>NVIDIA 16nm Pascal GP104 GPU Pictured in Full Detail – Powers 8 GB GeForce GTX 1080 and GeForce GTX 1070 Graphics Cards
http://videocardz.com/59266/nvidia-pascal-gp104-gpu-pictured-up-close
http://wccftech.com/nvidia-pascal-gp104-gpu-pictured-leaked/
>AMD: Polaris 10 mainstream desktop & high-end notebook, Polaris 11 for notebooks
http://videocardz.com/59248/amd-officially-confirms-polaris-10-and-polaris-11-market-positioning
>AMD Reports 2016 Q1 Results: Disappointment
http://www.guru3d.com/news-story/amd-reports-2016-first-quarter-results.html
http://www.anandtech.com/show/10270/amd-reports-q1-fy-2016-results-lower-revenue-but-new-ip-licensing-agreement-reached
>AMD’s Radeon Pro Duo Pictured In All Its Glory – $1500 US
>Three Fiji GPUs in a Box
http://wccftech.com/amd-radeon-pro-duo-glory-pictures/
>>54172284
>>Three Fiji GPUs in a Box
Is currytech smoking ignited shit again?
GP104 looks like a mid-ranged die, expect 980Ti performance or 40-50% of GP100 performance(which you won't see until next year anyway)
>>54172364
>40-50% of GP100 performance
Dream on, fuccboi. Big Pascal will be a 20-30% upgrade at best.
>>54172336
Probably three pins not three chips
>>54172570
>>54172336
No its 3 gpus. They give you a broken? GPU as like a collectible or some shit.
Maybe if you actually went to the link you would have seen that.
>>54172336
they give bad dies as a souvenir. shows how low the yield is for hbm (or maybe just Fiji in general)
>>54172284
>>54172336
>>54172763
>>54172901
HOLY SHIT. I have a Fury X but I'm incredibly jealous. Dat inner beauty... DAT DED CHIP
I would honestly love to have a dead Fiji chip just because.
>>54173455
Buy it in 3-4 years for nothing.
>>54173488
But that wouldn't be for free
>>54173488
See >>54173506
By that point, I'd be either over gaming entirely (my interest has been waning over the years) and thus wouldn't need such a powerful piece of hardware, or the newest piece of hardware would be much more important to me.
Whichever of the two it comes to be, I don't imagine spending that much on an old GPU just for a dead die would be worthwhile.
>>54172284
Fuck that's going to be much better than my 980ti
>>54172284
How much are these gonna cost, should I wait or just get a 970 now
>>54173598
Not sure
>GTX 980 Ti
2816 cores
>GTX 1080
2560 cores
>GTX 980
2048 core
Probably only cheaper
Man, I'd love a pro duo but I can't swing $1500 for it.
Plus it won't fit in the A4-SFX, kek
>>54173661
wait
What is the most possible entry price for "1070"?
I mean, I was planning to upgrade whole machine since few months, but you guys told me to wait, at least until Pascal will arrive.
I was planning to buy 960 before, now just wonder about that Pascal. I know that "1060" wont arrive yet but when could it?
I would probably import anyway, 960 cost around 220$ here.
>>54173860
Number of cores is irrelevant between different architectures. My 780 has 2304, yet is about half as powerful as a 980 Ti.
>>54174377
$300-$400 usd
>>54174427
If it would be 300$, it would still be lower price than what I would have to pay for 970 right now. But I would need to add transport cost.
I dont dont know, buy 960 in June or wait yet another year for 1060?
>>54174459
Suck some dick and get some extra cash
>>54172284
>shill "general"
fuck off back to /vg/ or reddit
>>54174459
1070 will cost around what an 970 costs now.
>>54174507
Shill what?
>>54174527
It will cost wherever its performance is, if it gets close to a 980ti, it will cost around $600.
It's like you people know nothing, the 780 launched at $700
>>54174507
Fuck off nvidiot
Fuck off ampoor
>>54174527
Yeah, but if I would want 970 right now, I would need to pay 1500 PLN, aka nearly 400$.
>>54174527
Nope. The 1070 will be priced around the 980's RRP and the 1080 bumped up to the 980 Ti's spot. They'll both move down next year when the Titan Supermegabest and 1080 Ti arrive. Absolutely no chance Nvidia passes up the opportunity for nearly a year of price gouging.
>>54174419
But the GP104 isn't a sort of Maxwell 2.0 @ 16nm?
>>54172901
lol
It's pretty cool though to be fair.
>>54172284
>AMD Reports 2016 Q1 Results: Disappointment
Except it's all the contrary, AMD overperformed
>>54172284
Too late to the market, we have just got this:
http://www.anandtech.com/show/10269/zotac-quietly-releases-pcie-x1-gt-710-graphics-card
>>54174954
Q1 2016
>Grossed = $832
>Operating losses = $68 milions
>Net losses = $109 millions
>Part = - $0,14
Q4 2015
>Grossed = $958
>Operating losses = $49 milions
>Net losses = $102 millions
>Part = - $0,13
Q1 2015
>Grossed = $1G
>Operating losses = $137 milions
>Net losses = $180 millions
>Part = - $0,23
They still don't make money
The grossed declines every quarter
Their net losses are higher than in Q4 2015
>>54175132
Except that they were expected to have higher loses in Q1 2016, they had much less than expected, along with the IP contract and contracts for the 3 next gen consoles, there's a reason their stock went up today
>>54174645
no. that's AMD FUD.
>>54175348
nah, both
Do you think Pascal will have AIO liquid cooled cards available at launch? That's the only reason I'd sell my 980 Ti's. They're loud and hot as hell. I don't want to wait a year for the board partners to finally release a good one though like that Gigabyte Waterforce one.
>>54177564
Liquid cooled cards are more of a thing now than before, so we might. If not, probably a few months after they launch
>>54177624
How does the TDP of these new cards compare to previous gen? Are they efficiency bullshit like AMD's new cards that will likely get away with air cooling?
>>54177664
With the die shrink they should be slightly more efficient than Maxwell
>>54177709
Fuck
>>54177715
Why fuck
>>54177731
Seems less likely that they'll make water cooled cards if they are more efficient.
>>54177748
There's watercooled 980tis all over the place
>>54177760
Reference card ones.
>>54177801
Oh ok. Yeah I doubt Nvidia would do that, they don't really need to.
Why would you want a reference water cooler though? Just buy a 3rd party one
>>54177564
>aio
Nop, it's not a housefire
>>54177564
>I don't want to wait a year for the board partners to finally release a good one though like that Gigabyte Waterforce one.
I bet one of those AIO merchants will release a mounting adaptor thing for it quickly enough so you could diy a 'hybrid' setup
I want to do a new build with a custom water loop (fucking aio pump whine reeeee) so I'll probably just buy reference and wait for EK or someone to release full-cover blocks
So if I game at 1440p / 60 Hz and I currently have a 770, should I buy the 1080, or wait until the 1080ti next year with HBM?
>>54178645
Also I'd prefer not to spend more than $600.
Guess i'm stuck with 1080 then?
I figure it'll still be 5x more powerful than the 770, so it shouldn't be too bad.
>>54178645
Do you want to wait longer for a more powerful card or not?
That's about it
>>54178690
There's sort of a game drought right now anyways so i'm in no big rush, but I don't want to sit around until march of next year for the 1080ti to come out, only for it to cost $900.
Then I'll have to wait 6 months for a price drop, when I should have just bought the 1080 all along.
>>54175303
>Be expected to lose a shitload of money
>Lose slightly less than a shitload of money
This is what is considered a success for AMD
Is it worth waiting for a pascal/polaris or should i get a 980ti
>>54178851
Wait
980ti is already a year old anyway
>>54178864
alright i'll do that.
i already have an i7 2600k, would there be throttle or can i use this cpu with a a pascal/polaris
>>54178890
Overclock it and you're good
>>54178924
awesome, i'll invest in a good cooler. thanks
>>54178645
1080 is good enough for current games and probably okay for a while to come. this is coming from someone who gripes all day about how shitty video cards are compared to modern resolutions and refresh rates.
>>54179098
I should add at higher refresh rates its a no go though unless the 1080 turns out to be a monster. Even my two oc 8gb 290X barely cut it for comfortable 1440p 144hz gaming and it definitely could be better.
>GDDR5X only in 1080
Fuck, I hoped for 1070 to have it too
I guess 1070 would be a good upgrade from 770.
I have 1200p/60 display, so I hope it will work for 4-5 years.
http://wccftech.com/nvidia-pascal-gp104-gpu-gddr5x-leak/
AMD's Polaris is going to walk all over NVidiashits pascall
>Smaller manufacturing process 14nm vs 16 for nm
>Better DX12 performance and Hardware Async Compute
>HBM1
Here goes the 490 mmmmmMMMMMMMMMMMMMM fap
I am looking to upgrade my 670 to a 970. Should I wait?
>>54180808
Polaris a shit and even AMD don't like the HBM1
>>54180850
I did the exact same transition recently. The 970 was just over £200, and it hasn't felt like a gigantic performance boost. I'd say wait.
My CPU is a 3570K clocked at 4.2GHz if you wish to compare that side of things too.
>>54180808
Oh boy...
>>54180870
Here is what I am working with right now. 4 years old. Want to upgrade so that I can maintain a healthy 60fps on High settings for BF4 and such.
>>54180864
why everyone parrots 800mghz 6000mghz 2300cus?
it doesn't make sense for polaris 11 to be stronger than polaris 10
all i care about we get to see 490x in june
there is no point to wait for vega if you go for upgrades every two years
>>54180909
>it doesn't make sense for polaris 11 to be stronger than polaris 10
Nop, 10 is stronger
>all i care about we get to see 490x in june
Fury X rebrand?
>>54180923
no rebrands, they reworked a lot of support hardware
CUs are somewhat the same but everything around it is new
>>54180808
>AMD's Polaris is going to walk all over NVidiashits pascall
Polaris are a mid range GPUs
>Smaller manufacturing process 14nm vs 16 for nm
The Samsung 14nm isn't mature also it's not really a true 14nm
>Better DX12 performance and Hardware Async Compute
Wait Pascal
>HBM1
4GB? No thanks.
>>54180948
>Polaris are a mid range GPUs
here is newsflash for you: gp104 is mid-range gpu as well
>The Samsung 14nm isn't mature
>trusting in TSMC
>>54180971
>gp104 is mid-range gpu as well
Polaris 11 and 10 are for laptops and mid rage cards, GP104 is the next GTX 1070/1080
http://videocardz.com/59248/amd-officially-confirms-polaris-10-and-polaris-11-market-positioning
>trusting in TSMC
It's the 1st time AMD use the Samsung 14nm FinFET
>>54180948
How is 14nm not a true 14nm because of "maturity"
>>54181029
>because of "maturity"
The SOC is not 14nm, only few transistors
>>54181006
490x is not mid range card as 1080
they can try and convince us however they want but $400 is not midrange
>>54178759
they are trying to steer an extremely slow moving titanic away from the iceberg, it's not easy
but if they manage it, everybody wins; NVIDIA and Intel don't get complacent
corporate tribalism is fucking stupid
>>54180971
Read that
http://www.techgrapple.com/apple-a9-14nm-vs-16nm-performance-test-with-iphone-6s/
>The Samsung Made A9 Chip is hotter than TSMC made; Samsung at 39.3 degree temperature and TSMC at 35 degree. Chip based on 14nm FinFET tend to get heat quicker than 16nm
>>54181051
yea, let me tell you a secret 16nm is not actually 16nm
>>54175067
why doe
>>54181060
>490x is not mid range card
I never said that...
The Polaris 11 and 10 (470x, 480/480x) are mid rage. We don't know what is the 490x.
>>54181065
Why the 14nm is hotter than the 16nm?>>54181064
>>54181064
This is an old article
>>54181120
>October 2015
I think it's the same foundry though.
>>54181084
we actually know that polaris 10 going to be a range of 480 to 490x
http://www.guru3d.com/news-story/new-amd-roadmap-gives-more-insight-in-polaris-10-and-11.html
AMD rumors always turn out to be true.
>>54181135
Anon, there's a large difference between LPE and LPP processes, plus at least 10 month difference as well as a completely different process when you compare tiny mobile socs to large GPUs.
As for why 14nm is 'hotter' than 16nm, it's denser and the logic is packed tighter together, heat dissipation is harder.
>>54181135
2w chip and 100w chips are two different stories
that article was debunked multiple times with larger sample sizes
>>54172284
>>AMD: Polaris 10 mainstream desktop & high-end notebook, Polaris 11 for notebooks
>>AMD Reports 2016 Q1 Results: Disappointment
Bye, bye AMD. You will not be missed.
>don't worry guys the console chip deals will save AMD
>>54181740
>the future.
You mean 980Ti performance at 100 less watts? Can't wait.
>>54181061
>NVIDIA and Intel don't get complacent
NVIDIA has moved away from desktop graphics as their main income for a long time.
And as for Intel, they just pretty much cut their PC CPU division in half, and I'm sure AMD will still find a way to fail even though they're basically competing against themselves at this point lmao.
>>54181766
AMD has 10000 employees, Intel around 100000, no matter how much Intel cuts their workforce they have more people working on the PC division and far bigger R&D
If Intel allows AMD to get to Broadwell/Haswell performance, then they're fucking shit, no company doing leading edge research should sweat over a company that has 15 times less money for research.
How much of a price drop can we expect from other cards once these ones are released?
Should I wait on picking up a 970?
>>54181800
Intel doesn't just make CPUs.
And they have a much larger marketing and support divisions than AMD - they kinda need them, because their shit actually sells.
As for AMD, they killed their own CPU business when they decided to go against the company that was the only reason anyone was buying their CPUs to begin with.
I've not thought of buying a shitty AMD CPU ever since NForce has not existed, and with good reason their own chipsets are garbo. Always have been, always will be.
Nvidia first to HBM2, first to GDDR5X
>>54181740
>AMD Reports 2016 Q1 Results: Disappointment
sure was dissapointed growing 40% after that report
>>54181800
intel doesn't have a wizard though
>>54181879
And AMD doesn't just make CPUs, your point?
AMD has had plenty of good chips everything from K7 to K10 has been decent to great.
Intel has had plenty of its own messes, but they had the money to quickly recover from it, AMD had to stick to Bulldozer for 5 years.
>>54181959
>they didn't have the money, it's not their fault
Maybe they shouldn't have bought Ati and ruined it if they didn't have money then.
A lot of AMDs problems can be traced back to AMD.
There's no market boogeyman, there's AMD being shit.
>>54181987
I'll love seeing your excuses once a company that has a 50 times smaller market value releases a line of competitive processors later in the year.
AMD IS BANKRUPT AND FINISHED
>>54182011
They'll be competitive with mid range Intel - they've never said anything else.
I'm neither impressed, nor significantly interested in garbo chips that can barely keep up with the high end on a slightly lower price point.
I'll give Intel a 300$ extra for a good chip any day.
>>54182044
They've said no such thing, they've said that their CPUs can handle 80% of the server market, which roughly translates to everything not needing heavy FPU workloads like AVX512.
Of course for your gaming build you suddenly really need that because Intel told you.
>>54182070
>lets listen to marketing blurbs
They've been saying they can handle the server business for years. Now it's magically going to happen amirite.
They can wipe their asses with their chips. No one is going to trust a multi million and sure as hell won't trust a billion dollar businesses on unproven processes manufactured on the cheap.
And as for me I've had an i7 3820 for years in my workstation and I've not thought of upgrading, let alone going to the lottery cpu manufacturer.
That's the laugh inducing shit people always say - hurr hurr Intel and Nvidia are the go to for gamers durr - when both Intel and Nvidia hold 99% of the pro segment.
People that make their money on a PC want good parts that won't blow the magic smoke. AMD can't deliver on that. Reality check.
>>54182124
So all you've got to say is "LALALA CANT HEAR YOU" ?
Good to know, see you in Q4, Mr. Professional shitposting on /g/ instead of doing his professional work-san
>>54182147
>work on Saturday
Better check your calendar once in a while.
Better check your calendar on when was the last time AMD was relevant for professionals as well.
Fucking armchair analysts of /g/ lmao.
So the 1080ti wont be seen until much later?
Like 5-6months from now?
I don't want to upgrade my 350Watt (gold standard efficiency) PSU.
It will come in handy, when the new cards are atleast as fast as a R9 390 and only have a TDP of less than 130 Watt.
>>54182147
>Good to know, see you in Q4,
amd always takes 15 months from tapeout to launch, so zen won't launch in q4. it'll probably be delayed into 2017, probably around this time next year.
>>54175067
Too long, not as kawaii as this card.
>>54182820
TDP =/= Energy draw
No chances of 1080ti having hbm2, or the new titan x?
What's GDDR5X? How does it compare to GDDR5 and HBM?
>>54183070
TDP=0.99x(Energy draw)
>>54183028
I would actually buy one of these if they could output 4K via DP.
>>54183028
>fans
LAFFIN
http://www.amazon.com/Sapphire-Ultimate-PCI-Express-Graphics-11215-04-40G/dp/B00ILALU28
>>54183684
>gpu
>no fan
Sure, if your case is rocking mad airflow, but if not then these cards are fucking housefires. Have a few GT 9600's at work with 'GIGABYTES PATENTED AIRFLOW TECHNOLOGY' that get so hot you could probably cook eggs on them.
>>54183711
god i hope you are not an engineer
>>54183711
>not having good airflow
?
>>54183770
I have had quite a few customers come in with these cards in their computers with no case fans whatsoever.
Ill only really consider an upgrade from my 980ti if they release the 1080ti or FuryX replacement.
Seems like they are dripfeeding releases again so im not holding my breath.
>>54172284
>GTX 1070
>falling for the 7.5 GB meme
>>54183777
It's not really for gaming, and even still, 0db.
>>54172336
I read it and its true, they just give you a seperate extra fiji chip
>>54183881
Still better than amd's 4gb
Do if I want a low end Nvidia, I should just buy it now? Pascal is basically just a die shrink of Maxwell anyway, so I'm not sure a 950 or even a 750 Ti is such a bad purchase.
>>54181944
>sure was dissapointed growing 40% after that report
it's the NASDAQ exchange.
Can someone explain how a GPU is bottlenecking the CPU and vice versa ?
I can't imagine that my i7 or FX 8350 going to have problems with the new released graphics cards.
>>54185931
>cpu can't feed commands and data to the gpu fast enough
or
>cpu always waiting for gpu to finish shit
http://www.amd.com/en-us/press-releases/Pages/press-release-2016apr21.aspx
http://www.tweaktown.com/news/51795/amd-positions-polaris-10-desktop-11-notebooks/index.html
>AMD demonstrated its “Polaris” 10 and 11 next-generation GPUs, with Polaris 11 targeting the notebook market and “Polaris” 10 aimed at the mainstream desktop and high-end gaming notebook segment
Polaris confirmed shit tier
>>54184624
?????????
AMD has had 8Gb for a while, ever heard of the 390 and 8g 290 brah?
>>54186562
He talk about Fury meme
>>54175067
>Fermi
>2016
why
>>54186562
>AMD has had 8Gb for a while, ever heard of the 390 and 8g 290 brah?
on housefire rebrands that don't support ASTC compression, which means you will end up fitting more in the upcoming 7.5gb than you ever will in the 390's 8gb
>NVIDIA FIRST TO GDDR5X
AMD BTFO!
AMD STOCKS ARE FALLING!
AMD ARE STUCK ON OLD GDDR5!
AMD ARE LITERALLY FUCKING FINISHED!
>>54186096
>nvidiots on suicide watch
feels good man. AMD stock is going up, low TDP GPUs and zen are just around the corner. it's AMDs year to shine.
>>54181800
>no company doing leading edge research should sweat over a company that has 15 times less money for research.
15x less and split between GPU and CPU.
>>54187659
>>zen are just around the corner. it's AMDs year to shine.
>>54180808
Polaris is mid-range. Vega will be walking all over Pascal.
Its AMD's plan to win the power struggle; let Nvidia release their powerful cards, see what they're like, release better cards.
>>54181987
>A lot of AMDs problems can be traced back to AMD's old management.
ftfy.
hector ruiz ruined the company by spinning off their fab and then jumping ship to go run it, but not before making an extremely long-term contract to force them into using the fab for a certain amount of wafers over a long time.
>>54187766
>>Polaris is mid-range. Vega will be walking all over Pascal.
Just like how Fury X beats Titan X. Oh wait, LOL!
>>54181987
Literally retarded. Their plan was to start entrenching AMD into HSA as well as custom SoC. The competitive edge for AMD products has never been able to overtake that of Intel because of a combination of issues, not just performance. Performance has only been a major issue for AMD in the past 4 years with Bulldozer. Radeon products are highly competitive with nVidia's offerings right now, and it's not even a big deal to them; console SoCs sell a lot more than discrete graphics cards, and are essentially vendor locked. Bulldozer's design isn't even that bad, but manufacturability was far less economic. CMT has a lot of advantages over SMT, and could've helped put AMD in a better position in the HPC market. However, AMD built big chips with low yields, ran hot, and performed worse for non-clustered consumer applications. Thus, AMD CPUs went to total shit. It remains to be seen if Zen can ameliorate these issues and actually swing the manufacturability and performance enough to beat out the marketing advantage Intel has.
>>54187792
Fiji was designed for 22nm process, but had to be pushed back to 28nm when TSMC fucked up.
>>54187651
>Meanwhile, Nvidia has NO HBM at all
>The first Nvidia release with HBM will cost $10,000 a piece
You nigs think a Zen CPU will have an iGPU capable of running 10bit, h265, 60fps 4k video?
>>54188348
I hope it won't have a retarded igpu at all
>>54188348
I wish they'd stop putting iGPU in there so we maybe get better prices
>>54188348
the cpus don't have an iGPU, the APUs do.
They should though, as long as they have a 10-bit h265 decoder.
>>54188367
>>54188404
Exept it will. I just wish it's good enough to run those type of videos.
Lmao that adoredtv had gotten to atmosphere levels of delusion. He thinks consoles will kill nvidia and Intel off to a point where amd will rule the market because of consoles.
>>54188512
This is probably the last we see of cumsoles anyway.
>>54187766
not gonna happen. the fact that polaris 10 is only 200mm2 probably means samsung 14nm isn't good enough for larger dies, it's suicide for AMD to target the OEM market like this considering OEMs never buy dgpus anymore.
>>54188555
Or it could be that they wanted to design & release the mid end cards first, which believe it or not, is where the money ACTUALLY is for the DGPU market.
>>54188555
Or maybe they want small dies, high yields
, low power and good margins.
Like they did with Evergreen.
If a notable smaller GPU is competitve, then this is AMD's win.
>>54188579
>which believe it or not, is where the money ACTUALLY is for the DGPU market.
polaris 10 is literally gtx950 performance, gaymers don't want cards that perform like that and igpus already fill that role for OEMs. it's been a dead segment of the market for ~5 years now.
>>54188534
The guy was saying he lost sleep over this shit lmao. He should commit suicide if amd fail like they inevitably will. His cancerous fan boy videos need to stop. He's taking this whole amd/intel/nvidia thing like it's some new 9/11 conspiracy.
>using NVidia
>any year
>>54188630
No its not (950 performance). Keep dreaming though.
>>54188665
benchmarks already leaked, we know that polaris 11 is 1/4th the perf of a 950 and that polaris 10 is on the level of a 950, idiot.
>>54188630
>polaris 10 is literally gtx950 performance.
Prove it, AMD already demoed Hitman running 1440p at stable 60fps, that's FuryX 980Ti turf.
Where is your evidence?
>>54188684
>Where is your evidence?
my evidence is under NDA for 6 more weeks. you are going to be disappointed.
> AMD already demoed Hitman running 1440p at stable 60fps, that's FuryX 980Ti turf.
demo was faked. the highest unlocked polaris 10 card is approx 380 performance with a maximum overclock.
>>54188679
Source or GTFO.
What a bullshit claim to stake.
And >Inb4 you're going to show us mobile GPU's and not desktop GPU's
>>54188679
Makes as much sense as the 'GTX380' being 60% faster than a 5870.
>>54188708
>NO EVIDENCE
Pajeet my shill, would you kindly go to the nearest designated shitting street and start eating it like a buffet.
>>54188708
Your uncle works for nintendo too?
>>54188731
kek, enjoy your r7 460 with 4gb of gddr5 on a 128 bit son
>>54188751
LOL okay.
>>54188767
POORFAG
POORFAG
>>54188767
>falling for the glued ihs meme
>falling for the 4gb of hbm1 meme
holy shit, no wonder there's so much amd damage control going on
>>54188781
>i7 4790k, Fury X, 2560x1440~144hz monitor & SSD
>Poor
Nope. Being poor would be getting 2 or more Titan X's.
>>54188787
If he bought a normal Fury he did good.
>>54188858
not really. it still has no OC headroom and struggles to perform the same as a reference 980 despite costing more.
the only positive the fury non-x cards have is that you can unlock the disabled shaders on some of them
>>54188679
we know that one variant of polaris 11 is 1/4th of a 950.
we don't know how mature the drivers were
we don't know how cut down the chip was
we don't know the power draw of that chip
>>54188630
>>54188679
>>54188708
If we assume they got their 2.5x power draw reduction, the 45W chip demoed (assuming 40W for the rest of the system) was equivalent to a 112W chip, or about an R7 370.
45W is nothing, that's an easy scalable architecture, and likely one of the lowest end binnings.
>>54188914
It's faster than 980 across the board.
>>54188914
are you from early 2015? Fury non-x is vastly better than a 980. The 390X is 980's direct competition, and it competes hard.
>>54188958
It should be faster than the 980 considering it costs more
>>54188982
post unbiased benchmark scores
>>54188958
>>54188982
that's false, a reference 980 performs better than the fury x. with an OC the gap would be even wider.
https://www.youtube.com/watch?v=nhckKSoidFQ
>>54188982
>The 390X is 980's direct competition, and it competes hard.
the 390x barely even competes with the 970, it's competitiveness is even worse since it's double the price for the same performance.
>>54188787
meme meme meme
literally go hang yourself
>>54189019
http://anandtech.com/bench/product/1442?vs=1522
>>54189019
Fury is the 3rd fastest GPU on the market, followed by the Nano and then the 980.
4th if anyone counts the Titan
>>54188982
The 980 stomps the regular fury with a moderate oc and can even beat the fury x in some cases.
https://youtu.be/pDxTcvZ4T2k
The 390x with a max oc can barely beat a stock 980 (pic related)
>>54189048
anandtech has been faking numbers for AMD since the buyout, nothing trustworthy about that considering every other reviewer has results that disprove theirs.
Nvidia first to HBM2, first to GDDR5X
>>54189019
who is this no-name you linked?
>video posted July 2015
ah, ok. i was right, you are stuck in 2015.
Please wake up. Latest drivers have the Fury/X competing with the 980Ti and the 390X competing with the 980.
![IMG_20160423_204554121[1].jpg](https://i.imgur.com/ej3OjdQ.jpg "IMG_20160423_204554121[1].jpg")
>>54188858
Its an X.
>>54189076
Year old drivers and refernce vs non reference¿
Okay.
>>54189090
>Latest drivers have the Fury/X competing with the 980Ti and the 390X competing with the 980.
>he thinks drivers magically improve performance
the entire reason amd made mantle in the first place was to deny and absolve them of the problems their driver's had, they have not improved performance on that front since 2012 when they dumped their gaymen architecture and replaced it with one that's only good at number crunching.
>>54189091
Nice oven, you forgot to put the food on your plate though
>>54189091
Unless you need FijiXTs high shader throughput, bandwidth and texturefill you dungoofd for a gaming card, unless you got it really cheap.
>>54189096
Yeah exactly. It's not fair they're using a high quality binned msi 390x vs a regular reference 980. They should get an msi 980 as well and the gap will be even wider between the 390x and the 980.
POO
Glad I bought a 390X in January for $125 out of pocket. This shit is trash. I will just get a second one for around the same price when the rubes upgrade and have stellar performance on the cheap.
>>54189110
only a nvidfaggot is this seriously this retarded
>>54189110
>>54189128
Already ate it. Finished just a minute ago or so.
>>54189131
I got it because I wanted the absolute best AMD offering. I'd already gotten my monitor just a month or two before and wanted the longest lasting choice possible.
I was sitting on a 7950 and the idea of waiting for Fury air was something I considered, but then I decided to say fuck it, plunge for the extra. Part of the attraction is how beautiful the card looks (in my opinion, at least.)
>>54174377
Let me tell you something about waiting.
I built my first PC 3 years ago. I used a 760, intending it to last me a year at best before upgrading again with the launch of Maxwell and when the new console generation set a standard of sorts for what to expect in PC ports. Then Maxwell was delayed for a year. Then Maxwell wound up being utterly inconsequential. Three years later and my 760 is just now starting to really show its age because I had to lower some settings in DaS3 to medium.
Waiting is a meme.
>>54189110
This is true. Even the ubershill adoredtv admitted amd haven't improved their drivers in years. They got lucky with dx12 forcing the devs to do all the work that would have been done by amd on the driver side.
>>54189208
>says nothing negative or meme's about nvidia products
>is somehow more a shill than your cuck ass
When will they be released?
>>54189192
Why shove the plate in a case is what i'm wondering
>>54189312
Only thing AMD cards do better than NVidia cards, keeping things warm.
>>54189077
Uh-uh.
Same anandtech that couldn't recommend a 290?
Pure shilling.
>>54189312
Its not in the case at all, that's the reflection.
>>54189380
Did you seriously just make that fucking picture
![IMG_20160405_091606219[1].jpg](https://i.imgur.com/j6vst4j.jpg "IMG_20160405_091606219[1].jpg")
>>54189312
>>54189375
that was before anand left, ever since then their reviews have been blatantly faked
>>54189409
Yes.
Hello friends
>>54189272
>says nothing negative or meme's about nvidia products
>keeps calling out the death of nvidia due to blatant hate for them cucking his beloved amd out of sales year after year
>thinks amd are the only company that should have global domination
Not a shill you say? Just like these console owner fanboys or ifags who want their beloved companies to have global domination are not shills right?
>>54189482
But he's not wrong in some regards. It WILL mean that Nvidia HAS to follow suit or get fucked, really.
Not saying that Nvidia won't see the sense in it and do the same, because its silly to assume that they'd keep brute forcing large dies just to claim strongest single chip GPU...
However, we all know that Nvidia doesn't SCALE as well as AMD, so that's what they'd have to really, really work on.
https://m.youtube.com/watch?v=aSYBO1BrB1I
New video just released. AMD poised to btfd Nvidia. This is just sick.
>>54189526
>we all know that Nvidia doesn't SCALE as well as AMD
in what? power consumption?
>>54189536
>being this late
>>54188512
>>54189482
If it'll have same shitty DX12 performance - I give up.
>>54189568
Yes, and we all know how perfect Nvidia are for their power consumption, don't we?
Don't forget its only since Kepler that Nvidia was better than AMD, and it could easily flip back to AMD's favour if they do what Nvidia did and gut anything they deem unnecessary for the short term out of their cards.
No, in multi-GPU, AMD is the reigning king for actual performance. What lets them down heavily is their gross incompetence in their slow release of crossfire profiles (if they bother at all).
>>54189746
multigpu is a dead meme thanks to temporal reprojection
>>54189847
That's on Alternate Frame Rendering (AFR).
The alternative option is Split Frame Rendering; which ought to fix the whole issue.
SFR was never very successful in the past due to bandwidth issues. If AMD or Nvidia put two dies on a single interposer - or at the very least, standardise 'multi-gpu single PCB' cards, SFR should have much less issue with bandwidth constraints - ESPECIALLY on HBM2 and onward.
Hell, my guess is that's exactly what they're banking on - HBM2 to make SFR a very, very valid possibility.
>>54189076
drivers are old af son, 390x beats the 980 in most games now.
in 4 months it will probably outperform the 980ti.
>>54189884
this is more adoredtv memes.
>>54189904
The fuck they're doing anyway, goyworks? Tesselation up the ass?
>>54189904
I wonder if anyone will ever test 390X with drivers set to x16/x8 tesellation against 980Ti/980
Should make things very interesting.
>>54189904
only in recent amd sponsored games like hitman and ashes does the 390x perform better as well as quantum break because of dx12. even far cry primal which amd had a massive lead over nvidia in is now better performing on the 980.
this always happens. amd gets advertisement licences for shitty q1 and q2 games and end up performing better in those games. once q4 comes and all the big AAA releases come out then nvidia always does better because of sponsorship's and general better dx11 performance.
>>54189920
Do some independent research, pleb. He's making absolute sense. Yes, I admit that it was him who brought the topic to my mind, and so I took what I knew about the topic, what he was saying and a quick google search and read up on it and, believe me or not, what I found is that he's right on that topic.
The issue with using split frame rendering has ALWAYS been the fact that the links between GPU's were traditionally slow as fuck. That made for an awfully choppy experience where one half of the screen could be well ahead of the other.
Alternate Frame Rendering fixed that issue by at least making it so you wouldn't have one half going smoothly and the other half of the screen poorly.
If HBM2 goes well and has the expected bandwidth speeds - if the links are strong enough to actually utilise those speeds between dies, you could EASILY make them match up and work on the ruleset of slowest to render sets the pace and not neuter the whole experience.
To get the ball rolling though, he's also right that you would NEED consoles to be behind the smaller pairs of GPU's working together to force developers into really putting in the effort to make MGPU work properly though.
There just isn't enough incentive to do it unless consoles force them.
So when is this shit coming out? I want to build a PC in the very near future.
Am I going to have to wait until Christmas to not get ripped off?
>>54189585
Nobody linked the new video yet so go shove it.
I'm excited for Polaris, but I'm probably switching with Vega. AMDGPU being up to date with these new chips would be swell, I want to get rid of this 970.
>>54189076
>old driver benchmark meme
pajeet stop trying
>>54187659
I don't give a fuck of Polaris 11 or 10, I want the 490/490x...
>>54190004
thats the whole problem. this guy firmly believes the next gen of consoles will be multigpu. we already know the ps4k isn't multigpu and that console will last for at least the next 5 years. the next xbox is rumoured to have the exact same specs as the ps4k so that puts the next xbox out of the picture as well. if nintendo are the only console company doing multigpu then that means nothing because nintendo rarely do multiplats so it wont benefit the pc space in any way. if amd wants to do something big then they'll need to do this multigpu stuff right now, which we all know wont happen. the earliest they can try and do it will be in at least another 5 years when the ps5 and xbox 2 replace the ps4k and xbox 1.5 and by then it'll be way too far gone for amd.
>>54190018
June-July.
>>54190154
Yeah, but you're forgetting that AMD isn't going balls out MGPU straight away either.
Navi is scheduled for 2018 onward, and is their start for MGPU.
Even if consoles don't go MGPU in 2018, there's nothing wrong with AMD preparing their next big move ahead of the game.
Perhaps their plan isn't to force consoles onto MGPU even, but to make MGPU much, much more common on the mid-tier PC market by matching up two weaker dies for their midrange market cards?
It would be a gamble for AMD, hoping that Nvidia go for the same kind of logic - but its still possible.
If they relied on the concept of treating two chips on a single interposer and treating it as a single GPU, it could definitely work and be their way of moving past the fact you can only get so small before you lose profit.
Whats stopping me from just going out and getting a 980ti? Regardless, it's still going to outperform consoles for at least 5 more years. Isn't that one of the reasons we like pc gaming, /g/?
Is it really that bad if I don't wait?
>>54190216
as adored said, it will only take off if consoles start using it because it'll force devs to take advantage of those dx12 multigpu features which will obviously benefit the pc space. since the consoles are confirmed not to have multigpu then the devs have literally no incentive to optimize for it even if amd beg them. it'll be more work on their part and we all know how lazy devs can be. i honestly can't see amd being successful with this.
>>54190258
Nothing is stopping you. Just remember that you won't be able to do UHD at 60fps or 1440p at 120fps comfortably, VR is going to need much more power very fast, and Nvidia are assholes with their closed source bullshit.
>>54190258
Depends on what you're currently on. I mean, if you currently have a craptop, go ahead and upgrade. But if you already have a decent or even good GPU, you'll likely get a significantly better deal once next gen GPUs drop AND you won't have to worry about Nvidia starting to gimp maxwell performance as soon as pascal comes out.
>>54190212
That early? I thought it would be much longer.
>>54190330
nice meme
>>54190345
>meme
Remember when 780ti performed better than 290x instead of on par with 280x?
>>54190311
>>54190330
Fair points, I'm on a craptop now that I got as a gift. It overheats so fast and can only handle games at a decent fps from 2011 - 12.
How soon could we expect non-reference cards to come out for pascal cards? wasn't it like several months for the 980ti's?
>>54190304
Dude, consoles don't even use DX12. They use their OS as the actual API.
THAT is half of what makes a console so efficient with the power they have. Their OS is literally designed with gaming as the #1 priority and so before anything else, its bare metal as can be.
The other half is of course the fact that developers know the exact limitations and the exact specifications.
DX12 and Vulkan are trying to make PC's behave as close to consoles as they can.
>>54190417
Yes it was.
>>54190460
i didnt mean it in the sense that the consoles will use dx12. i meant it as if the consoles had multigpu they'd have to port it to pc via dx12 and utilize those multigpu features because it would be the logical thing to do. it would be a straight port between a multigpu console to a dx12 multigpu amd r9 4xx.
>>54190460
>DX12 and Vulkan are trying to make PC's behave as close to consoles as they can.
dx12 is the gaymer API, vulkan was just as much about gaymen as it was about providing a proper API for developing for tiler gpus (powervr and similar mobile shit)
>>54187127
>The ZOTAC GeForce GT 710 1 GB (ZT-71304-20L) video card is powered by a cut-down version of NVIDIA’s GK208 GPU with 192 CUDA cores, 16 texture units and 8 ROPs. The GPU is based on the Kepler architecture
>GK - Kepler architecture
>>54190460
>The other half is of course the fact that developers know the exact limitations and the exact specifications.
People should look up the sort of crazy shit Insomniac games pulled for the later ratchet and clank games. It got to the point that they were building their game engine around the exact read/write speed of the ps2's dvd drive to squeeze maximum performance from the machine (this is also why R&C 3 is one of those games that barely works on dying ps2 machines).
This is also what Carmack was on about when he made the claim of consoles being 2x as powerful as equivalent pc hardware. Its the software holding shit back in the pc space (though naturally some of the bastraction can never be lifted due to the million combinations of hardware out there).
>>54190495
By very definition multi-gpu is not as good as simply having a single gpu that is twice as fast - there will always be additional software overhead to sync the two chips.
See also: sega saturn (okay that was cpu not gpu, but the point stands).
>>54190529
Kepler is still basically fermi underneath - its mostly gimped fermi with software picking up the slack (equally maxwell is mostly kepler with some numbers shifted about). Sure there are improvements made to the underlying architecture, but fermi is still making its presence felt.
On the AMD side GCN hasn't evolved too much from 1.0 found in tahiti - hawaii is simply more of everything and fiji is hawaii with some numbers shifted about (primarily HWS replacing the ACE units).
>>54190548
By very definition multi-gpu is not as good as simply having a single gpu that is twice as fast - there will always be additional software overhead to sync the two chips.
this is one point which contributes to my opinion of the consoles never having multigpu setups.
>>54190640
Off the top of my drunken head no console has ever had two graphics chips. Generally the cost alone of two bleeding edge semi-custom chips would price said console beyond what most people are willing to pay.
Then you have to factor in the learning time it takes development teams to wrap their heads around the nuances of two seperate chips needing to communicate - thats a disaster in the making.
Like the previously mentioned sega saturn - while it had two cpus virtually nobody knew how the fuck to make it work.
>>54190723
the argument people are making in favor of multigpu is that dx12 has a wrap layer which sees both gpu's as a single big gpu. very deluded people like adoredtv think this is actually the future and will give amd world domination within the gpu market. its all a load of bollocks if you ask me.
>>54190820
Well it certainly alleviates a lot of the issues that plague multigpu under DX11 but even ignoring that lets be real - how many people have two or more gpus in a single machine? The cost is pretty high on cards that see the biggest gains (primarily high end chips).
Especially when you consider some low end cards - like the 750ti - are blocked from multigpu on the driver level (I don't know if AMD blocks any of their cards like that).
That said I could easily see a developer (perhaps crytek) using this feature heavily if they are maknig a game designed to bring pc hardware to its knees, not unlike the first crysis. Still, that still makes it a niche feature - but if the option for scaling is there this can only be a good thing for end users.
>>54190820
Very deluded people see it as world domination for AMD, and I agree with you on the world domination part - but being part of the future? Like I said earlier; if AMD and Nvidia can get two dies on a single interposer or make it so HBM2 and onwards kills the bandwidth between chips issue at least (though it'd still mean single PCB for that), then treating the two chips as a single unit could very easily become the future for getting more performance without having to go to to chip sizes as big as Titan X/Fiji again.
>>54190953
That would tie in nicely with AMD's old "small die" strategy. Instead of monster chips they would simply bolt two of their smaller chips together to get performance. If the software support is there is not a bad strategy all told.
Certainly it curbs down on certain limitations a single chips has (like the ones fiji run into).
Who waiting for Volta/Vega here?
