AMD Radeon R9 300 Series MegaThread: FAQ and Resources

Page 46 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.


You don't know everything. There is infinite history in this world, and you, someone else on here, well, nobody knows everything.

There are people who claim Hitler died in his bunker. Where was their research? All evidence points to that he didn't. I've done my research on that. I have better means of my time than looking up marketing history of AMD in 2003 or whatever, if there even would be such a book. Maybe I'll read it when I'm having trouble sleeping. But to be fair, my sentence was that Nvidia seems to market more than AMD, which implies present tense, not past. And it was opinionated.
 
Pretty impressed with my 390X so far, got it in a spare rig right now with an i5 4440/16GB RAM and it runs games extremely well while keeping surprisingly cool in a mini-ITX case (380T); with 76C being the highest temp I've hit so far while gaming. Power draw for the system is ~450W. Not bad at all.
 
only reason to doubt a claim is the source or if the claim does not give concrete performance values.

will Polaris be that old V8 GCN we know with a bigger turbo and direct injection?

likely.

does that mean it cant be the fastest thing ever and live up to claims?

no.

example for engines again.

my 03 montie carlo has a 3.8L V6 @200 HP and gets 25mpg.

my new 2013 mustang 3.7L V6 has 305 HP and gets 30mpg.

granted some of that extra mpg comes from aerodynamics but most is from engine management and transmission gearing.

the HP though. almost the same size engine but pushes 50% more power running on less of the same fuel.

point is their claims come from a reliable company as most of amd's claims do seem to be within 5% accurate to independent testing, and they give you concrete numbers to power draw (mpg).

what I would dispute of their statement on the 86W vs 140W isn't if the wattages are bias toward other componets or if their testing methods are poor, but if the Polaris gpu is equalivant in performance over the broad range of games to the 950, which without drivers I cant see how it could be...
 


The only problem I see is that they are building up early hype by comparing it to what is out now when they know that when they launch this new GPU they will be facing something else. Much like the FX8150 that was compared by them to the 980X when it had to deal with SB-E and then Ivy Bridge when it actually launched.

It just builds up a false hype. If Pascal launches at the same time are you going to care how Polaris stacks up to Maxwell V2? Most people wont even care, they will worry about how it compares to Pascal. Much like when Zen hits people will compare it to Skylake/Kabylake/Canoinlake (whatever is out) and not to Haswell.
 
you have a valid point.

http://cdn.wccftech.com/wp-content/uploads/2016/01/AMD-Polaris-15.jpg

appears to have a few new componets and "built for finfet"

http://wccftech.com/amd-polaris/

they also have a claim that PCperspective guy was told first Polaris gpu to be out in 2 months.

too soon to be true?

likely.

jimmy amd cant compare to pascal because nobody knows how fast pascal is... amd compared the 8150 to the 980x because we knew how fast the 980x was. told me exactly what to expect.
 


I understand they don't know how fast Pascal is. Instead they should compare Polaris to GCN 1.2 and let reviewers show how they compare to what is out when they launch instead of building up a false hype.

And the FX series built up a false hype. It was compared to three different CPUs: an i5 2500K, an i7 2600k and a i7 980X and only in what it won compared to each respective CPU. It is marketing hype that does nothing but cause crap in the long run.
 
The point is it is a bit useless to compare with competition product when they actually comes out later they will be competing with different competitor product. It is more meaningful if they compare with they own old product instead to see actual gaon they get from their product.

And about the card launching in two months time is not that strange if you have been following gpu news all the time. Even nvidia big pascal was rumored to tape out as early as june last year. The interesting bit is AMD choose to release their low end card first before everthing else.
 


its not any more meaningful to me, and it can't be used to uneducated consumers as PR.

to say 460x will be 1/2 power use of 360x or 1/3 of 950 is the same thing. we know the 950 is more power efficient than the 360x. It makes amd look better to those who don't read the fine print or know specifics of what current cards are capable of to compare to competition, but makes to the educated crowd no difference.

to somebody who can do comparisons and math any data can be used to make relative comparisons to Polaris performance scaled up to any number of cores based off of even one benchmark within 15% accuracy. for example you can take that a gtx 950 uses 60W for 60 fps @1080p for said game and R7 460X uses 40W for same experience and you can scale that to similar card setups (aka same memory setup) and make the assumption that a R9 490x (the 3500ish core chip) will use 200W to the 980ti's 300W.

not that it will tell us if we should buy Polaris or pascal (which is what you were mad at 8150 comparisons for) as we don't know any data for pascal, but it does tell us a relative scale to what Polaris is capable of. this data isn't hype the same as 8150's data wasn't hype as the benchmarks they showed you were real, they are just to help you get some early knowledge on what they have been working on most.

im still waiting for more information before I give Polaris the stamp of approval as I don't care much about power use, but its a step in the right direction.

people! just remember this isn't telling you much useful information and there is no reason to be real worked up about it!




that's what I was thinking.

 
^You just have not been here for the hype train. Trust me this will keep getting bigger and bigger. Every time a new piece of hardware is coming the marketing media makes it worse than it should be by creating that false hype. It is then spewed forward by the more intense fanboys of the party.

I am always hopeful but I take everything with a grain of salt until TH, Anand and the others take a crack at it to give us real world numbers instead of cherry picked scenarios.
 


Well said.
 


I wasn't on Toms for BD hype but I was watching it all the same on other forums. I don't remember much marketing media hype so much as cherry picked benchmarks that I was leery of not telling the full scenario.

I do remember the fanboys though. plenty of people spamed forums about how BD would crush intel and how 8 cores were faster at everything than 4 cores hyperthreaded. I try to stay away from those sites. everyone here seems pretty honest with expectations and keeps final verdicts waiting until full reviews are out.



as always that's what we really want to see. the skylake review TH or anand (cant remember who wrote it) did comparing IPC at the end from SB all the way up was phenomenal! I want to see a review like that for zen and all graphics generations. get a card with same number of spu's and ram downclock them to same clock and test to see generational improvement. then show a stock vs stock comparision for same core counts and finally a top end card vs top end card to show what max is capable from the new gen. then break that down into a what the new gen improved with for example 5% from ipc and 5% from clockspeeds and 10% for more cores. we always have to wait till after release for those benchmarks.
 
jimmysmitty is right. the hype will get bigger and in the end when the expectation are not met what supposed to be a decent or even good product turn out to be a failure. take 285 for example. it is a decent step up from GCN 1.1. the first info about tonga was coming out in the form of firepro. the said card was rated at 150w. some people take this fact and start speculating about AMD power efficiency improvement for GCN 1.2. they said now AMD have something that can compete evenly with nvidia maxwell in terms of efficiency. some even predict that GCN 1.2 will be maxwell killer in term of power efficiency. so how that's turn out in the end?
 


Nvidia seem to prefer to do it via shills on public forums/message boards and reviewers, paid and unpaid, that's worse IMO.
Edit not to mention underground oceans etc
 
Well, in all fairness to AMD, Tonga in 2014 (was it?) was very good for the power usage and performance level it showed. That was was anti-hyped and no one really expected anything from it, but it did impress a lot of fellas out there. Unfortunately it wasn't a crown stealing card, so no "enthusiasts" were swayed by it. My point goes into saying AMD will now put a replacement people actually wants to see: 7750. It is VERY long due and the HTPC and lappy friendly fellas will be pleased with it, I'm sure. Even if nVidia has something better, AMD will more than likely regain market share thanks to it.

Focusing on the low end first is a smart move IMO. It's OEM focused on one hand and will drive bulk sales for them at the beginning. It might be better for AMD to scale up instead of scale down. Although... There is so much MPG you can get from a V8 block (GCN) before hitting a roadblock. Secret sauce not withstanding, I have good expectations for this new card.

Cheers!
 


Interesting that you mention that because AMD were exposed a few years back. 😀

http://alienbabeltech.com/old_abt/viewtopic.php?f=44&t=24247
 


...Who believed it? I'm the one who brought it up initially, and everyone simply jumps to the conclusion that they think turkey3_scratch believes it all. No, I just think it will be awesome if that is true. There are products in the world that have exceeded expectations (sliced bread), delivered to expectations, and delivered below expectations. I want both Pascal and Polaris to at least deliver to expectations. I'm not taking these statements as facts, but as someone who games I'm eagerly waiting for the upcoming cards from both sides, as I'll be picking between Polaris and Pascal.
 
well you better believe the power efficiency. I watched the demo... it does work on identical systems (aside from gpu)

point isn't that you believed it so much as you realized its importance.

this info does nothing to say how fast Polaris will be, just that its more power efficient than Maxwell 2.0 (the current most power efficient gpu uarch)

as a huge amd fan im sure ill get Polaris regardless as my Tahiti is getting a little dated, but would still love to have pascal for some testing as im sure all hardware enthusiasts on this page would love to do as well.
 


That didn;t say much more than we know and didn't really show us anything about the system setups except what settings really. I would like a Speccy or CPU-Z screen on both to show what the exact hardware setup is.

Until then it is all just marketing.
 
Heard that this new gpu will launch in Q1. Though in laptop they might appear much later. Well at the very least AMD do need immediate refresh to their low end offering. 750ti launch in feb 2014. And right now still the fastest card on sub 75w segment. For nvidia part they probably have no intention to come out with pascal based sub 75w card yet since they just launch 950. Though due to the it's low end nature it is not impossible for nvidia have the card ready already. The only problem is when is the correct time for it.
 


low end is said to release first for Polaris, so appears that the immediate fix is already in production, and we have a working demo at ces.

^^ only took 6 months to release Carrizo after their working demo 😛