and as such people should have the facts at hand before forming a firm opinion on the cards worth.

It should be alright. AMD's targeting the more lucrative segment the $200-$300 range. They did this with the Radeon 4850. Nvidia was bleeding money on high cost, but more powerful chips when AMD made a cheaper and powerful enough solution for 90+% of the market.

And yet AMD's discrete desktop graphics market share dropped from 38% to 31% in the year following the HD 4850/4870's introduction (in Q2 2008) when Nvidia fielded the big die GT200, and a bunch of warmed over G92/94/96 products. The RV770 was a great GPU, and the HD 4850 succeeded primarily because it wasn't a salvage part - it featured the full fat GPU of the HD 4870, but it did not displace Nvidia's rebranded 9800GT/GTX+ principally because of Nvidia's better OEM relations (leading to vast numbers of beige boxes sporting boring 9400GT/9500GT cards) and the halo effect of the GT200.

 

[Squid Surprise]

it doesn't matter, $380 is still a high price for a video card. it's not what sells and makes money for AMD/Nvidia.
if AMD plays their cards right they could seriously eat into Nvidia's market share.

Except they DO sell and they DO make money... They make far more profit off of more expensive cards than cheap ones....