• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Polaris architecture – GCN 4.0

The only problem with the GTX 1080 is AMD enthusiasts !!!!

If only loyal AMD enthusiasts would buy more Fiji and Hawaii based cards it would force NVidia to keep their prices lower.:D:)

"Putting my money where my mouth is since the 8800GTX"

;)

5850
7850
280X
Fury Tri-X
:D
 
AMD need to be focusing on the low and mid end or they will have more of the bulk GPU owners moving to Nvidia.

The bulk being GTX 970/390 owners, they are not going to buy another GTX 970/ R9 390 like card even if it is sub £200, those people want to step up for their money (about £300)

If AMD cant compete in performance with the 1070 at least their share will get dangerously close to single figures by October.

Like many other millions of GPU owners my next GPU is not going to be a third 290 perf like card in 3 years.
 
Last edited:
IMO,it is simply because Polaris is as sated by AMD, a mainstream and low-end part. It may be absolutely fantastic, better performance er watt than Pascal, and way better priced. But if they don't have cards that compete with the 1080 then they have a hard marketing message to do right now. people want to see a 1080 vs Polaris 10 benchmark leak, not a 970 vs 390 vs P10 showing Poalris smashing the old gen cards out the water.

Why a polaris 10 vs 1080 benchmark?

It's not competing on that level. Also If there was a 970 vs 390 vs P10 and it was "Smashing" those out of the water then I would be pretty excited about that. That could mean that the P10 was approaching Fury/Fury X performance, however an Nvidia users use of the term "Smashing" usually means about 2 fps
:p
 
IMO,it is simply because Polaris is as sated by AMD, a mainstream and low-end part. It may be absolutely fantastic, better performance er watt than Pascal, and way better priced. But if they don't have cards that compete with the 1080 then they have a hard marketing message to do right now. people want to see a 1080 vs Polaris 10 benchmark leak, not a 970 vs 390 vs P10 showing Poalris smashing the old gen cards out the water.

Even if P10 does not compete with the far more expensive 1080, if it has performance near to or matching the old top end at a far lower price then people will buy it.

Not everyone is in the market for the most expensive and top end card. And Polaris is targeting a far larger market than the 1080 is targeting.
 
Why a polaris 10 vs 1080 benchmark?
It's not competing on that level.
That is exactly my point. The most reliable information we have, i.e. straight form AMD, is that Polaris 10 is aiming at a difference market segment to the 1080. So there is nothing AMD can say or leak benchmarks out at this time that will stop people buying the 1080.

Also If there was a 970 vs 390 vs P10 and it was "Smashing" those out of the water then I would be pretty excited about that. That could mean that the P10 was approaching Fury/Fury X performance, however an Nvidia users use of the term "Smashing" usually means about 2 fps
:p

Indeed it would be ice, and is at the price point I would be buying at. However, AMD probably aren't loosing sales to nvidia for the 970 at this precise moment, the damage is already done I don't think AMD releasing such a benchmark will drive a lot of excitement and stop that many people buying a 9760/970, not least the 390 and 390x already do a very good job in terms of price-performance.
 
Even if P10 does not compete with the far more expensive 1080, if it has performance near to or matching the old top end at a far lower price then people will buy it.

Not everyone is in the market for the most expensive and top end card. And Polaris is targeting a far larger market than the 1080 is targeting.

I agree but if AMD leak out an internal benchmarks showing that Polaris 10 is similar speeds to a FuryX it wont stop people buying a 1080. They are different markets.


i'm just coming up with reasons why AMD might nt be releasing benchmarks and churning the runours mill regarding Polaris. the simplest explanation given the evidence is the cards are just in different market positions so there is nothing AMD can do to counter the 1080 sales. It might just make AMD look worse if they release performance figures early without a proper release so the public firmly know what Polaris is and isn't.
 
The only problem with the GTX 1080 is AMD enthusiasts !!!!

If only loyal AMD enthusiasts would buy more Fiji and Hawaii based cards it would force NVidia to keep their prices lower.:D:)

Also once again your spouting utter nonsense mate, my last cards

5850
6950
7870
290 ref
290 Tri-x

Not to mention the 50 or so brand new AMD cards I bought when bitcoin mining, intact I have not bought an Nvidia card in over 12 or so years.

And I bet I'm not the only one
 
That is exactly my point. The most reliable information we have, i.e. straight form AMD, is that Polaris 10 is aiming at a difference market segment to the 1080. So there is nothing AMD can say or leak benchmarks out at this time that will stop people buying the 1080.

That was a conclusion drawn from Roy saying AMD are targeting the affordable VR market.

Given that at this point entry VR is GTX 970/R9 390, this is how people have concluded Polaris must be around 390 performance. which is one of many conclusions one could take from that comment

Here is the only thing we do know, the only benchmarks of Polaris that have surfaced are Laptop variant GPU's, those are at 390 performance.
the full fat Desktop parts as always are likely to be a great deal faster than their Laptop counterparts.
 
X1900xtx
Another x1900xtx for crossfire
Another x1900xtx for younger cousin
8800gts...boooo
3870x2
4870
3850 second comp
7970
270 second comp
Fury pro

What do I win?:) or are loyalty points subtracted for purchasing an NVIDIA card?

Edit: Just seen oldgamer out AMD'd me!
 
X1900xtx
Another x1900xtx for crossfire
Another x1900xtx for younger cousin
8800gts...boooo
3870x2
4870
3850 second comp
7970
270 second comp
Fury pro

What do I win?:) or are loyalty points subtracted for purchasing an NVIDIA card?

Edit: Just seen oldgamer out AMD'd me!

We could ask kaap or someone willing to make up a Fanboy of the month thread?
 
On the AMD call thing today apparently, taken from another forum.

This is what they were allowed to share:

FreeSync up to 4K120
DP 1.3 (and probably higher, but as the spec is not accepted they don't advertise with it)
HDMI 2.0
HDR Ready
Designed for DX12 and Vulkan
VR Ready (They are going to reveal new LiquidVR features on the Polaris launch)
14nm FF (2,5 again confirmed)
Full 4K60 Hardware H.265/HEVC encode & decode
Low latency 4k game streaming
AMD TrueAudio technology


Nvidia appears to only be stating DP 1.4 'ready', and only listing actual support for DP 1.3. They use the same bandwidth, up from DP 1.2, the only difference is the signal uses a different encoding so it looks like both support DP 1.4 in theory, if/when DP 1.4 is completely finalised, uses the expected spec and people support it everything new from here on in from both sides looks like it will support DP 1.4. The difference appears to be AMD being more honest about it, stating flat out DP 1.3 and that they should support DP 1.4. Nvidia are going up on stage claiming DP 1.4 support rather than being more exact and saying they should be fully compatible.
 
On the AMD call thing today apparently, taken from another forum.




Nvidia appears to only be stating DP 1.4 'ready', and only listing actual support for DP 1.3. They use the same bandwidth, up from DP 1.2, the only difference is the signal uses a different encoding so it looks like both support DP 1.4 in theory, if/when DP 1.4 is completely finalised, uses the expected spec and people support it everything new from here on in from both sides looks like it will support DP 1.4. The difference appears to be AMD being more honest about it, stating flat out DP 1.3 and that they should support DP 1.4. Nvidia are going up on stage claiming DP 1.4 support rather than being more exact and saying they should be fully compatible.

DP 1.4 was finalised in march, which is why many people questioned Nvidias marketing of DP 1.4. But as you said the main difference is the compression tech in DP 1.4 along with support for BT2020.

Apparently AMD is supporting BT2020 over DP1.3 which is probably where the higher than DP1.3 spec comes from in terms of support.

I wonder if the new liquid VR features will be single pass rendering and Multi projection. Considering those two techs have been doable for years anyway. A few racing games and every single flight sim supports multi projection. Heck, most CAD type software uses multiprojection, its just a fancier name for using multiple view ports.
 
Last edited:
This is what they were allowed to share:

FreeSync up to 4K120
DP 1.3 (and probably higher, but as the spec is not accepted they don't advertise with it)
HDMI 2.0
HDR Ready
Designed for DX12 and Vulkan
VR Ready (They are going to reveal new LiquidVR features on the Polaris launch)
14nm FF (2,5 again confirmed)
Full 4K60 Hardware H.265/HEVC encode & decode
Low latency 4k game streaming
AMD TrueAudio technology

I just want to see the benchmarks so badly at this point (and of the GTX 1070 too).

It's interesting they keep pushing the 2.5x performance per watt now, since that's extremely good if true. Nvidia has managed ~1.56x performance per watt over the 980 with the 1080.

Still depends on what AMD is using as their benchmark for 2.5x though. According to Techpowerup, if AMD are referencing:


  • 270X, or Fury X/non-X/Nano - It'll be higher perf/W than the 1080
  • 290 X/non-X - It'll be almost identical perf/W
  • 390 X/non-X - It'll be mildly behind
  • 280 X/non-X, or 285 - It'll be noticeably behind, so you'd probably need 180W AMD to match 150W Nvidia, something like that.
 
When they talk about performance per watt, they are more than likely referring to the theoretical peak performance, which GCN4 should be closer to achieving in the majority of cases compared to GCN1-3. Especially if it is true that a 1280 SP P11 at 1ghz with 7ghz ram has 380-380x performance. That is quite a jump when 1280 GCN4 SP's now has the performance of 1800-2000 GCN3 SP's.

Also on another note, it appears they have bumped the ROP's per SP ratio back up to pre GCN3 levels and a bit higher. since this 1280 SP p11 is supposed to have 80TMU's and 40 ROPS. Pitcairn has 1280 SP, 80TMU's and 32 ROPS in comparisson.

comparing shaders per rop

This P11 = 32:1 (GCN4)
Pitcairn = 40:1 (GCN1)
Hawaii = 44:1 (GCN2)
Tonga = 64:1 (GCN3)
Fiji = Tonga

Pretty much doubled the Rops per shader compared to what they did with fiji. I think fiji would be pushing some higher FPS if it had the same shader to rop ratio as hawaii. Now vega should be very different if it does follow what this P11 part seems to imply.
 
Last edited:
Back
Top Bottom