AMD Radeon R9 300 Series MegaThread: FAQ and Resources

Page 56 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Amd reveal us so much info on polaris since january so many expect that amd will be ahead of nvidia launching new cards. And i thibk many hoping fully enable polaris 10 at least as fast as if not faster than fury x.
 
I hope Polaris 10 surprises me, because for the first time ever I'm debating on buying a Nvidia card. The 1080 looks to be a complete gaming powerhouse. Being that much better than the 980Ti and apparently even faster than the Titan X at nearly HALF the price... that's bordering on irresistible.

I really hope AMD can pull something out of the bag. Good competition means better things for us as the customers.
 
The 1070/1080 preview makes me believe even more that p10 and p11 are the 270/270x and 260x class cards, the top one around 390 level in dx12 but a bit better in dx11 and older api.
 


If the leaked specs are accurate, full fat polaris 10 should be essentially the same core config as Hawaii, with a narrower memory interface (but including the memory compression tech from Tonga + much higher frequency to close the gap). Then take into account the new version of GCN (rev 4), a new geometry engine and a process that should clock considerably higher than anything on 28nm based on NV's numbers, Polaris 10 should be comfortably faster than the 390x. If it isn't then AMD have got something very very wrong.

My prediction, Polaris 10 'X' should be around radeon Fury performance. Non X should be 390X equiv. That would place them both slightly below the 1070, so in the $200 - $300 range? My guess, Polaris 10 = 480X / 480 gpu's.

If things go really well for AMD, they might have a situation when the 'X' version slightly outpaces the 1070, with the non X just below. That will boil down to how well GF / Sammy's process clocks compared to TSMC though... If NV are pushing over 2ghz, AMD must be able to push higher than the circa 1ghz we've seen (I think that is probably a laptop part).
 


Current rumor is that the 1080 should be about 25% faster than the 980Ti. If true then AMD has a real challenge ahead of them and not even ASYNC can save them because, well ASYNC is not the best solution for every game so it will help in some and not in others.

Should be an interesting next few months. Either AMD launches early and works on drivers or launches later and loses sales.
 


I don't think they do though- the 1080 is a higher end GPU than Polaris 10. Almost everything (including AMD's own comments) point to that. They aren't releasing a direct competitor. The are releasing a smaller gpu that will go up against the 1060 / 1070 cards. So long as it's price competitive I don't see the issue. The number of sales of the $500 can't be that large.
 


If so then the only "solution" to the 1080 would be Vega 11 which is not slated for release till end of this year or even beginning of the next year because as it stands in most cases the 980Ti and Fury X is within 5% of eachother at 4K with the 980Ti taking 1440p and 1080p. Give the 1080 a 15% performance gain/advantage over the 980Ti and AMD again has nothing to compete with nVidis in the top end.

To me that only allows nVidia to bide it time and gain/keep market share until it launches its true Titan X/980Ti replacement with HBM2.
 
http://store.steampowered.com/hwsurvey/videocard/

I think AMD is right on the money on where to put Polaris. They need something to displace the 970 / 1070 market. By all accounts and means, I do agree they don't need to take the crown, just take the big chunks of market.

Cheers!
 


I ma not saying they are wrong. I am saying that they need to think of perception. The biggest issue they have is that they have not had a top end GPU that is consistently or vastly better than nVidia for a while.

We also have to consider that 1070 which is currently being seen as possibly near 980Ti performance.

I just think that AMD needs to try to put out a killer product. They need more than just the middle of the pack market share.
 


All true, but it's worth pointing out that AMD's stock price is higher than it's been in more than a year. Looks like people who bought in during their recent all-time low might be making some serious money!
 


Stock price means nothing though. TRhey hit a low of almost $2 before and jumped up when a new CPu or GPU was being launched. What matters is if the performance is there to compete on a level playing field with their competition.

What worries me is that all information points towards their new CPUs competing with older Intel products, even with Intel slowing down, and their "new" GPU competing with nVidias new mid high end. We all know that there will be a Titan variant of Pascal and a Ti variant so the 1070/1080 will be mid high end with the Ti being high end and the Titan variant being the "dear God I have way too much money and should not be allowed to buy things" GPU.

Guess we will have to see where Polaris 10 lays. I still think if they wait to long to answer the 1080 they will not gain any market share.
 
But my point, Jimmy, is that "the crown is not needed" for AMD to be "successful" this time around. Remember the 4870 effect? They need something like that again, badly.

Appealing to enthusiasts is good and all, but they *have* to win where the money actually is. That is in the 970 and now 1070 territory. Consumers just want good value at the end of the day. nVidia, at least at the top tier, is nothing of that. Titan gets quickly displaced by the 980ti and the 980 overshadowed by it's lesser sibling. Other than bragging rights, they don't really drive profit for nVidia in the consumer sector.

All of this does not mean I don't want AMD to succeed with Polaris, but most information around it, points to a slightly better Fury X performance at lower wattage, putting it on par with current nVidia top dogs. Again, nothing confirmed as usual, but the hope is the last thing to be lost, I guess, haha.

Cheers!
 
Well the thing is will amd willibg to compete head to head with nvidia? Back in HD4K era AMD are not cash strip as they are right now. Engaging price war with nvdia probably the last thing they want to do right now. So how they can avoid from going into price war with nvidia? By releasing card that perform in different class altogether.
 
Looks like the Pascal release may be causing AMD to re-think their release plans for Vega.
AMD VEGA10 now scheduled for October launch
http://www.guru3d.com/news-story/amd-vega-now-scheduled-for-october-launch.html


It looks like that means Nvidia is set to hold the performance crown undisputed until that October release.
"Vega10" is lined up to replace the 390/490 series. Vega11 will replace the FIJI Fury parts. Rumors right now indicate that Vega10 has 4096 stream processors and could compete with the GeForce GTX 1080 and GTX 1070. Vega11 would feature 6144 stream processors and would be lined against the GP100 big pascal GPU (think Titan).
Click to expand...
 
I would really love for that to happen, renz. AMD charging $600 USD when they actually deserve it, I would have no complaints with that at all, but the problem is what they have showed and the current rumor mile is, that won't happen. Polaris won't leap over the improved Hawaii (Fiji) used in the Fury siblings, at least, not with *very* drastic changes to GCN (IMO, that is). They could do for the next generation (Vega?), but I don't think they can with Polaris, given current available information.

That is why their best bet would be to pull another 4K series "win". Engaging in a price war is better than selling no cards at all. It's like saying, they are better off with the position they had when the HD4K series came out than the HD2K and HD3K series before it. And the side benefit of actually going into a price war, is they will get some brand recognition from it.

In any case, let's see if they release something additional. Maybe we will know where the new cards performance will land.

Cheers!

EDIT: Nice timing, 17seconds 😛
 

Problem is that Vega 10 is supposed to have HBM2 which is not available in full volume yet. Samsung is producing it but probably does not have enough to satisfy the needs of AMD and nVidia.

Plus this is a rumor from a post on another site so who knows what will actually happen.



Thing is you are looking at it from an tech stand point instead of a normal consumer stand point. Consumers don't look at it the same and while we can say what we want, we are always small fry compared to consumers and right now the buzz is that nVidia has the power house GPU coming out this month while AMD has no official day of release for Polaris 10 which right now is just a mid end GPU replacement and not a top end gaming GPU.

Lets just say that people have held off on new builds because of Pascal and Polaris and right now nVidia is ahead with the ball in AMDs court and everyone is waiting.
 
I still don't see the problem. If AMD were to release a massive die GPU again like Fiji on 14nm and it got spanked by the 1080 *then they would have a problem*. That isn't the case here though.

They are planning on launching a fairly conservative (i.e. smaller than Tonga) mid range gpu and an even smaller entry gpu to replace their oldest GPU's, whilst nVidia is releasing their much larger main gaining GPU. They aren't going to occupy the same price points, they aren't directly comparable as a result. I don't see the problem. If AMD have done everything correctly, the larger Polaris gpu will bring high performance down to a much lower price point than anything NV is offering (I'd expect the cut version of Polaris 10 to be in the £130 - £160 price bracket), where it is up against last gen NV and AMD stuff.

If they pull that off they'll have a value leader- and this time it won't be a respun old gpu with a load of drawbacks. I mean I think one of the reasons they lost *so much* market share with the 200 and 300 series was their entry and mid range options were so out of date. They sold the oldest GCN gpu until the launch of the 300 series and then it's replacement wasn't much better (basically being a copy with the missing features added in). They sold Pitcarin all the way through to now in one form or another. Those GPU lines *desparately need replacing* with something higher performing, lower power consumption and with all the new features AMD have added so they have a full line of GPU's with a common feature set.

What we'll have with Polaris is:
- New entry gpu to replace Bonair with performance hopefully pushing on R9 380 territory for the top part and better power consumption.
- New mid range gpu to replace Pitcarin / Tonga again with full feature set and last gen flagship performance and again great power consumption.
- The Fury X / Fury Pro Duo maintain the top end, and already have the key hardware features so True Audio, Freesync, Target Frame Rate Control and so on are common across whole range.

The key then is pricing, so long as the new gpu's are priced low NV has nothing to counter atm. The 1080 is a bigger die relying on first gen of new memory standard (a bit like fury), it's not going to be cheap. 1070 is using older memory to get price down but is same die so also isn't going to be profitable for NV at too low a price point (I don't see the 1070 going for much less than £300).

Polaris 10 is smaller than GP104 and isn't using fancy memory. AMD can price it to sell. Gamers aren't stupid, if you can get performance close to the 1070 for £50 less, what are you going to buy? If AMD are really lucky Polaris 10 will split the 1080 / 1070 cards- that would be a very nice place for them as they have a smaller, cheaper to produce, higher performing part. NV simply couldn't win that price war without hurting themselves in the process.

Edit: direct price conversion for 970 in GBP is £260, so add on import duties that brings us to £300.
 
^There is no problem. I am just stating that if AMD holds off too long they will not gain enough market share. They have not had a good chunk of the GPU market since Bitcoin became impossible to mine with GPUs and moved to ASICs.

I am just thinking as a normal consumer and most people are not going to be looking at the entry level. AMD already has a very tight competition with nVidia and if nVidia has a wide enough performance gap do you think people will want to wait longer to build a new system or buy the new king of the road?

The Fury Pro Duo wont be in most gaming systems when you can get two Fury Nanos or Fury Xs for 2/3rd of the price or two 980Tis for less and get better performance. It is priced way too high.

If current rumors are correct then Fury X will be relegated to high end and not top end which will drop sales for it unless people are either die hard Radeon fans or want the cheaper option which I am sure the Fury X will drop in price if the 1080 is pushing the 25% better than 980Ti performance it is rumored to currently have. The features you describe are not what sells GPUs. They are nice but very few use GPU audio, most use either a sound card or a USB DAC in their headphones. TFRC is nice but nVidia has it and they have a version of Freesync, a bit more but still not an "exclusive" feature that sells. What will sell is the FPS the GPU will push, especially at 1440P and 4K since those are the big resolutions right now that people want to move to.

Then again who am I to say. AMD has only just begun to not bleed tons of money so maybe they are on to something....
 


Actually Jimmy, I would argue that what I am saying makes *perfect* sense for a consumer: bring better value, not king performer. The Steam Survey is a reflection of a "value", putting the 970 at the top. AMD *needs* that.

I think cdrfk said it very nicely, so no need for me to go more in-depth.

Now, you do have a valid point. AMD needs to act fast. The more time passes, the more people will flock to the *current* value king instead. That might as well now be the 1070. We don't even need reviews to tell us that, haha. We all know that is going to happen.

Cheers!
 
There is no doubt that Nvidia has GP104-150 waiting in the wings for release to counter Polaris as a GTX 1060 (Ti). Also safe to say that they have GTX 1050's also ready to go when the time is right. It's very much like a chess game, always thinking several moves ahead.
 
17seconds and mousemonkey, that is the key fight imo. Polaris (either model) vs gp106 and gp108.

Gp104 is evidently a bigger gpu, although I think the rumours suggest polaris 10 is possibly in between 104 and 106 in terms of specification and performance, I guess like a gp105. I mean you are correct nv could counter with a cut down 104 if needed, but that is still a poor move financially, cutting a big die down, unless you have really poor yields... As always price will be the key differentiator, as much as performance, and hopefully power consumption won't be an issue this time.

I mean I know it's nice to see the two companies go at each other's throats for the top end, but I think the reality is sales volume are in the mid range bracket. I've never purchased a gpu for more than 150 quid in my life and I am an enthusiast (just one with limited funds). Thing is, my 150 quid r9 280 handles everything I throw at it fine at 1080p, albeit on high details rather than ultra.

Something like Polaris 10 at a good price might make a good upgrade. Last few generations I've not even considered the NV '60' parts simply because they were so cut back in comparison, can't see them ageing well with narrow 128 bit memory interfaces compensated for with silly memory clock speeds (i view life span as a serious issue with such cards when the vram wears out, have had that with older nv gpus in the past).

Hopefully nv will not release another hamstrung '60' part this time around, guess we'll find out soon.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom