(GURU3D RUMOR) AMD Polaris 10 GPU To Offer Near 980 Ti Performance For 299 USD?

Page 5 - 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.

96Firebird

Diamond Member
Nov 8, 2010
5,712
316
126
What rumors point to means jack as long as it's performing better than nvidia.
because that's the thing which matters.

if rumors point that polaris10 will be 3x faster than GP104. and it ends up being faster by just 10%. what do you think is best option ? GP104 ?because it's just 10% behind? LOL

same thing with furyx, doesn't matter what rumors said. it's the fastest 4k/dx12 card atm with better cooler. even on dx11 it's faster in some games.
unless game has gamewrecks.

I don't agree... Rumors point to a certain performance level, and are typically wrong. Take your vendor bias out of it, rumors of Fiji being faster than Titan X were wrong. That is what Udgnim pointed out, and he was right.

The rest of your babble was more nonsense.
 

maddie

Diamond Member
Jul 18, 2010
4,787
4,771
136
The article writer also has some ignorant beliefs. He's obviously clueless about the relationship between Mantle, DX12 and Vulkan.

"I'd argue AMD hasn't always ploughed what money it has into the right areas. Its current plan to kickstart a VR controller ecosystem—as revealed during Taylor's keynote speech at VRWC—seems misguided, particularly when history has shown developers and consumers prefer a single standard over an overwhelming amount of choice. And then there's Mantle, the short-lived but influential API that brought low-level hardware access to PC developers, but ultimately failed to take off in the wake of DirectX 12 and Vulkan."
 
Mar 10, 2006
11,715
2,012
126

Headfoot

Diamond Member
Feb 28, 2008
4,444
641
126
The article writer also has some ignorant beliefs. He's obviously clueless about the relationship between Mantle, DX12 and Vulkan.

"I'd argue AMD hasn't always ploughed what money it has into the right areas. Its current plan to kickstart a VR controller ecosystem—as revealed during Taylor's keynote speech at VRWC—seems misguided, particularly when history has shown developers and consumers prefer a single standard over an overwhelming amount of choice. And then there's Mantle, the short-lived but influential API that brought low-level hardware access to PC developers, but ultimately failed to take off in the wake of DirectX 12 and Vulkan."

Ars Technica is a joke these days. I've stopped reading it entirely and I used to be a daily click. They lost their technically competent people and gained a lot of people with big opinions and small knowledge.
 

Headfoot

Diamond Member
Feb 28, 2008
4,444
641
126
Is he serious? Is Roy Taylor really claiming NVIDIA won't have mid-range and low-end Pascal chips, too?

There have been spy-shots of GP106 and obviously there is GP104 coming too which isn't going to go into a $1k Titan card.

By the way, kinda funny coming from a rep of a company that just released a $1500 card for VR. LOL!

Probably just the context of the interview -- from what other interviews I've seen Koduri at least was talking for the first release. Rumor is that gp106 is not going to make the June launch but follow a few months after. I'd lean towards that AMD believes that Pascal midrange is first with low end / laptop to follow thereafter, there's no way they believe no laptop chip is coming ever
 

kraatus77

Senior member
Aug 26, 2015
266
59
101
I don't agree... Rumors point to a certain performance level, and are typically wrong. Take your vendor bias out of it, rumors of Fiji being faster than Titan X were wrong. That is what Udgnim pointed out, and he was right.

The rest of your babble was more nonsense.
lol, i have never used Amd card since 5770, currently own 670/970/210.
Green eyes as always. LOL , open sometimes maybe ?

http://www.techpowerup.com/reviews/Gigabyte/GTX_980_Ti_XtremeGaming/23.html

Furyx is fastest reference GPU. unless you care about peasant resolutions.

And i never said rumors about pol10 are correct.
 

maddie

Diamond Member
Jul 18, 2010
4,787
4,771
136
Is he serious? Is Roy Taylor really claiming NVIDIA won't have mid-range and low-end Pascal chips, too?

There have been spy-shots of GP106 and obviously there is GP104 coming too which isn't going to go into a $1k Titan card.

By the way, kinda funny coming from a rep of a company that just released a $1500 card for VR
. LOL!
Did they? Well foolish me, I thought it was mainly for developers and maybe the 0.0001 percenters with more money than ideas on how to spend it.
 

kraatus77

Senior member
Aug 26, 2015
266
59
101
Just not getting it are you...

Hint - This whole post has nothing to do with the topic at hand.

Hint2 - Rumors. What do they represent?

Hint3 - Timing. Are rumors for predicting a year ahead?

Keep at it, you'll understand soon enough!
I do know what you are trying to say. that furyx wasn't faster than tx on launch. thus i said " people don't buy gpus on launch day and throw it in dustbin afterwards" means even though furyx was slow, now it's faster than tx which rumors also said that time. and most likely will happen to pol10 too. it's pretty clear Amd keeps improving. (hint - 7970 which is now equal to 780 in many games )

anyway it's getting too ot, so im leaving at this.
 

Adored

Senior member
Mar 24, 2016
256
1
16
The article writer also has some ignorant beliefs. He's obviously clueless about the relationship between Mantle, DX12 and Vulkan.

"I'd argue AMD hasn't always ploughed what money it has into the right areas. Its current plan to kickstart a VR controller ecosystem—as revealed during Taylor's keynote speech at VRWC—seems misguided, particularly when history has shown developers and consumers prefer a single standard over an overwhelming amount of choice. And then there's Mantle, the short-lived but influential API that brought low-level hardware access to PC developers, but ultimately failed to take off in the wake of DirectX 12 and Vulkan."

Perhaps unsurprisingly also appears oblivious to LiquidVR being based on Mantle.
 

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
Yeap that concludes the R9 390/GTX 970 performance at lower price. Full stop.

Ya, which is nowhere close to 980Ti level of performance for $299. It was already possible to get 390/970 level of performance years ago. Countless after-market R9 290/290X on sale. Recent times had R9 390 for $275 which after selling the Hitman game coupon for $30 US translated into a $245 390. There have been deals for $240-245 GTX970 in the US. $249 Polaris 10 with 390 level of performance is only great for people outside the US who cannot regularly get these deals or OEMs/users with crap power supplies. For people who regularly visit this forum, if Polaris 10 is $249 and is only as good as a 390, that's mediocre.

As I repeated many times ago, someone who looks in the Hot Deals, could have purchased a Sapphire R9 290 Tri-X with 3 games for $200 and PowerColor PCS+ R9 290X for $255 around November 2014.

Using today's market prices of R9 390, R9 390X and 980 is almost BS imho to justify Polaris 10's price/performance because all 3 of those cards are flat out overpriced. If you look at market prices of R9 390/390X, they have barely budged from their MSRP about a year ago. In the old days when AMD/GPU passed on manufacturing node savings to us, R9 390/390X would already be $249 and $329 cards a year later.

This going to be another HD4870 vs GTX 280, unless NV will lower the price of the top cards to $500.

No, not even close. This comparison is completely flawed. 4870 and GTX280 were flagship cards of their generation, only superseded by marginally faster HD4890/GTX285 before Fermi/HD5870 launched. Not a chance that Polaris 10 is like 4870 and GP104 is like 280. Also, if Polaris 10 is only going to reach GTX980/390X's performance tops, 1080 will smoke it by much more than GTX280 beat HD4870.

GTX280 beat HD4870 by about 18-20%.



If NV releases 1080 that's only 25% faster than the 980Ti, that's almost 70% faster than the R9 390. (1440p = 83% x 1.25 / 62% = 67%)
https://www.techpowerup.com/reviews/Gigabyte/GTX_980_Ti_XtremeGaming/23.html

Roy Taylor must be dreaming if he thinks the 3 rumoured GP104 cards will land in the $600-1000 US price ranges.

If this is HD 4870 all over again, I'd buy two!

EDIT: It has to be 90-110% of whatever NV puts out, if both are aiming for some where near 980 Ti, I'd go for the better perf / price option like I did last time.)

Ya, which is inconsistent with everything that's coming out from AMD => Bring VR spec to much lower prices than $349. He continues reiterating VR spec around 290X/390/970 level. If he wanted to tell us they are bringing flagship performance at mid-range prices, he'd tell us Fury X/980Ti level of performance at more affordable prices. Who knows of course if Polaris 10 can scale from 110-175W TDP it could reach Fury/Fury X performance but hints after hints from AMD imply this is a low-end to low mid-range part.

Also, simply based on rumors of GP104 beating 980Ti by 25-30% at high resolutions and the best case rumours of Polaris 10 matching Fury X or slightly beating it, again it's impossible to have a 4870 vs. 280 situation as I already posted above. For that to happen, Polaris 10 would need to be faster than 980Ti. Besides, HD4870 demolished previous gen AMD and NV flagships by miles. The comparison of 4870 vs. GTX280 being thrown around is used incorrectly.

I Believe GP104 (full die) will be more than 10-15% faster than GTX 980Ti. The thing to look at is the performance at DX-12 games, Polaris may be extremely close to GP104 at a significant lower price for the 1080/1440p resolutions.

I call BS on that. Pascal has pre-emption and should handle DX12 better than Maxwell.

The fact alone that Polaris 10 is rumoured to have 8Gbps GDDR5 (256GB/sec) while the high-end GP104 is rumoured to have 10-12Gbps GDDR5X (320GB/sec - 384GB/sec) should already be the biggest red flag that these products don't compete anywhere close to in the same class.

I don't know why AMD fans on these boards are trying to desperate jam Polaris 10 into performance classes it was never meant to be in.

In summary, AMD fans are ignoring:

- Polaris 10 will have lower memory bandwidth (clearly not a Tahiti successor)
- Polaris 10 will have smaller die size (clearly not a Tahiti successor for the new 14nm node)
- Polaris 10 is clearly being aimed at lower pricing segments to increase total install user-base
- Polaris 10 is clearly not AMD's high-end product. The 2304-2560 shaders vs. 4000+ for Vega 10 should be another red flag here
- Even if Polaris 10 matches Fury X, that's nowhere close to where after-market 1080 would be unless NV failed miserably or purposely gimped clocks speeds on the GPU

If you have 2x 290 i would wait for Vega, unless Polaris 10 at $299 is ~GTX980ti . You could sell your 2x 290 and buy a single Polaris 10 for the same price. Or, by 2x $199 Polaris 10 Pro and be faster than GP104 at lower cost when CF is supported

This is bad advice. 980Ti OC is already on average better than R9 295X2 for consistency if someone plays Day 1 launch games, which is what he mentioned. So how in the world is a $199 Polaris 10 Pro which in the best case ~ 390X be superior to say a $400-450 1070 OC? You are not making any sensible projections again. Besides, recommending two much lower-tier CF cards against a much faster single GPU is not great advice either. SLI/CF makes sense when it's something blatantly superior such as $600 R9 295X2 vs. $550 980 or if buying flagship cards.

I can't imagine Polaris 10 will be slower than a 390x

I can't imagine it being much faster either. Fury X is only 21% faster at 1440p against 390X using TPU's latest benchmarks. Start ethereum mining on those 290s and by May 31st, that's easily $100 US in profits that you can instead use towards a GP104 card that's a worthwhile upgrade.

What rumors point to means jack as long as it's performing better than nvidia.
because that's the thing which matters.

Sure but the chance of Polaris 10 performing as good as higher tier GP104 cards is close to 0%. What people on our forum are doing is trying to manipulate AMD's road-map because they cannot accept that Polaris 10 is not a GP104 competitor. They don't want to accept that AMD may need to lower prices on the entire Fiji line and suck it up until Vega.

This is actually a perfect scenario for NV and AMD. NV gets the entire $350-650 market all to itself while maximizing profit margins, while AMD regains market share in the $300 and below segments until GP106/107 launch.

Even common sense logic prevails here.

4096 Vega on paper is already 60-70% faster than Polaris 10 (this is because Vega 10 has even better perf/watt which suggests shaders are more linearly utilized vs. Fury X against 290X). So now if Polaris 10 is a GP104 competitor, do people expect Big Pascal to beat 1080 by 60-70% as well? It's not making any sense now.

The only way to reconcile these differences is that Polaris 10 is a much slower card than GP104 1080 and that it's not aimed at $350-650 markets. For some reason though this is hard to accept so let's start making up facts like 85% of GP104's performance for half price. :sneaky:

This is how AMD products are set up to fail on this forum. :thumbsdown:

Just not getting it are you...

Hint - This whole post has nothing to do with the topic at hand.

Hint2 - Rumors. What do they represent?

Hint3 - Timing. Are rumors for predicting a year ahead?

Keep at it, you'll understand soon enough!

His point is that Fury X was hyped to beat Fury X and at high resolutions it does stock vs. stock. Since not everyone buys a $650-1000 GPU to keep for 12 months, and many keep it for 3-4 years, Fury X isn't as bad as people make it sound when discussing stock vs. stock performance. He has a point here. Your point is that Fury X was hyped to beat Titan X by much more than a couple percentage points and that ultimately 980Ti's overclocking gave it the win anyway. So you also have a point.

Regardless, the hype behind Polaris 10 is far worse than Fury X. People expected Fury X to beat 980Ti at high resolutions and it certainly does -- stock vs. stock. OTOH, expecting Polaris 10 stock to offer 85% of 1080's stock performance (HD4870 vs. 280 situation) for half price is starting to leave any realm of reasonability.
 
Last edited:

Dribble

Platinum Member
Aug 9, 2005
2,076
611
136
The article writer also has some ignorant beliefs. He's obviously clueless about the relationship between Mantle, DX12 and Vulkan.

"I'd argue AMD hasn't always ploughed what money it has into the right areas. Its current plan to kickstart a VR controller ecosystem—as revealed during Taylor's keynote speech at VRWC—seems misguided, particularly when history has shown developers and consumers prefer a single standard over an overwhelming amount of choice. And then there's Mantle, the short-lived but influential API that brought low-level hardware access to PC developers, but ultimately failed to take off in the wake of DirectX 12 and Vulkan."

He says it was influential - what more should he have said? It cost AMD a lot of money to develop and in the end was killed, and then just given away for free to make Vulkan. AMD (the company not the charity) is meant to be investing to make money not giving stuff away for free.
 

Atreidin

Senior member
Mar 31, 2011
464
27
86
Yes, that's the new marketing spin...it's a "developer" product...right.

Considering the Pro Duo has workstation drivers, yes, it is targeted to developers. You can also play games on it. I assume you trash talk Quadro cards too?
 

Udgnim

Diamond Member
Apr 16, 2008
3,664
111
106
if Polaris 10 is actually equivalent to 980Ti, then all the more reason to hold out for Vega
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,361
136
Ya, which is nowhere close to 980Ti level of performance for $299.

Nobody talked about 390 performance for $299, he said 390/970 performance at lower price.

GTX280 beat HD4870 by about 18-20%.

So if GP104 at $650 is 20% faster than Polaris 10 at $299 in DX-12 Games , dont we have the same HD4870 vs GTX280 situation ??


If NV releases 1080 that's only 25% faster than the 980Ti, that's almost 70% faster than the R9 390. (1440p = 83% x 1.25 / 62% = 67%)
https://www.techpowerup.com/reviews/Gigabyte/GTX_980_Ti_XtremeGaming/23.html

That is only IF Polaris 10 performance is the same as R9 390.

Roy Taylor must be dreaming if he thinks the 3 rumoured GP104 cards will land in the $600-1000 US price ranges.

I dont see a 300mm2 die + GDDR-5X at lower than $500-600. Perhaps 1070 or whatever it will be called to be at $400 range.

I call BS on that. Pascal has pre-emption and should handle DX12 better than Maxwell.

If you only take DX-12 games for Pascall, you should do the same for the Polaris 10. Because R9 390 is close to GTX 980Ti in DX-12 games

The fact alone that Polaris 10 is rumoured to have 8Gbps GDDR5 (256GB/sec) while the high-end GP104 is rumoured to have 10-12Gbps GDDR5X (320GB/sec - 384GB/sec) should already be the biggest red flag that these products don't compete anywhere close to in the same class.

Nobody said they compete in the same segment/class.

I don't know why AMD fans on these boards are trying to desperate jam Polaris 10 into performance classes it was never meant to be in.

In summary, AMD fans are ignoring:

- Polaris 10 will have lower memory bandwidth (clearly not a Tahiti successor)
- Polaris 10 will have smaller die size (clearly not a Tahiti successor for the new 14nm node)
- Polaris 10 is clearly being aimed at lower pricing segments to increase total install user-base
- Polaris 10 is clearly not AMD's high-end product. The 2304-2560 shaders vs. 4000+ for Vega 10 should be another red flag here
- Even if Polaris 10 matches Fury X, that's nowhere close to where after-market 1080 would be unless NV failed miserably or purposely gimped clocks speeds on the GPU

As i have said before, Polaris 10 is in the Pitcairn segment/class with Hawaii-Fiji performance at lower price than both of them.



This is bad advice. 980Ti OC is already on average better than R9 295X2 for consistency if someone plays Day 1 launch games, which is what he mentioned. So how in the world is a $199 Polaris 10 Pro which in the best case ~ 390X be superior to say a $400-450 1070 OC? You are not making any sensible projections again. Besides, recommending two much lower-tier CF cards against a much faster single GPU is not great advice either. SLI/CF makes sense when it's something blatantly superior such as $600 R9 295X2 vs. $550 980 or if buying flagship cards.

I clearly said for games that support CF, I have also made that particular statement BOLD so it was clear in the first place. And yes in games that support CF, 2x $199 cards will be faster than a single OC $400-500 card.
 

airfathaaaaa

Senior member
Feb 12, 2016
692
12
81
The article writer also has some ignorant beliefs. He's obviously clueless about the relationship between Mantle, DX12 and Vulkan.

"I'd argue AMD hasn't always ploughed what money it has into the right areas. Its current plan to kickstart a VR controller ecosystem—as revealed during Taylor's keynote speech at VRWC—seems misguided, particularly when history has shown developers and consumers prefer a single standard over an overwhelming amount of choice. And then there's Mantle, the short-lived but influential API that brought low-level hardware access to PC developers, but ultimately failed to take off in the wake of DirectX 12 and Vulkan."

did roy actually told that mantle failed while its in both dx12 and vulkan? D:
 

Erenhardt

Diamond Member
Dec 1, 2012
3,251
105
101
Considering the Pro Duo has workstation drivers, yes, it is targeted to developers. You can also play games on it. I assume you trash talk Quadro cards too?

Contrary to Titan line which was a totally gaming card with gaming drivers.

Also, Polaris will not match FuryDuo, not even close.
 

Riek

Senior member
Dec 16, 2008
409
14
76
That is only IF Polaris 10 performance is the same as R9 390.
And only if 1080 is > 20% faster than 980ti. For which i haven't seen any proof. (We are also talking about a chip which only recently has some samples and for which no ram is available in volume)

But more interesting would be the 1070 which will have regular ram as some believe (rumours). Now we are talking about a chip that is close to 50% bigger than polaris 10 with the same available bandwidth.
 

airfathaaaaa

Senior member
Feb 12, 2016
692
12
81
Contrary to Titan line which was a totally gaming card with gaming drivers.

Also, Polaris will not match FuryDuo, not even close.
so titan was a gaming card and 980ti that is a cut down version of titan x is often ahead
what does this make it then?
 

linkgoron

Platinum Member
Mar 9, 2005
2,334
857
136
I have to agree with RS - it seems as though AMD products start with their expectations at one level (for example ~390 or 390x perf) and slowly creep to much higher expectations (for example almost ~980ti perf). That's what happened with the Fury X, and it seems to be happening with Polaris 10.

I would really like Polaris 10 to be a great chip, but these talks about 980ti perf. just seem unreasonable given all that we know. I wouldn't be surprised if these cards do indeed end up with the 470/480 naming. Sites are just spreading these rumors as there is basically nothing new to say about these cards.

BTW - while this is not the Pascal/1080 are the ~20% rumors are against the 980ti or against the 980ti OC? 20% from stock 980TI would put the 1080 at around 980ti OC versions, which is not that impressive IMO.
 
Last edited:

maddie

Diamond Member
Jul 18, 2010
4,787
4,771
136
He says it was influential - what more should he have said? It cost AMD a lot of money to develop and in the end was killed, and then just given away for free to make Vulkan. AMD (the company not the charity) is meant to be investing to make money not giving stuff away for free.
Mantle did exactly what it was created to do. Enabling the development of mainstream low level APIs allowing their GPUs to shine. Look at the wins in DX12.

Did you and the writer really expect financially constrained AMD to maintain a separate Mantle pathway going forward?

What he could have said, at the very least in the case of Vulkan, was that Mantle morphed into and is at the core of Vulkan. Vulkan would not be where it is, at present, without Mantle.
 
Status
Not open for further replies.
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |