http://semiaccurate.com/forums/showthread.php?t=8635&page=52
>256-bit GDDR5 in Polaris 10
>128-bit GDDR5 in Polaris 11
holy fuck currytech wasn't kidding when they said these new cards were mid range and low range
Fiji just came out, that is AMD's high end GPU for 2016. They aren't going to replace a brand new die until they get a substantial return on investment and sell off the majority of inventory.
Small Polaris is their new entry level GPU, and it will scale down to mobile as well. The larger die is a mid-high end part.
Waitfags BTFO
>THEY FELL FOR THE AMD POLASHIT MEME
why do you faggots eat up every rumor like it's your last meal?
they'll release official specs in a couple weeks and then you'll have enough time to shill.
>>53661407
GDDR5X, and yeah, HBM2 on Vega 10, we already know that. This is at best a Fiji replacement.
NVIDIA FIRST TO HBM2 & GDDR5X
ENJOY YOUR GDDR5 CARDS, AMDPOORFAGS
>>53661469
based on that very brief Hitman demo the Polaris 10 performs up to as well as a Fury X
http://videocardz.com/58639/amd-polaris-10-gpu-specifications-leaked
http://videocardz.com/58634/amd-confirms-polaris-10-is-ellesmere-and-polaris-11-is-baffin
IT'S OVER, AMD IS FINISHED & BANKRUPT
AMDPOORFAGS CONFIRMED ON SUICIDE WATCH
>>53661612
AMD FIRST TO HBM1
ENJOY YOUR GDDR5, NVIDIOTS
>>53661612
We already have HBM cards ;)
Enjoy your curent 256 bit and 128 bit cards!
At least my GDDR5 card has 512 bit.
>>53661722
>512bit
>Get TROUNCED by 256bit card
STAY MAD, AMDPOORFAGS
>>53661739
but HBM is AMD technology stolen by jewvidia :^)
>>53661801
HBM2 is JEDEC spec, AMD can't even into HBM2 until 2017
>>53661739
Since when?
A $300 card now is beating out a Nvidia $500 card on modern titles using dx12.
They are highly efficient entry level cards that can run games like Star Wars Battlefront on 1080p and not high end cards. We know this since january.
>>53661820
There is no HBM2 spec. It is the exact same thing as HBM1, the only difference is a manufacturer distinction.
AMD is the main party behind its creation, just like with GDDR5.
is there any way to tell if its GDDR5X or not??
>>53662447
99% chance it won't be. Any GDDR5X available now would be risk production in limited quantities.
>>53661407
The ultimate poorfag CPU/GPU combo has been released. All hail AMD. :^)
>>53661407
So it's confirmed, P10 die size doesn't go over 250mm^2
>>53662525
yes we've known it's incredibly small for quite some time now
>>53661407
>>53661687
Those specs are 470 and 380 tier...
I'm calling Vega will be 490 and "Rage"
>>53662537
250mm2 at 14nm isn't incredibly small at fucking all.
That would around 550mm2 on 28nm
>>53661857
>AMD is the main party behind its creation, just like with GDDR5.
And gdd3 and gddr4...
So polaris 10 = r9 380 tier?
>>53662553
>>53662537
>>53662525
What do you think is the TDP of polaris 10?
>>53662584
Probably, or it might be the chip that will replace the aging Pitcarin that's in the 370.
>>53662582
>With that in mind, the GPU actually did well in the benchmark and came ahead of the GTX Titan Black. Which features a full GK110 chip and coincidentally is faster than a 2816 GCN core Hawaii XT GPU. Also funnily enough this is very close to the performance we’ve seen from the leaked GTX 1080 and GTX 1070 benchmarks.
This is also a ES at 800MHz
>>53662584
No, Polaris 10 = 390X
Polaris 11 = 285/380
>>53662553
The are scaling advantage is over 70%.
A 750mm2 28nm die would be around 225mm2 in 14nm silicon. A full Fiji die would be around 180mm2.
>>53662668
Is that storage or logic scaling?
>Samsung's 14nm process turns out to be inferior to TSMC's 16nm process
>no HBM2 until 2017
>tiny memory bus widths all around
>most likely no GDDR5X
That's it, we're done here. It's over.
AMD is finished.
>>53662675
Thats an average, not specific to SRAM cells or logic.
>>53662702
And then you stopped dreaming.
>>53661824
DELETE THIS!
I-ITS NOT OUR F-FAULT AMD HAS ASYNC COMPUTE AND WE DONT
Are you telling me that the latest iteration of /g/'s "i-it g-gets b-b-better n-next y-year!" favorite amd marketing meme is falling on its face just like it did every single time before for nearly a decade?!
STOP
THE
PRESSES
This looks like AMD's version of Maxwell.
Smaller dies all around, smaller memory controllers, less but more powerful shaders, high clocks.
>>53662724
nice shilling
>>53661407
>no reason to upgrade my 290x yet
JUST
>>53662702
but how can TSMC's process be superior when they couldn't even get a pascal GPU to produce on stage
enjoy Pascal being released in 2017
>>53662838
FUD, Drive PX was not meant to display working pascal chips, it was just meant to show the reference PX board.
>>53662871
like I said, enjoy Pascal being released in 2017
>>53662899
AMDHOUSEFIRE VICTIM DETECTED
>>53662920
I'm actually still using a GTX 670 :^)
Polaris 10 = R7 450
Polaris 11 = R9 480
>>53662937
>poor
>defends AMD
lelelel
>>53662937
GTX 680 here, I know the upgrade is coming soon but I have feeling it's going to be Nvidia again because fuck waiting around for this shit.
>>53662960
Go be retarded somewhere else.
>>53662962
>implying I would upgrade to a 3.5gb memecard
>>53662974
http://wccftech.com/nvidia-pascal-3dmark-11-entries-spotted/
>The first entry we have is of an unidentified Nvidia GeForce graphics card with 7,680 MB of graphics memory, 512MB short of 8GB, very likely a test chip.
Pascal literally confirmed 7.5GB memecard
>>53661469
This is the only fucking guy i ever seen on /g/ that knows what the fuck is going on. AMD also stated GPU manufactures make a large amount of money on the low to mid range cards. AMD cares about low power consumption for Polaris because of mobile and laptops. I think they are finally going in the right direction with this. Build a house but make sure the foundation is good.
>mfw running an i7 950 with 660GTX
>>53663083
Nvidia #1
I will never buy nvidia again since their drivers ruined my previous GPU
they can go fuck themselves
>>53663126
Thanks for your contribution, Rajesh
>>53663105
Pic related is why I'll never buy another Nvidia card.
>>53663126
BAAAAW BAAAAAAAAAAAAAAAAW
Go on, buy the GPUs without working drivers at all
You'll be back to Nvidia as fast as AMD's bankrupcy
>>53661824
No notable game developers will use DX12 until Nvidia gives the go-ahead. And Nvidia will give the go-ahead when they are ready to rekt some AMD butts.
>>53663105
At least it didn't kill GPUs like Crimson did
>>53663083
>Not having adblock
top lel
>>53663158
That makes every thing better.
>>53663185
Well yeah, it does, unless you want to advocate that having a dead GPU is somehow better than facing errors on a driver you can rollback from. If that's the case, be my guest.
>>53663188
Get the fuck out of here their are people all over with dead nvid cards cause of those 2016 drivers just good it and stop being a fag.
>>53663296
>so fucking mad he can't even type properly
>>53663314
resorts to my typing. awww so cute
>>53663296
>driver update by itself killing cards
yeah no
Unless the driver locked the fans at a low RPM despite the card running hot (which is what happened with crimson), this is literally impossible. This is on par with downloading a virus and its code "destroys" your PC or downloading more ram. There's no such thing. What happened most likely is that they had multiple monitors connected, which makes the gpu send no signal, precisely the bug within this driver update. One of them even mentions having 2 monitors.
>>53662749
>you will never build a time machine and go back just in time to stop the AMD/ATI merger
670 here.
Just want to upgrade for VR. Don't know what to do. It's been so long. ;-;.
https://www.forum-3dcenter.org/vbulletin/showpost.php?p=10980597&postcount=1966
>Same graphics level as Fiji/Tonga
IT'S OVER, AMD IS FINISHED & BANKRUPT
>>53663171
>using adblock that got sold
>not using the superior ublock origin
topzozzle
>>53663412
>what is firmware
>what is microcode
>what is BIOS
Drivers have bricked hardware on multiple occasions, but just GPUs.
>>53663446
I feel you
>months ago we thought that summer 2016 was going to revolutionize the already stunning PC GPU market
>its just a power consumption update that doesn't even definitively outperform old GPUs
literally what the fuck
>>53663471
I think the point is to sell them for like $200. A 280x still is on par with a 960 and the 380/x beats it in nearly every single game
>>53663476
>not using both
step up your game
>>53663471
b-but muh lower price/power consumption
>>53663471
What the fuck is that post?
>>53663503
holy shit nigger wtf are you doing
Not even joking here, this will most likely be amd's final nail in the coffin. If pascal gets no competition whatsoever when amd is in need of literal miracles as of now, amd is downright dead.
How is pascal shaping up?
>>53663412
you so incredibly are stupid
>>53663526
why would I use only one when two is always better?
>>53663314
>>53663412
Nvidia™ damage control out in full force
>>53663538
t-there's always c-consoles, amirite?
>>53663544
Adblock got sold, now they show "acceptable ads"
>AMD is literally going to die in a year's time
AHAHAHAHAHAHAHA
WE FUCKING WON
SUCK MY DICK AMD FAGS
>>53663591
no problem, ublock will block those ;)
>>53663598
>Nvidia™ damage control out in full force
>>53663598
>AMD kicks the bucket
>Firesale on remaining stock
>Get mid-range gaming PC for dirt cheap
>>53663612
>damage control
>when nvidia downright won and amd is going bankrupt by fucking up yet again
lmao
Q Q
Q
M O A R
O
A
R
AMD FAG
>>53663538
Pascal is Maxwell with FP64 bolted on.
It's a joke.
>>53663598
>Underage Nvidiot's want amd to go bankrupt
>Will be totally OK with with spending 800 per card for marginal performance gains because of a Nvidia monopoly.
Can you honestly be this retarded?
>>53661469
I have four Fiji cards but no Vulkan
Where is it
>>53663627
>no more driver support
>ever
>mid-range gaming PC is obsolete as soon as you buy it
If AMD goes away I'm gonna have to write a GPU driver and that's gonna suck
>>53663646
>amd in charge of providing competition
FUCKING KEK
POORFAG IN SUICIDE WATCH
>>53663670
So nothing changes then?
>>53661828
>perf/w optimizations
Literally, what you do when you can't push the performance envelope any further.
Build more god damned power plants, fuck efficiency.
SHIT AMD SHIT AMD SHIT
FUCK OFF AND DIE DIE
aahaha amd stock just lost 4$
>>53663690
I honestly should just start on an open source Vulkan driver for the Fury R9
Maybe once I graduate I'll spend the summer working on it
>>53663704
But it's okay when Nvidia does it.
>>53663727
I don't have a horse in this race, it's shitty whoever is going it.
>>53663704
>I have no idea about anything: the post
Perf/watt is the single most important thing for a GPU. It is an array of several thousand small ALUs. The less power they use the more of them you can pack together in a larger GPU.
>>53663412
My 1.5 year old r9 270 died 2 hours after a driver update. I didn't play anything graphically intensive, it could not overheat in those 2 hours, I don't think that was a coincidence.
>>53663840
AMD
DRIVER
QUALITY
GENTLEMEN
>>53663471
if it has the same performance as Fiji and is a $250 mid range card I would cum
>bought a 290x only last summer
>tfw want a fiji GPU
the fury or fury nano is calling me
i really want to do a mini ITX build with a nano
>>53664017
Fury is nice
>>53664017
>tfw like mini-ITX systems
>tfw have external bay devices I still use
>>53663840
Nice anecdotal.
>>53663668
and this is different to what AMD fags face everyday how?
>>53663840
You do know that GPUs also die out overtime right? They don't just last forever until they overheat and go poof. Defects are a thing. Your case was just a coincidence.
>>53664017
>>53664054
Why the preference for mini itx over mino atx /g/?
>>53664017
Get a laptop with a full voltage i7 and thunderbolt 3. AMD already has full support for external GPUs over thunderbolt 3
https://www.youtube.com/watch?v=uraNu-H9S6I
![1442219127612[1].jpg](https://i.imgur.com/MmjQyft.jpg "1442219127612[1].jpg")
>>53664113
kek
CUE UP YOUR DRAKES LADS
#TEAMGREEN WINS AGAIN
>amd
>AMD is over
the market disagrees
>>53661407
AMDOOMED
>>53664180
AMDEAD
>>53661614
There are many complaints from general owners of that game that it's poorly optimized. It's ratings are seriously bad, so I'd take any benchmarks from that game with a grain of salt.
>>53664134
>mino atx
I assume you mean micro?
Because mini ITX takes up less space. I did micro for my first build and I regret it, I don't need a fuckheug case.
>>53664191
AMDONE
>>53664200
miniITX is retarded small. MicroATX is a lot easier to work with while still being significantly smaller than ATX
Looks like Polaris 10 is the one for me to get. No point in getting a 380x now.
>>53664218
>gettin mad about how people prefer their computer sizes
this pleb doesn't even a e s t h e t i c s
>>53664216
AMDESTROYED
>>53664163
>it raised a few cents
>in a company whose shares used to go as high as 40+ dollars each
WEW YOU HEARD IT HERE GUYZE, THEY ARE TOTALLY SAVED!!!
>>53664243
9.5/10 build.
how are temps?
>>53664200
Ah I thought there was something else, I still use Micro because I am possibly the only person on Earth who still uses a sound card.
I can't find a mobo on the cheap range that got enough ports for my logitech speaker set.
AMDECIMATED
>>53664243
That's a regular FT03, the mini is mini-ITX. Nice build though.
>>53664257
Its surviving until zen, bristol ridge, and polaris.
Fuckin retard wanting a monopoly
AMDISMANTLED
m u h
d i k
>>53664314
Heh, that's a good one since Mantle = dead too.
>>53661469
I mean it's already been explained countless times now that Vega is going to be the Fiji replacement, how do people no understand this?
>>53664292
>Fuckin retard wanting a monopoly
No yeah because amd is totally providing some hot competition yo. I too want a company to be kept in place for the sake of pity so we can have pretend-competition and keep dwelling in this massive stagnation. After all, that's how the market drives amirite? It totally wouldn't benefit everyone if amd went bankrupt so they could be bought off by companies that actually have money for proper R&D and know what they are doing, nah, nuh-uh, PROTECT MUH SPECIUUL COMPANY.
Idiot.
>>53664327
>so they could be bought off by companies that actually have money for proper R&D and know what they are doing
Funny that AMD are the ones making innovations on both the CPU and GPU side while Intel and NVIDIA just roll in their R&D money and give zero fucks.
Maxwell and Skylake are pretty clear examples of this.
>>53664327
Not trusting the shit wrecker keller
>>53664257
You clearly don't understand how these things work.
>>53664353
>making innovations
Yeah I can totally see those innovations with re-re-re-re-re-re-re-re-re-re-re-rebrands of the same GPU and half-cores with less IPC than cpus from 6 years ago being sold as the full thing.
>Intel and NVIDIA just roll in their R&D money and give zero fucks
Maybe if they actually had any competition from something other than a pretend-company, they'd actually bother giving a fuck. But thanks for making my point for me.
>>53664163
14 cents surely stops the 5 year slide
>>53664373
Oh right and you do huh? Please teach us how the ideals of the free market involve sustaining a company out of pity and how bankruptcy and buyouts aren't part of its system. I'm looking forward to you stumbling upon your own words.
>>53664327
>there are people actually this fucking stupid
>>53664398
>muh amd fanboyism
>>53664376
>HBM
>Extremely efficient GPU (Polaris)
>Mantle, which reulted in Vulkan and DX12
>Already working on HBM3/Next Gen
Meanwhile at NVIDIA
>Shoddy drivers destroying shit as usual
>Let's just get rid of DP from the Titan like and sell it to retarded kids with parents' credit cards, who make up the majority of our consumer market
>3.5GB
>7.5GB
Meanwhile at Intel
>5% IPC EVERY YEAR GUYS
>GUYS WE NEED MORE WAYS TO LOCK SHIT DOWN
>SOMEBODY RELEASED A BIOS THAT ALLOWS OVERCLOCKING ON NON-Z BOARDS? GET THEM FIRED
>>53664448
>meanwhile intel and nvidia take those innovations and wreck amd with them and it all starts over again
>>53661407
Indians won once again
![Ptqu0PF[1].jpg](https://i.imgur.com/HhDuvfA.jpg "Ptqu0PF[1].jpg")
>>53664502
>>53664493
>thinking otherwise
Even the documentation is near identical dumbass.
>take those innovations and wreck amd with them
What's it like to be completely retarded?
>>53663098
>2600k and 660ti
Does the job and cost me next to nothing
Feels good man
>>53664327
I don't want to go back at the time when Creative was all there is to "gamingsound" and the alternative was static-laden onboard
It was a most shitty time. I'm glad onboard has reached the level of "good enough"
>>53664448
>HBM
Does literally fucking nothing. We aren't even starved bandwidth wise and what did hbm change for amd? That's right, nothing. And even in that field nvidia is beating them since they're gonna have hbm2 before.
>Extremely efficient GPU (Polaris)
So you're telling me that consuming normal amounts of energy like competition have been doing since ever is innovation now? Wew lad.
>Mantle
Went absolutely nowhere and you're delusional if you think that vulkan and directx 12 wouldn't exist without it. It got absorbed because its open source and that's it.
>already working on hbm3/next gen
While they can't even get their hands properly on hbm 2. So much for that.
>Shoddy drivers destroying shit as usual
You mean Crimson? Meanwhile AMDfags have to wait weeks~months for a driver when a game comes out to play them properly. Such quality, such innovation.
>3.5GB
Wouldn't happen if we had competition, but we don't because retards like you keep amd afloat so nvidia has an excuse to do literally whatever the fuck it wants.
>7.5GB
Are you fucking retarded? What does this have to do with anything? They never said anything about it being an 8gb card you memeing retard.
>>5% IPC EVERY YEAR GUYS
And still over TWICE the IPC of AMD, exploring the near limits of silicon while amd does MOAR HALF-COARS when barely nothing benefits from that.
>GUYS WE NEED MORE WAYS TO LOCK SHIT DOWN
>SOMEBODY RELEASED A BIOS THAT ALLOWS OVERCLOCKING ON NON-Z BOARDS? GET THEM FIRED
Again, that's literally what happens when you only have pretend-competition from AMD. Now if they were bought out by someone who actually knows anything other than how to run TWO companies into the ground (amd and ati), maybe we could properly respond to intel's fuckery.
But no, let's all buy amd out of pity everyone, that will surely fix things up. It's been working out so well throughout these years of stagnation huh?
Idiot.
>TFW bought a Fury X
>Polaris will offer me nothing, even Vega probably won't be all that exciting to me
>Generation after that will bring refined designs, even higher efficiency and more power
>By then, my Fury X will be starting to show a little age
Feels good, I made the right choice I feel.
>>53664591
>bought amd
>made the right choice
1
>>53664612
>Nvidia
>Right choice
I'm not 13.
>>53664531
>implying using a feature of a bit of open source code is integral
> implying that something can't exist without it
You're the idiot chief
>>53664327
I, for one, welcome an nVidia monopoly.
If it weren't for AMD we might've gotten sweet video cards with x1080 tessellation.
>>53661407
Just give me cool, quiet 390 level performance in the 480 with 8GB RAM and some OC potential. I'll be happy to buy it used in 2018
>>53664961
Its funny because tessellation was an AMD innovation
>>53665053
>AMD can't even make their own innovations run properly
>>53663009
Actually it's 5gb you fucking bandwagon memer.
I WAS RIGHT, HBM1 WAS USELESS, YOU ALL FELL IN THE AMD CRINGE MARKETING, 4GB OF HBM1 WAS A FUCKING NONSENSE
>>53661857
Are you stupid?
http://www.anandtech.com/show/9969/jedec-publishes-hbm2-specification
http://www.jedec.org/standards-documents/results/jesd235a
Read the last page of the PDF to see the differences between hbm1 and 2.
>>53662582
see above so you dont go thinking what that other anon was.
>>53662466
IIRC GDDR5X chips are exactly the same. The spec is for a different memory Controller that enables the faster speed.
Further reading on GDDR5X: http://www.jedec.org/standards-documents/results/GDDR5X
>>53663643
Where have you been reading this? Is Nvidia finally admitting that fucking over their FP64 units was a mistake?
>>53664257
econ illiterate spotted
>>53664575
>arent starved for bandwidth
You have no idea what you're talking about man...
>already working on HBM3
Of course they are. But "they" in this case isnt AMD or Nvidia. It isnt even Samsung or TSMC, its JEDEC working on the spec that they'll ship out in a few years.
>rest of post
Damn anon. What ever they pay you to shill this hard, its too much. Even >pooinloos are more convincing.
>>53665315
>Unity
Anon, you do know that game was the center of a major controversy surrounding its poor performance on all devices dont you?
>>53665158
please, nobody will believe that the 3.5gb memecard has any long term value
>>53661857
>>53665390
In case you cant be arsed to download it and read.
>>53665411
>implying an AMDead card will live to see the long term
>>53665390
>Anon, you do know that game was the center of a major controversy surrounding its poor performance on all devices dont you?
and GTA V, Far Cry 4. @max texture, the games aren't smooth because of 4GB limit
>>53665471
Okay, that's possible. I personally dont know, just thought it made you look stupid using Unity as an example.
>>53665465
>implying it won't
>implying Nvidia isn't just jewing you into using a card that will be useless when DX12/Vulkan becomes standard
>>53665390
>I'm a shit eating tech illiterate retard
AMD designed HBM.
AMD designed GDDR5.
ATI designed GDDR3.
JEDEC is only the body who ratified them as open standards, they did not create anything. Tech illiterate faggot children such as yourself should do more research before posting on topics they know nothing about.
>>53665315
>3.5GB
>1080p
>>53665500
>implying anything gets done without a spec
I want Asia to leave.
Getting away from the Graphics circlejerk, When do you kids think we'll start seeing HBM in things like APUs or in low power applications like Mobile Phones?
>>53665528
>970 = Fury X
>>53664198
And if a Polaris card can perform well then that is a good sign
>>53665528
>meme gtx 970 vs fury
Both are shit retarded
>>53665583
Phones, never. APUs, maybe 2018. It would be a miracle if Raven Ridge next year had HBM on package, but AMD not mentioning a single thing about it doesn't leave much hope.
>>53665623
What makes you say never for Phones? This seems like the kind of thing Apple would jump on as soon as they could. Similar to how they adopted NVMe storage on their newer models
>>53665587
>>53665618
>Can't read
>Samefagging
That's a 390 next to the 970
>>53665644
>Can't read
it's Fury X vs 980ti, not the 390 vs 970....>>53665315
ITT: Nvidiots buttmad a $200 GPU is going to be faster than the $500 GPU they just bought
>>53665636
Throwing bandwidth at things doesn't improve performance. Bandwidth is only useful if something is actually using it. Mobile phone GPUs are absolutely nowhere near needing 128GB/s. A HBM module also draws about 3w, not great for battery life.
>>53664243
I'm not even going to pretend that doesn't look stupid.
>>53665644
>dat off topic post
My point is a high end card with 4GB of HBM1 was a mistake like the 970 3.5 was a mistake. AMD should use 8GB of GDDR5 and not the useless HBM1. Read the thread next time...
>>53665722
The PHY for GDDR5 is larger, and uses considerably more power. They would need 8000mhz GDDR5 to provide the same bandwidth, which again would take a ridiculous amount of power.
>>53665465
The 7950 already did it.
>>53665722
>970 3.5
>mistake
nah
>tfw waiting for new high end card
>tfw I'll have to wait another year
Feels batman.
>>53665759
>considerably more power
Not really, only ~+15w
>>53665787
>tfw bought a card just last year
>will be time to upgrade again once decent HBM2 cards come out as well as possibly new zen CPU
Just in time.
>>53665675
A fully stacked HBM1 module draws that much, but all in all HBM is more efficient than even DDR4. It would be an attractive option for highpower mobile applications.
>>53665768
>tfw still running two based AS FUCK 7950s
feels fucking good anon.
>>53665823
That is twice the power draw, and its not counting the enormous leakage from the memory pads which contribute to total power draw for the GPU.
Don't try to correct people who know more than you.
>>53665842
Its more efficient in terms of bandwidth per watt, but mobile GPUs in smartphones are incapable of utilizing that bandwidth. Its pointless.
>Nvidia: muh cars
>AMD: muh performance/$
one of these strategies benefits gamers. The other is a meme.
>>53665873
>15w
>>53665873
>That is twice the power draw
yes and no
12GB GDDR5 at 7Gbps = ~31.5w
4GB HBM1 = ~14.6w
hypothetical ''8''GB HBM1 = ~30w?
>>53661824
Yup, I get a constant 80 fps in the paris level in Hitman with a 390X on DX12 with the newest hot fix driver. Shit is dank as fuck man.
DirectX12 actually has some pretty sweet advantages, more games are confirmed using it coming out soon too :^)
>>53665997
HBM2 will have significantly higher power draw over HBM1, it's a big concern for AMD which is why they are only planning on HBM2 for Vega then moving onto a new standard in 2018
>>53665997
No.
Higher density HBM packages are using smaller process DRAM dies with a finer pitch TSV connecting them. They draw less power. HBM2 draws less power than HBM1.
>>53663158
but that was another lie made up by Nvidia because one jackass with a custom BIOS screwed themselves.
the driver itself was fine
nice try but you know, ouch.
>>53661407
>256-bit
>128-bit
Holy shit, AMDrones BTFO
They are going BACKWARDS for the Rx 400 series.
>>53666051
Wrong
>>53666030
>$500 GPU
>1080p
Wew
WHAHAHAHAHAHAHAHAHAHAHAHAHAHA
http://videocardz.com/58639/amd-polaris-10-gpu-specifications-leaked
http://wccftech.com/amd-polaris-10-gpu-specs-leaked/
>>53666154
come back with some real sources anon.
>>53666089
NOPE.
This is from a bullshit presentation and was 100% speculative.
HBM2 is lower power. The higher density DRAM slices have to be made on a smaller process. They use less power. Straight from Samsung, Hynix says the same thing.
>>53663704
perf/w is going to be hugely important to me, however im a rare case scenario as im planning on living on an entirely solar powered narrow boat withing 12 months. looking at things depressingly realistically im going to have to add a diesel genny for winter time, UK climate sucks.
>>53662960
Nigger
Polaris 10 = R9 470 -> $100-200
Polaris 11 = R9 480 -> $200-300
>>53666195
I don't even want to ask
>>53666208
Polaris 11 is the low level/mobile GPU
>>53666208
>Doesn't know Polaris 10 is the bigger chip
>Thinks he can predict product names and prices
>>53663627
Intel to give AMD 1.2 BIL for Graphics Patents because they don't like Nvidia.
Why such hard feelings? Can't you just let both companies compete and calmly pick what's best for you?
FYI, AMD cards are superior for fast hashing (cryptocoin mining) by order of 2 to 3.
Friend has nVidia GTX 980Ti. It can hardly do 7.5 Mhash/s (SHA-51). That's CUDA program, the OpenCL program is even slower. AMD R9 380X can do 21Mhash/s.
Pick what works best for your case.
>>53666257
Maybe AMD will be able to make better use that of that money than NVIDIA.
Nvidiots are so eternally BTFO they don't know what to do when AMD gets some good news so threads like this pop up.
>>53666250
>>53666243
Just flip them around then
>Polaris 11 = R9 470/X
>Polaris 10 = R9 480/X
One will be between $100-200 and the other will be around $200-300 based on past price segments and the shrinking costs of GPU dies
>>53666270
Considering their won't be a stumble like the jump to 20nm, AMD will be fine, and Samsung is on the second generation of their 14nm process
>>53666311
What?
>>53663715
>lost 4$
>2.(xx)$ currently
>-2.(xx)$ after price drop
>major problem.
>Do you sell or buy?
>>53666365
Nvidia is eternally BTFO. Polaris looks like a real winner. Pascal is going to be a continuation of Nvidia's failure to plan for DX12. Like I said, green team utterly blown the fuck out. 2016 is AMD's year.
>>53666416
>Polaris looks like a real winner.
It looks like underpowered shit.
>>53666416
and Pascal?
>>53666109
He wants to keep it above 30fps.
>390 has 384gb/s stock bandwidth
amd is probably going to rebrand gcn 1.1 and 1.2 cards, they'll prob drop gcn 1.0 though
>>53661612
nvidia already said they are not going to be using gddr5x.
they are using newer samsung gddr5 thats able to clock to 8000mhz effective instead of the normal 7000mhz on the 1080 and below.
nvidia might be using hbm1 on their ti and titan not hbm 2 due to cost, yields, and heat.
hbm 2 might be limited to only quadro / tesla cards.
>>53666513
They will only rebrand "GCN 1.2" GPUs, so that's the 380/X and the Fiji cards. All GPUs using GCN 1.0 and 1.1 will be retired.
the 680 was roughly 30-40% (40% on only a few) faster[/URL] than the 580.) (http://www.anandtech.com/bench/product/1348?vs=1350)
we are comparing successors here. I.E. the 680 was the 580 successor
remember the 1080 is the successor to the 980. NOT the 980 ti. that's the 1080 ti's job
1080 being roughly 30 - 40% faster than the 980 (which would put it equal to or 10% greater than the 980 ti.
the "x70" has traditionally been 10 - 20% lower than the "x80" counterpart. this goes all the way back to the 6xxx series. I.E. the 6800gt was about 15% slower than the 6800ultra. just like the 970 is roughly 20% slower than the 980 (but the 980 has 30% increased price for that extra 20%! which has made the 980 a ****ty value)
if we take the 1080 being a full 40% faster than the 980
take the max 20% reduction for the 1070
that would put the 1070 around 20% faster than the 980 and 10% slower than the 980 ti
that's practically the 670 over the 580.
i think a lot of people forget about the last die shrink is that there wasn't a "ti" card in the mix. the 580 was the fastest single gpu solution on the market for nvidia. it was nvidia's single solution flagship, top dog. yes they had the 590 but it was NOT a single gpu card. it was a dual gpu card on a single pcb.
the 600 series? same exact thing. 680 was the top dog for a single gpu solution. not only that but both the 570, 580, 670, and 680 were the "big chips." they were the full fermi and keplar gpu's of their series.
starting with the 700 series this changed. the 780 was not nvidia's top dog. a new tier was added. the 780 ti followed by the titan was. maxwell 2 continued this trend but took it one step further with the 970 and 980 (gm204) not being full maxwell 2 chips. gm200 is the full maxwell 2 "big chip" and were exclusive for the 980 ti and titan x. the 970 being a cut down gm204 chip and the 980 ti being a cut down gm200 chip.
we went from "570 and 580" to "970, 980, 980 ti, and titan."
the "1080" is aimed at taking over the 980 spot. NOT the 980 ti spot. that's the "1080 ti" job.
you cannot compare the 1080 to the 980 ti since both cards are on different tier's.
another thing to note is what nvidia said: "pascal built on 16nm finfet process will offer 2x the performance per watt compared to maxwell 2." the 980 ti is 250 watts. half of 250 is 125. so that means, with no gimping, a full big pascal chip, at 125 watts, should offer the same performance as a 980 ti. the full pascal chip at 250 watts? that should offer the performance of two 980 ti's in sli. that's incredible.
>>53666605
seeing as the "1070" and "1080" are not the full pascal chips, that's being reserved apparently for the ti and titan, similar to maxwell 2's setup, the 1080 having performance near a 980 ti is perfectly reasonable. that puts it at roughly 30% increase minimum over the 980 since the 980 ti on average is 30% faster than a 980 with a similar tdp as a 980. also toss in the faster clocked gddr5 and increased memory capacity it will probably be even more faster. near beefy 980 ti overclock levels (extra ~7 - ~10%). rougly a 30 - 40% increase over the previous 980 that occupied its tier level. that's a big increase.
Time to wait for Vega...
Based on the number of transistors and alleged die sizes:
>R9 "Vega 10" <= 2 Crossfired R9 Fury/Nano (1.75* Fiji)
>R9 480X "Polaris 10" < R9 390X
There is a performance gap here. Vega 11?
>AMD will go bankrupt this year and Samsung will buy them and hire actually competent people and fire AMD's current retarded monkey engineers
What a time to be alive.
>>53666109
nah more like $360 but whatever crybaby
>>53666790
It truly is
>meanwhile pascal is DOA
http://www.bitsandchips.it/52-english-news/6785-rumor-pascal-in-trouble-with-asyncronous-compute-code
>>53666870
>http://www.bitsandchips.it/52-english-news/6785-rumor-pascal-in-trouble-with-asyncronous-compute-code
All the pieces of the puzzle are starting fall into place. Nvidia fucked up and to look like the good guy they released their game works stuff to the public and so devs will optimize for their shit cards.
>>53666584
>>53666605
>>53666619
so true
>>53666870
i always believed pascal was nothing but maxwell 2 optimized a bit for 16nm finfet.
the true upgrade wont be till the "1080 ti" and volta.
Since most people playing games either have integrated graphics or low end gpus(steam hardware stats) I can see this working in AMD's favor for first time graphic card buyers.The only thing I see with AMD is they can make the best GPU they want but if nvidia offers more at the store lvl, they lose out on a sale.Let me give you a example,Last summer I was in the market for a new card and picked a r9 290+dirt rally for $280 with the game it's actually a $220 card,but noticed nvidia was selling gtx 970s for $350+ game(witcher/batman at the time) making the actual card $290.At the time they were selling the 390s at the time for $345 but with no game and they weren't selling at all.
>>53666870
>>53667005
Nobody should be shocked by lack of async compute in Pascal
> Fermi - hot as fuck but actually does shader scheduling in hardware
> Kepler - removed hardware scheduler, relies on driver tweaking to not waste compute capacity
> Maxwell - stripped out more stuff. fp64 capacity in particular pretty low
> Pascal - you magically expect MORE GPGPU capacity?
async compute is in a way a "lite" version of hardware shader scheduling - it just slips instructions from different contexts into the ALU array is a graphics shader is waiting on memory or results from fixed function units.
Maxwell (and to a lesser extent Kepler) is power efficient because it relies on software engineers to manually balance everything out, which is great until Bethesda releases a patch with new shaders in it.
>>53666870
Fake news posted by shitty Italian site trying to slander Nvidia
>>53664253
AMDEFUNCT
>>53664279
AMSOMETHINGABOUTOVERHEATINGN
>>53661407
hahahaha AMD retards are gonna eat this shit
ENJOY YOUR GDDR5 AND LOW BANDWIDTH, AMDPOORFAGS
>another retarded anti-AMD shitposting thread
~70watt
~380-390 performance
whats the issue?
>>53667175
But it's a big news
>>53666047
Assuming AMD makes it to 2018.
>$2 per share
>>53667196
right now You shitposting...
>>53666268
Why even come to 4chan if you aren't going to argue over petty bullshit
>>53666870
Can driver optimizations actually provide the boost? Interesting if true that nVidia will be the brute force option instead of AMD. NVIDIA has a good driver team, so interesting to see how well the driver work will leverage an excess of raw power.
>>53667480
>driver optimization on low level
kek
Look at their "driver optimization" on the DX12 games. They've been "working" on it for the last 7+ months and nothing has come up.
tl;dr a lie
http://www.anandtech.com/show/10075/early-exynos-8890-impressions
>One observation I made today which was particularly concerning, was that both with the Snapdragon 820 LG G5 as well as the Exynos 8890 Galaxy S7 got considerably warm after running some heavy workloads. The fact that the Galaxy S7 touts having a heat-pipe thermal dissipation system is a quite worrying characteristic of the phone and should in no way be seen as a positive feature as it points to high power draw figures on the part of the SoC.
Polaris is nothing but a low end GPU, it's not even relevant and AMD built it on a bad HOUSEFIRE 14LPP process and GloFo licensed this shitty process from Samshit
Pascal is top to bottom performance and power efficiency leadership, Apple is moving entire A10 production to TSMC 16nm FF+, they wouldn't do that if Samshit's process wasn't shit
>>53667534
AMD BTFO
>>53667534
AMDESTROYED
SHIT PROCESS
SHIT GPU
SHIT BANDWITH
HOUSEFIRES
SHIT MARKETSHARE
DEAD DEAD DEAD DEAD DEAD !
>>53667534
http://wccftech.com/amd-polaris-10-gpu-pictured/
>polaris 11 passively cooled
>the city of you
>2W phone chips
>In any way relevant to desktop GPUs
>Nvidiots are this desperate
