>Pajeet rolls out AMD's "gaming scientist"
>Science man: "Here we have Polaris 10 playing Hitman on DirectX12 at 1440p 60 FPS"
>10 Seconds of him playing before he gets shot to death
>Pajeet: Wow that is amazing
Nice going AMD
>AMD will never get HBM2 because they'll be bankrupt by New Years
SHIRT CANNON
Hahahahahaaha, this is embarrasing, AMD and Nvidia are making the imposible to delay the biggest jump on PC in years.
Thanks consoles hahahahahaha PCtards on suicide watching hahaha
>>53493543
Sad, but this isn't really AMD or even Nvidia's fault this time around. Blame Hynix and Samsung for not being able to produce HBM2 fast enough.
Looks like I'm waiting for Vega. Should be cool since Zen comes out then anyway.
>>53493543
Uh, Pajeet works for nvidia...
>>53493543
>NO HBM2 TILL 2017
STICKY THIS
>>53493543
Thanks, Hynix and Samsung.
>>53493585
You know it's true
>>53493833
In 2018 AMD will change HMB2 memory to another new memory, so there is no point on getting that HMB2 shit now.
guys what's the diferencces betwen hbm 1, hbm 2 & gddr5 in layman's term?
>>53494182
GDDR5 runs really quick, but is thin
HBM1 is slow, but super fucking wide so it ends up better; maximum of 1gb per chip
HBM2 has the same speed and width, but a maximum of 4gb per chip
Fury X has four chips, with HBM1 thats 4gb; with HBM2 that four chips could allow 16gb. More likely we'll see 8gb (2gb per chip) for consumer cards though.
They could of course make a design with more chips of HBM1 as a stopgap.
>>53494182
HBM1 is much faster then GDDR5 but limited in size to about 4GB.
HBM2 is supposed to be even faster and to at least double capacity.
>>53493585
>followed by NVidia being split into 3 as a result of anti-trust suit, GPU progress set back 5 years
>>53494182
>>53494240
oh forgot to mention
HBM2 also can transfer an extra time per clock; which typically results in it being twice as fast despite having the same clock speed and width.
There is also GDDR5X, which does a similar thing of more bits per transfer, but will run at a lower clock than GDDR5 at first. It's more traditional and I wouldn't be surprised if this first Polaris cards use it.
>>53494182
what other anons said and also how its manufactured and structured
2d vs 3d
so what will happen to the gpu market if amd went bankrupt?
will jewvidya ruled the market or is there any law preventing that very same situation ?
>>53494336
500+ dollar midrange cards.
>>53494240
>>53494241
>>53494287
so what is this gddr5x i keep seeing ?
is it gddr5 2.0 or entirely new ?
>>53494240
>GDDR5 runs really quick, but is thin
HBM1 is slow, but super fucking wide so it ends up better; maximum of 1gb per chip
HBM2 has the same speed and width, but a maximum of 4gb per chip
>Fury X has four chips, with HBM1 thats 4gb; with HBM2 that four chips could allow 16gb. More likely we'll see 8gb (2gb per chip) for consumer cards though
what chip ? vram or ????
how does this chip affect game performance in general?
>>53494367
Gddr5x is pretty close to HBM1 in performance.
>>53494336
AMD has a bunch of assets they would sell before going bankrupt, another company wanting to get into the game would buy them or just outright buy AMD.
>>53493921
audiophiles disgust me
So is Polaris 11 supposed to be the successor to the Fury X? Any news on pricing?
I'll be honest, I'm disappointed. Especially after finding out that the 1080 is going to have GDDR5X5.
Might just wait another year for HBM2
>>53494506
I pity them.
It's a mental illness but they dont know they need help.
![Slowpokefeelsgood[1].jpg](https://i.imgur.com/9LXCm43.jpg "Slowpokefeelsgood[1].jpg")
>>53493734
>mfw I get to cruise on my shit card for 1-2 more years
>mfw I buy something 30% more powerful than the ps5 3 months after it comes out for $200
>>53494564
WAITING EVOLVED
THE WAIT IT'S MEANT TO BE PLAYED
Fuck everything.
When are the new generation of AMD cards due? I want HDMI 2.0 so I can upgrade my htpc. Need to put some price pressure on nvidia as well.
>>53494819
Kek my man. As cucked as AMD fanboys feel right now, theyll feel better after Nvidia does everything in their power to make sure they do the bare mininum to compete. We all lose. Thanks, Pajeet.
>>53494825
New gen in a few months, hbm2 gen 2017.
>>53494564
Polaris 11 is mid range, Polaris 10 is going to replace the Fury X.
what fucking games do you even need more graphics processing for?
They haven't made graphics better! they look the same or worse since crysis they just don't even try with optimizing anymore.
>>53494934
polaris 11 is low end and polaris 10 is midrang, vega will replace high end stuff
>>53494934
>>53494970
LOOKS LIKE IM WAITING ANOTHER YEAR
>>53494968
Yeah, dog shit coding definitely hasnt helped things. But don't worry! We'll hire more multicolored hair bull dykes! Then we'll have progressive code!
This is our bleak future.
>>53494997
should wait until nexgen memory desu senpai
>>53493543
Okay guys. Buy that 980/970/960/950/390/380/370 that you've always wanted. Because the good shit is still going to take a while. Unless GDDR5X is good enough for you.
>>53495090
GDDR5x is a marketing meme that will not be in any cards this year.
>>53495090
I can wait
>>53495090
I just might do this since the good cards likely won't be out until 2017.
>>53495120
Geh. What a terrible setback for 14nm.
Ok. I'll retract my statement. Wait for Polaris/Pascal. Then laugh at everyone else who is waiting for GDDR5X/HBM2.
>>53495204
waiting for anything in tech is stupid in general
buy the best value that can be overclocked for futureproofing
![tumblr_ll57usiaUF1qzj8r6_large[1].gif](https://i.imgur.com/oKMrcVI.gif "tumblr_ll57usiaUF1qzj8r6_large[1].gif")
>I didn't buy a 390 for $225 on BF
>I didn't buy a sapphire nitro 390 for $240 a couple weeks ago
>>53495090
Do ya'll think getting a AIO GTX 980 Ti will still be worth it? If not I'll probably drop a couple hundred on a 390X until Vega/1080 Ti
>>53495258
Sure, why not. It's either that or wait forever for 1080ti.
>>53495250
I know this feel
I know it all too well
When will see nvidicucks leak some real info about pascal?>
My titan black that I got brand new for $600 rides for another year.
>hueheueheueh
>>53495801
A titan black will be good for the next several years.
>tfw snapped up a 290 Vapor-X for 325 eurobucks when the 300 series launched
It was a good decision.
>meanwhile Nvidia cucks my 780ti with every new driver update
>>53493543
I don't see why everyone is so butt blasted about this. Gddt5x has similar bandwidth to HBM1. It just means the cards will have more power draw and be a bit larger. The new cards are still 2.5x more efficient than last gen.
>>53495906
>A titan black will be good for the next 2 or 3 driver updates.
>head mounted VR AIO on a freaking r7.
I thought that was a good 5 years away. Yeah, I know it was just a proof of concept, but still impressed especially after considering that VR sets soon to launch gonna need chinese supercomputer clusters to run smoothly.
>>53495090
I wouldn't mind slapping a 290 or 290x into my computer, but that shit's $550 canadian loonies
fricking redonkulous
>GTX 980Ti will still be the best card
WAITKEKS BTFO
>WAIT FOR PASCAL / POLARIS
>m-muh HBM!
regret nothing gtx 980ti
>>53494017
>imblaiing 2017
AMD IS FINISHED & BANKRUPT
NVIDIA FIRST TO HBM2
>Polaris will use HBM1
LOL ENJOY YOUR 4GB BOTTLENECK
TOLD YOU RETARDS HBM1 IS SHIT
>>53498856
>bottleneck
>512 bit bus
retards always make themselves obvious.
Will it be worth it to upgrade my GTX 460 to a Vega when it comes out?
>>53494264
you know that's not going to happen you fucking retard
>>53495906
>next several years
>when nvidia is known to kill their older GPUs when they come out with a new gen.
>>53499121
Don't be a fagot and wait for Navi, duh.
>invent HBM2
>keep that shit to myself.
>can take my time,
>samsung and hynix partner up with me
>competition salty because htey are stuck on gddrx5
>samsung and nvidia doesnt go to court and samsung doesnt have to supply hbm2 to nvidia
>hbm2 gpus come out fall 2016
>my cards outperform nvidia by 25-50%
> have gpuopen, dx12, and vulcan in control
>competiion still doesnt have true SFF cards
>cometition doesnt have full async.
But no, AMD had to fucking fuck up AGAIN
>>53493921
what.the.hell is that idiot doing?
>invent HBM2
>keep that shit to myself.
>can take my time,
>samsung and hynix partner up with me
>competition salty because htey are stuck on gddrx5
>samsung and nvidia doesnt go to court and samsung doesnt have to supply hbm2 to nvidia
>hbm2 gpus come out fall 2016
>my cards outperform competition by 25-50%
>have consoles, gpuopen, dx12, and vulcan in control
>competiion still doesnt have true SFF cards
>cometition doesnt have full async.
>no way we can lose this guys
But no, AMD had to fucking fuck this up as well.
>>53498856
HBM1 is not limited to 4GB.
HBM1 is limited to 1GB per stack.
Polaris 10 is likely to have more stacks than Fiji.
Nvidia only has GDDR5X in the same market segment
>>53499069
1024 bit
>>53498773
Nobody is getting HBM2. The fabs were late.
>>53498856
GDDR5X has twice the bandwidth of GDDR5. There will be no bottleneck although it will be more power hungry than it would be on HBM2.
>>53499899
Samsung and Hynix where late not AMD
>>53493543
did they say it would be 2017?
the graph isn't very precise and makes it look like late 2016
>>53500013
¨
yeah ok, but who gets shafted? AMD. they could have forseen it.
https://news.samsung.com/global/samsung-begins-mass-producing-worlds-fastest-dram-based-on-newest-high-bandwidth-memory-hbm-interface
Samshit isn't late, AMD's partner SK Hynix is very late with only Q3 mass production
AMD BTFO
Cucked by gameworks
Cucked by HBM
Feels good
novidia will start supporting freesync right guys? r-right?
>>53500035
AMD always said two Polaris chips in 2016, one in 2017.
Obviously Vega is highest end Polaris.
Obviously AMD is only using HBM2 on the highest end chips just like Nvidia.
This is not news. People just pretend it is.
Nvidia will not release their full Pascal lineup this year either.
>>53494506
>>53499897
If you newfags had actually seen the real video. He's not cringing at the audio quality. He's cringing at how shitty the quality is compared to how much he spent.
>>53500051
To late.
Won't be implemented until Q4 2016.
>>53500076
as long as they release something as good as the fury for <400$ I'm ok with it.
>>53500104
That's what they demoed yesterday.
Polaris 10 is Fiji-ish in performance
Polaris 11 is Tonga-ish in performance
>>53500076
But nvidia is stated to release 1080 in may/june.
if amd cant compete with it in may/june, amd is truly fucked. seriously, why would anyone wait?
>>53500144
1080 will only feature GDDR5X (if not GDDR5) and compete against Polaris 10, which will also come out in summer and probably feature HBM1
Nvidia will only release the Titan with HBM2.
>>53495250
i did, sucks to be you
>>53500144
it will be another maxwell, AMD is fucking up big time.
>>53500176
>1080 will only feature GDDR5X (if not GDDR5) and compete against Polaris 10, which will also come out in summer and probably feature HBM1
no it won't, you can only fit 4 1GB stacks of hbm1, any more reduces memory bandwidth to lower than gddr5 levels for half of the memory like the 660ti and 970 suffered. AMD will be forced into using gddr5.
>>53500073
in 5 years when marketing on gsync slows down and nobody will be buying gsync monitors
as things progress every single monitor on the market will support freesync for free while you still will have to pay additional 200 for same thing in gsync screen
http://www.pcper.com/news/Graphics-Cards/AMDs-Raja-Koduri-talks-moving-past-CrossFire-smaller-GPU-dies-HBM2-and-more
>Curious about the HBM2 inclusion in Vega on the roadmap and what that means for Polaris? Though he didn’t say it outright, it appears that Polaris will be using HBM1, leaving me to wonder about the memory capacity limitations inherent in that.
>Why is Polaris going to use HBM1? Raja pointed towards the extreme cost and expense of building the HBM ecosystem prepping the pipeline for the new memory technology as the culprit and AMD obviously wants to recoup some of that cost with another generation of GPU usage.
AMD IS FINISHED & BANKRUPT
NVIDIA FIRST TO HBM2
>>53500322
>you can only fit 4 1GB stacks of hbm1
wait wait, but isn't it only size to fit issue that doesn't matter at 14nm?
i thought hbm2 is about efficiency and even higher speed not node size
>>53500322
>AMD will be forced into using gddr5.
You mean gddr5x. It has 2x the bandwidth of regular gddr5.
>>53500322
>you can only fit 4 1GB stacks
That's only true for Fiji. Polaris is different
>>53500322
>it will be another maxwell, AMD is fucking up big time.
what do you mean?
And people are forgetting that AMD is phasing out HBM1. i dont think we will see more HBM1 cards.
>>53500420
doesn't matter if 'polaris is different' (it isn't), there's a massive tradeoff to having more than 4 stacks (massively decreased yields, much higher cost, more assembly steps for chink workers, etc). it's not going to happen, and HBM2 will fix the density problem anyway. there really isn't a point to continuing with HBM1 or trying to find workarounds to fit more than 4gb.
>>53500444
Why would more memory stacks have an impact on yield?
My only concern is when the fuck we get Displayport 1.4 on the cards. Why bother buying a graphics card now if it's not going to support your fancy 4k 120Hz VR a few years down the road?
>>53500538
because HBM uses many small chips to achieve it's bandwidth rather than a few large chips like other gpus using gddr5, each fury card using hbm1 used 4 256MB chips per stack, double the amount used in other conventional cards with the same capacity.
>>53500592
HBM2 just adds more chips to the stack so there is no significant cost difference between using 8 HBM1 stacks and 4 HBM2 stacks.
HBM1 has also been in mass production for over a year now. It should be available in quantity, unlike HBM2 which is only available for low volume $1000 cards.
>>53500658
>HBM2 just adds more chips to the stack so there is no significant cost difference between using 8 HBM1 stacks and 4 HBM2 stacks.
no it doesn't, the entire point of hbm2 was to allow denser chips, not more per stack.
>>53500144
Nvidia is stated to release their new GPU in q3 this year at the earliest.
They'll show off their first prototype in May or June.
>>53500690
Doesn't change the fact that HBM1 production lines are running while HBM2 will only be available in small quantities.
Great. AMD fucked up again.
And I was so sure they'd get it right this time I even bought AMD stock for 2000€.
>>53496610
He should just not update it anymore.
http://www.gputechconf.com/
Keynotes
Jen-Hsun Huang
CEO & Co-Founder, NVIDIA
Tues, April 5
09:00 AM PT
IT'S OVER, AMD IS FINISHED & BANKRUPT
>>53500420
Why don't they just photolothographize the hbm with the gpu? Or is the memory too big?
>>53500544
>4k 120hz
You're not getting that out of one card for awhile
The fury x2 maybe
>>53495258
>buying 5 year old 28nm
14nm is a big deal even without hbm2.
>help to invent HBM
>Nvidia releases cards with HBM2 first
AMD are really the cucks of the tech companies.
HBM is a dead end.
For 2018 they already have something way better than HBM.
>>53494710
Just like feminists
>>53495250
>$225
>$240
what are you talking about
the r9 390 is $400
>>53500035
baka desu senpai
>>53493543
sweet i was hoping my old 770 would last a bit longer.
only playing older games atm anyway.
only 2 more years
>>53501923
That "1x perf/watt" is most likely the first GCN1.2 chip in late 2014. This >>53500035 is correct.
I might get Polaris even without the HBM2 to be quite frankly honest. Navi as a next upgrade then will be a good upgrade.
>>53502141
What the fuck
>>53501889
>Choosing the most expensive store on purpose
The R9 390 can be gotten for 74627Ft+áfa which is $266 + VAT.
http://www.xplaza.hu/termek/gigabyte-234338-vga-gigabyte-pcie-r9-390-oc-8gb-ddr5
>>53502279
yes but ÁFA is 27%
74627 + 20150 (áfa) = 94 777
which is $340
but still not $220 and $260
also I was looking for sapphire nitro model and not >gigabyte
Is Polaris just going to be low to low high range?I want to get rid of my 970 as soon as possible.
After seeing that depressing livefeed from yesterday. I guess its time to start my nano build.
>>53502956
It looks so...Polaris is to be low power consumption and low heat GPUs aimed at the average user around $300 to $500 range.
>>53497974
r9 390 is already significantly better.
>>53497974
$10 more and get the Fury X
You 980Tifags sure love to rationalize your purchase. It's a great card for DX11, yeah, but it's getting blown to smithereens in DX12. Both the 390x and Fury are very very close to it already, the former being at least $200 less.
>>53503365
Fury X is 4gb VRAM and 10USD pricier, 980ti is 6gb and 10USD cheaper ,hmmmm what a hard decision to make.
>>53503408
HBM > GDDR5
>>53503370
390's great against 970 i agree but Fury X isn't gonna be VFM against 980ti .
>>53503433
that faster speed's not gonna substitute the additional 2 gigs.
IF my game/app's optimised for a full 6gb its not gonna cut it with faster 4 gb.
>>53503408
Let me correct this for you.
"Fury X is 4gb[HBM] VRAM and 10USD pricier, 980ti is 6gb[GDDR5] and 10USD cheaper ,hmmmm what a hard decision to make."
GDDR5->28GBs
HBM->128GBs
>>53503477
>faster speed's not gonna substitute the additional 2 gigs.
Imagine your moms mouth can fit 128 dicks and your mouth can fit 28 dicks. Both of you have 1 mouth. Your mom will finish off 128 dicks per second and you will have only sucked off 28. Bigger bandwidth.
>>53495801
Kek is Charlie still around?
Ages since I went to SA
>>53503540
oh ok, if you put it that way.
>>53503370
DX12 is doa, performance on RotTR with the DX12 patch and Hitman has been spotty for a lot of people, and Gears was shit and didn't even work on AMD. Microsoft fucked up, and AMD's performance on DX12 is irrelevant since Gameworks won and will be in almost every AAA game from now on.
>>53503540
From this day forth, this is how I am going to explain any form of bandwidth to people.
>>53503574
The 980ti is a better card in literally every way. You have to be a fucking mindless retard to buy a Fury X.
>>53503636
t. nvidiot.
>>53503636
>what is DX12/Vulkan/Async
>what is HBM
>what is low noise, space, and heat requirements
>what is the Fury for 500
Fury X isn't an all round superior card but is by no means all round inferior, you fucking mongrel.
>>53503949
>low noise, space, and heat
thats because its watercooled
>>53493543
I just want to upgrade my 780 to Pascal x80 with HBM2. ;_;
>>53503636
In life you will be a better man when you admit when you're wrong. It is also better to ask a question then to walk a path of lesser knowledge. Follow these 2 rules and you will make it in life. Trust me.
>>53504094
I just want to upgrade my Geforce 8600GTS to ~literally anything~ with HBM2. ;_;
>>53499185
It absolutely will happen if they don't buy AMD and save them you dumb faggot.
Microsoft only barely skated by last time and that was because they kept Apple from going under completely.
If they become the only game in town, there will be only one cure and that is splitting them up to create competition.
>>53493585
>AMD finally gets bought by Microsoft
Jesus I don't even want to think what would happen to PC gaming if Microsoft had it's foot in the GPU market.
I was going to upgrade my 280x to a 390.
Should i wait?
>>53504466
thats not a big upgrade
>>53504513
For someone other than you!
>>53504077
Yes.
Based on the timeline, it seems to me that most of these cards will probably be HBM cards made for the mainstream (sub 400) with a serious die shrink. I think this will probably mean you can get a sub 450 or sub 400 card with Fiji grade performance. The other cards will probably be Tonga shrinks with some minor core improvements. That's a good thing though, as it would mean a 20%-30% perf bonus for more than a hundred bucks cheaper and almost half the power.
>>53504466
You better be getting a 390x not a 390
>>53504466
Why not? Both of those cards are great, I'm holding up with the bad 670.
>>53504531
I upgraded from a shitty laptop
upgrading from 280x to 390 is not that big upgrade
you should >wait for polaris
>>53504754
Not bane?
>>53504642
>390x
cancer
Is waitfaggotry btfo? Should I just go buy a 980ti now?
>>53505019
wait™ pascal desu
>>53504647
>>53504754
Yeah i'm going to wait and save that money so i can get a cintiq/huion or a new cellphone since my moto g 1st gen ia diying.
Thanks anon.
>>53504801
I intentionally did not responded with baneposting
>>53505019
Yes good goy, buy often, buy worse!
>>53495090
i bought a GTX 980
>>53505019
>980ti
>not fury x
an hero please
>>53504324
Nvidia would collectively shit their pants if this happened
>>53505201
but the ti isn't faster?
>>53505019
Even if you want a top-tier card right now, it's better to wait and hope the price lowers a bit, or you might be able to get one used since people will be selling their old cards and buying new ones.
>>53495250
>dodged a bullet
>twice
be thankful.
>>53505266
faster a bit for now
1-2 years later the fury X will be faster
just look at r9 200 and the gtx 700 series
>>53505266
It's not, except in non-patched gameworks games.
>>53505269
>buying a GPU used
Retarded desu senpai.
Enjoy your 1 month lifespan GPUsed.
GeForce GTX 650 here. Would it make sense to wait it out or upgrade now?
>>53505336
Wait. When polaris/pascal is out, decide if you want to wait for HBM2 or buy. Don't wait until 2019 for "next-gen memory" or whatever, it's long enough to use up a GPU's worth by then.
>waiting this long
For fuck sake my 5770 is on life support for 2 years already. It cant hold on anymore
>>53505359
The next generation launch will drop prices on the current gen at least, and we'll still be looking at significant performance improvements if the slides are accurate.
Wait to see what Polaris/Pascal have to offer, then decide.
>>53505359
You can get a filler card for the time being. I have seen GTX 760's going for $50 used on craigslist locally and $90 on eBay. It is not going to win any races it'll be a hell of a lot better then that 5770.
>>53505289
>faster a bit for now
>wait 6 fucking month for driver optimization
good job AMD, good job
>>53505359
eBay is your friend after a new generation is launched. Lot's of people trying to ditch their last gens to get some money for the new gen. Way more supply then there is demand = cheap prices for a couple months.
>>53503297
>this is what amdfags actually believe
>>53505359
Dude, I'm on a 8600GTS. It was on life support 6 years ago. What modern game is going to be released in the next half year worth playing? (because we all know that's the only thing you use your GPU for, or you wouldn't care about upgrading). It's not like you need to buy $90 + tax + season pass + dlc + tip games at release either.
>>53504324
>>AMD finally gets bought by Microsoft
that sounds scary
>>53505359
Buy a 750Ti or something in the meantime.
>>53505444
>wait 6 fucking month for driver optimization
quite the opposite senpai
nvidia will gimp the 980ti and THEN the fury X will be better
>M-MUH HBM!
and what is that a problem? memories are only 5-10% of gpu performance.
AMD still eats dirt and Nvidia never had any bandwitch issues
>>53503370
>implying Ill ever fucking use windows 10
You wanna sacrifice all it takes to use win10, be my guest.
>>53494361
Isn't that what we have now?
>>53505570
>Nvidia never had any bandwitch issues
I can't believe the AMD Defense force faggotry going on in this thread
>Fury X is better, buy that!
>Except it literally isn't, in any benchmark
>BUT IT IS I SWERE
>ITLL GET BETTER OVER TIME
You people are how AMD stays in business.
>tfw stuck with nvidia because i play pso2
>>53505651
difference is 1-3 fps and same lead at 4k in every benchmark but with full dx12/vulkan support, while 980ti will have same fate as 780ti nobody wants it for free now
so i got 390 nitro in october for 1440p last purchase as lucky as this was i72600
feels good to be right
>>53505651
As time goes on, it's pretty likely that Fury X will be better in games that offer DX12 support over the 980ti but how much better remains to be seen. Also, there is no denying the 980ti pretty much crushes the FuryX in anything DX11 "optimized" right now especially at 1440p and 1080p.
The other thing to consider is that by the time DX12 really starts being more mainstream we may already have Polaris and Pascal out, with whatever memory configurations they end up having, BTFO'ing both Fury X and 980ti anyway.
There's no denying that waitfags are kinda BTFO today, though.
>>53505720
>1-3fps
My ass
>leads 4k
Who gives a flying fuck about 4k? Talk about one of the most forced memes in tech since the pentium 4. Shit wont be economical until fucking 2020 or better. By then 980 Ti will be ancient and waitfags will have died of anal tension
>dx12
>win10
>nuff said
>Vulkan
Anybody ever notice it's only AMDfags that rant endlessly about Vulkan? Wonder why that is.
>980 Ti same fate as 780 Ti
You mean AMDfags will meme endlessly about it like they did the 970 because that's all they have to comfort themselves with, is Memes? Because it sure isn't fucking good graphics cards.
I bought my 980 Ti a year ago. By the looks of things, It'll continue to be the best fucking GPU for the next 6 months to a year, It'll remain competative for the following year, and since I'm not upgrading to anything over 1080p till that shit becomes more affordable, a 980 Ti is exactly where I need to be.
>The biggest kicker
I actually got my 980 Ti for 350 bucks thanks to EVGA's upgrade program and 970 as a gift.
>meanwhile, AMDfags of all kinds are stuck with a 3 year warranty if they're lucky and no warranty at all if they're brainwashed by ASUS. Usually only 2 years.
Go ahead, let's see your response to all this shit. I Can't wait for the memes and illogical arguements.
>amd btfos itself again
>>53505823
>980ti on 1080p
Even at 144Hz that's pretty overkill but I don't really care about that. If you want to max out your games comfortably at 1080p and not have to worry about your card holding you back there is nothing wrong with that.
However, the fact that you only had to invest $350 in your card makes you pretty biased towards it. You got a great fucking deal. Now it's a year after you had your card and people are looking a the future landscape and you can't blame them if they don't want to pay fucking double what you did for the same exact card given Nvidia's recent track record of having their former flagship cards kind of fall off the map when the next gen hits.
>>53494367
gddr5x is slightly faster than gddr5.
high end gddr5 runs at 7ghz atm, gddr5x runs at like 10ghz. With gddr, the main thing influencing performance is the bit depth of the memory bus. So a 512 bit bus is gonna be key if you want to keep up with hbm.
>>53505651
>go on any Digital Foundry video
>comment section is full of AMDrones trying to start a GPU war, screaming and calling everyone shills
I don't know what's wrong with AMD owners, I don't want AMD to go out of business obviously but their fanbase is literally children.
>>53505910
>Rational argument
Well ok. That's fair.
The biggest thing I think, is people hanging on the idea that the fury X will get better. Yes Tahiti was a gold mine, the 290x can still *kind* of compete, but there is nothing promising that this trend will continue. It's irrational to hang on the idea that driver support and API changes will make a card better.
>>53505823
>Anybody ever notice it's only AMDfags that rant endlessly about Vulkan? Wonder why that is.
Also Linuxfags. It could finally make Linux gaming a thing, since it's platform independent. Unlike DX12.
>>53506033
>Linux gaming a thing
If that winds up being true, I'll eat my 980 Ti, then buy an AMD card and switch to linux full time.
I guess the only question now is if the 490/490x will get HBM1
Feels fucking bad man
>>53497007
> VR
You mean memeVR.
Just a screen in your face on a box
>>53506057
>Linux
>AMD
Might want to wait a bit longer, AMD is shit on Linux for gaming. Not even supported in the next LTS release of Ubuntu yet as they transition to a new driver. Sure, it "might" be better in the future, but that's how AMD always works, you're always left hoping it "might" be better eventually.
>>53506158
Yeah that's pretty much my whole argument.
>>53505823
>Who gives a flying fuck about 4k? Talk about one of the most forced memes in tech
how is higher resolution a "forced meme"?
>>53506236
You almost did it, you almost rused me. I as forming an arguement, and looking for a picture of factual evidence.
But you know what? Fuck you.
If you aren't rusing and you can't see why 4k is a big god damn joke, then you never will. The very same fucking mentality that makes you an AMDfag fanboy for life is the same stupid ass thinking that has you thinking 4k is great.
The only time 4k and higher resolutions will be good is when the screen is two inches from your face, and doubled for VR and that isn't coming any time soon. Ergo neither are relevant. Not to mention the fucking cost.
>>53506024
>people hanging on the idea that the fury X will get better
The newest version of Ashes of the Singularity has asynchronous compute fully implemented-- it wasn't complete in the previous versions.
The Fury X is 20% faster rendering this version of Ashes than the last.
>irrational to hang on the idea that driver support and API changes will make a card better
Both AMD AND Nvidia release drivers that improve GPU performance. Nvidia does this for individual games. It's not irrational at all to think that better software translates to better performance. Just look at how poor performance is on Linux for proof of this. You know what IS irrational, though, is to attempt to make the case that AMD cards aren't gaining performance from DX12 and Vulkan adoption when there are benchmarks available that clearly show that this is exactly what's happening.
>Rational argument
I don't think you're in any position to make judgments like that.
>>53493543
hbm2 is literally a meme. hb1 is still more than enough for everyone's needs.
>>53506024
That is also fair. I mean the definition of insanity is doing the same thing over and over again and expecting a different result, so expecting AMD to come through with a slam dunk on the driver/software side of things is optimistic at best and maybe crazy at worst.
One thing I'd point to is if you believe what's posted out there from developers, you'd have to come to the conclusion that AMD's drivers can't get much worse than the way they are now, so that might be some reason for hope if you're looking for them closing the gap as time goes on.
The reality is that we'll probably get something in the middle where AMD makes some gains due to better software support but will still trail Nvidia as far as being the company that has easier cards to work with.
I highly doubt a company as big as Nvidia is going to be incompetent when it comes to handling DX12.
>>53505530
Look the benchmarksof Fury X vs 980ti before the new Crimson drivers at the launch...
>Zen!
>Polaris!
>DX12!
>Vulkan!
How many times do you have to get boned before you stop believing AMD's lies?
>>53506324
Paid nvidiot please go.
>>53505580
>implying I even run Windows
>>53506324
>>53506024
>>53506320
Here's one of the articles detailing the new Ashes benchmarks:
http://www.anandtech.com/show/10067/ashes-of-the-singularity-revisited-beta
>>53506482
>>53506320
>GAME DEVELOPMENT FUNDED BY AMD
>>53494564
Polaris 11 replaces R7 360
Polaris 10 replaces R7 370
Vega 11 replaces R9 380
Vega 10 replaces R9 390
We have to wait until Navi for a Fury X successor.
>>53506508
Nvidia has access to parts of the source, just like AMD does. Nvidia can also contribute code to help their cards perform, so long as it doesn't hinder performance on other hardware.
Oxide has no bias against Nvidia, they're simply using the tools available to them. If Nvidia wanted their hardware to perform, maybe they should have developed an open source API. In lieu of doing that, perhaps they should have added some support for Mantle, seeing as Microsoft thought Mantle was good enough to add to DX, and the Khronos Group thought Mantle was good enough to throw away OpenGL and make an new API based on Mantle.
Nvidia are literally the only people in the computing world that don't want to use the APIs developed for Mantle. What does that tell you?
>>53506555
Excrement from a male cow.
>>53506402
I've only owned ATi/AMD cards since I bought a 9600 Pro to replace a GeForce Ti 4600 11 years ago. I always just buy what the best price/performer seems to be at the time I need a card and that's just happened to be red team. Fanboyism on either side is cancer 2 the max.
>>53506318
>The only time 4k and higher resolutions will be good is when the screen is two inches from your face
this is wrong.
there will come a day when you upgrade to 4k and see how much better it is but there's no point in arguing with you now since you made up your mind although you've never seen a 4k panel.
>>53506555
>R7
>relevant
>>53506591
>80%+ Market share
tells me that Mantle isn't relevant yet. Which is also why they didn't waste time on async compute and other dumb shit like that.
Their cards are targeted at 1080p/1440p and highish frame rates,
Pascal and then Volta will incorporate these other features only if they are necessary.
>>53506057
>I'll eat my 980 Ti, then buy an AMD card and switch to linux full time.
No need. Nvidia drivers support Vulkan.
>>53506682
>80% market share
>Mantle isn't relevant
Mantle got incorporated into DirectX 12. You know, the API that ships with Windows 10? You know, the one that works on Xbox One and PS4?
How the hell do you figure that it's not relevant?
>Pascal and then Volta will incorporate these other features only if they are necessary.
Free performance is unnecessary? Is that the case you're trying to make here?
>>53506555
>making things up: The Post
>>53506747
You can count the games that use this new API on one fucking hand, nigga.
The. Shit. Aint. Relevant.
...Yet.
A year or two, and every new game will use it, and in a year or two, it will have been THREE YEARS since the release of the 980 Ti and GP100 will be in full swing with all these fancy features and amazing support still kicking the ass off of AMD cards.
Meanwhile, all along, AMD suffered set backs and stupid shit like yesterdays blunder because, in part, they wasted time making sure their shit had support for an API that ISNT YET RELEVANT.
>>53506861
>You can count the games that use this new API on one fucking hand
Ignoring the 17 PC games that have announced DX12 support, there's every console game currently in the works. One hand?
Seriously, you need to relax on the Nvidia shilling. I own a GTX 960. I'm talking about an API.
>ISNT YET RELEVANT
I guess that's why Microsoft fell over themselves to bolt DX12 onto Gears of War, right? (for the record, I am aware this game is a flaming turd). Or maybe that's why DX12 just got added to RotTR (and before you say it doesn't do anything-- it does: the game runs much smoother and with fewer bugs on DX11 and AMD hardware).
DX12 is relevant RIGHT NOW whether you like it or not.
>>53495250
What's BF?
>>53507254
Black Fridee
>>53503560
Indeed he does.
>>53506861
>DX12 isn't relevant guys!
>every developer is tripping over themselves to add it
the only reason you wish it was irrelevant is because it gives a massive boost to AMD cards
>>53507330
It hasn't given a massive boost to anything yet since every game it's been implemented in has been implemented badly. RotTR, Hitman and GoW haven't experienced any magical performance increases from DX12 and in some cases run worse. Gears is one of the worst PC ports of all time despite being the DX12 flagship. DX12 is worse than irrelevant, it's a disaster.
>>53507399
>Gears of War
They took code from 2006 and bolted DX 12 onto it and you think it's representative of what DX12 is and can do? What the shit.
>RotTR, Hitman
I'm not so sure about Hitman, but in RorRT at least, the patch has made the game run much more fluidly than before. The minimum FPS has increased by an order of magnitude, in some cases. There are fewer onscreen bugs than before, as well. And that's with a hacked together patch, not even full support.
>disaster
>I DON'T LIKE IT AND NEITHER SHOULD YOU
Man, it's a new technology. Either get on board, or get out of the way.
So, now that AMD has disappointed me and Nvidia is going to release a 1080 with a 10% perf increase, should I drop £580 on a Classified?
I know the 1080 will probably be a disappointment to people running a 980Ti or Titan X, but it will probably be fine for me since I have a 960.
JUST WAIT bro, HBM2 will change everything just wait! Don't have fun now, WAIT. WAIT FOREVER, JUST WAIT BRO!
What are you doing bro? Wait bro! Don't hurt me! WAIT WAIT WAIT STOP! STOP!!!!!!!
>>53507810
HBM is a meme. Look at Fury X, it barely beats the GDDR5 980ti.
>>53508001
>>53507810
Ebin, ebin for the win.
Truly the waitfags are the only ones that are BTFO.
I'm sure we all had a part in it. I know I told a few waitfags come April/June HBM2 will be out.
Now it's 2017.
Fucking lol
>>53494240
>gb
Guys, when talking about individual memory chips please be more precise with your spelling. Gb and GB are different unit and both are used in conjunction with memory chips.
>>53508187
I'm only going to wait for pascal to release, sitting on a GTX 670 is pretty shit and HBM2 is probably not going to be worth waiting for since the next generation memory standard is only 2 years away
not mad, just dissapointed in AMD, I thought with 14nm combined with HBM2 they truly would have blown nvidia out this year but its probably going to be a stalemate
>>53508455
Calm down autismo, do you know what context is or do you want me to spell it out for you.
>>53508187
I think most people actually waited because of the die shrink, not HBM2.
>>53508482
I'm pretty sure you couldn't do that properly even if you wanted to, kiddo.
>>53508530
Here let us try, repeat after me
C - O - N - T - E - X - T
Now try it yourself.
>>53508561
Sure is kinder/g/arden here, kiddo.
Don't worry guys, it's just around the corner!
>>53493543
Then 2018 then 2019 then 2020 then never
>falling for the waiting meme
>>53506318
This is the dumbest shit I've seen all day, this is the kind of retard who buys iPhone because "it just werks"
>>53493543
>tfw bought a 980ti back in July
>tfw get to stay at high-end for the whole year nother year before upgrading
Who else /goodvalue/ here?
I don't understand these threads.
If you were expecting AMD to magically beat all their competition this year then don't be surprised when your unrealistic expectations don't happen.
Here's my personal view on it:
>Currently:
1. AMD is still leading on the APU market (CPU+GPU).
2. AMD is "1 step behind" Nvidia on dedicated GPU market.
3. AMD is "5 steps behind" Intel on CPU market.
>After 2016:
1. AMD is still leading on the APU market, perhaps gets a better lead on the CPU performance thanks to Zen, more on that in the (3).
2. AMD is still "1 step behind" Nvidia on dedicated GPU market.
Current announcements from both companies show that the trend from past years, which is basically "Nvidia opens up, AMD is late to the party but provides better long term solutions", will continue.
This of course is an unfavourable position, hence still "1 step behind". It is unlikely that they will lag further (2 steps behind) or even more unlikely they will catch up completely (even stride) but it's not out of question.
Basically AMD is not prioritizing GPU market this year, and a "rematch" from last year is what's most likely going to happen, with less mistakes on the way.
3. AMD is less than "5 steps behind" Intel on CPU market. This is because Zen is such a huge leap technologically, that even if it flops (Buldozer-tier) it's going to provide invaluable insight that *will* even the odds a bit.
Sure, Intel is still *far* ahead and that's illustrated by how many steps they are behind them. Yet still, some catching up is pretty much a guarantee at this point.
I'd estimate that after this year AMD will be 1-4 steps behind Intel, leaning towards 2-3 in my predictions, but who knows.
All-in-all: they stay relatively even on all markets while closing the CPU market gap (which, by the way, is their focus this year).
If that's not a successful year then I don't know what is.
Please don't have unrealistic expectations for Zen or Polaris.
You're up for disappointment.
>>53510182
underrated post
>>53510182
Bay Trail and Braswell fucking eviscerated AMD in the mobile market, the only growing sector in the CPU world. Puma doesn't stand a fucking chance against Bay Trail, which is the older 22nm-based SoC CPU, and Zen will likely not help that much because Braswell is so efficient. Zen+ won't stand a chance in hell either, since Braswell's successor will basically double down on the efficiency gains by the time it comes out. Because they can't cut it in the ultra-mobile and mobile market, AMD will continue to lose ground against Intel in the CPU realm.
There's a reason why 95% of all x64-based tablets are powered by Intel chips.
>>53510272
Apollo Lake(Goldmont) comes out Q3
Brand new TOCK CPU architecture, Skylake Gen 9 GPU & other improvements in 14nm SoC
AMD can't compete with Intel or Nvidia at all
>>53510272
Oh of course.
This is something that we will keep on seeing until AMD somehow reaches at least "1 step behind", which is most likely NOT happening this year, unless Zen exceeds everyone's expectations.
They will however make some progress towards this goal this year.
>>53510300
I agree they still can't really compete with Intel, bar "budget builds" which isn't real competition but market niche-filling.
Though I wouldn't say that they can't compete with Nvidia.
Sure they are behind, no shame in admitting that, but they undoubtedly were competitive last year - so why are you expecting them to fare worse this year?
Like they won't release new generation at all or it will be severely underwhelming? I haven't seen an announcement yet that would sway my predictions this way.
Point me in the right direction.
>>53494444
But AMD already did that with ATI and it only helped them sofar.
guys i want to switch to AMD but i can't realistically see myself waiting till 2017 to upgrade. unless polaris performance is god like then i might just buy a used 980 for SLI and call it a day till vega.
>>53499899
>>53499854
>implying AMD will ever be that much faster than its competion in the GPU or CPU sector.
>>53510595
Good thing you don't have to wait 'till 2017.
1. Wait for Polaris.
2. Nvidia's cards will probably be available by then.
3. Take your pick. No clear winner? Do your 980 SLI thing.
Profit!
>>53500073
yeah just as free as mantel was or how open and source it was right
>>53503540
Yes, but that doesn't not help you if you have to hold 50 apples for your job, when you can only hold 40.
Your boss will fire you and hire the guy who can hold 60 apples at a time.
You see there are cases where bandwith and cases were size is important.
What does 5 times the bandwith help you if it is not actually need, but your V/RAM is much to small to actually hold all the stuff that is needed.
Obviously shit will have to be discarded and other stuff has to be reloaded, so shit has to always be shoved from on part to another while waiting for slower components to thus slowing down that nice bandwith advantage.
>>53504586
RAM drwas like no power compared to most stuff.
So I always wonder where AMD supporters see these huge power draw improvements with HBM.
