>AMD Radeon R9 470 and R9 480 at Computex
>AMD Radeon R9 480 and Radeon R9 470 Series Detailed – Polaris 10 “Ellesmere” Has Around 100W TDP, Polaris 11 “Baffin” Under 50W
http://videocardz.com/59065/amd-radeon-r9-470-and-radeon-r9-480-to-be-released-at-computex
http://wccftech.com/amd-radeon-r9-480-470-polaris-10-polaris-11/
>Nvidia is planning to launch its next generation Geforce, pascal based lineup a few weeks before Computex 2016
>The full uncut GP104 die (GP104-400-A1) will constitute the GTX 980 Ti successor.
>A cut down variant of the GP104 die (GP104-200-A1) will constitute the GTX 980 successor.
>Another cut down variant of the GP104 die (GP104-150-A1) will constitute the GTX 970 successor.
http://wccftech.com/nvidia-launch-pascal-editors-event-may/
>currytech
Just fuck off already.
>>54035815
If all AMD have to offer this year is mid-range trash, they're in big trouble.
but which has simpler passthrough for VM use
>480 instead of 490
>Even more reason to wait for Vega
I should've bought a Fury X.
>>54035833
AMD, Nvidia already tried to prevent you from passing through their cards by producing an error when a card is virtualized
>>54035827
The news are from benchlife.info
Maybe they are simply going back to the old naming scheme with 490x being the highest end model with no "Fury" thing on top of the numbers.
110-135W with "2.5x perf/watt" would put Polaris 10 in the same performance range as Fiji.
>New GPU Architecture (not GCN “1.5”)
AMD already said Polaris is an evolution of GCN.
Bullshit clickbait rumor
>GP104-400 = 1080
>GP104-200 = 1070
>GP104-150 = 1060
pretty much confirmed
>>54036177
>Polaris 10 in the same performance range as Fiji
If it doesn't have its balls cut off by the stellar 256b memory bus with blistering fast GDDR5.
>>54036226
>waitforgddr5x.jpg
>AMD
Poo belongs in loo, not a PCIe slot
silly AMDcurries
>>54036279
I've been waiting for a long time, I just fucking want to get 4K 60FPS without paying NVIDIA for 2 980Ti's or AMD for 2 Fury X with 4GB VRAM and shit CF support.
>>54036331
okay...
>>54036175
literally who you could have written your mom instead, at least everyone would have know who
>The full uncut GP104 die (GP104-400-A1) will constitute the GTX 980 Ti successor.
More like a sidegrade that's a little cheaper
>>54035831
I want to buy midrange trash and even I don't want to see only midrange trash out of them.
>>54035815
>under 50w
AMD WON
JEWVIDEA FAGS ON SUICIDE WATCH!!!
>>54038054
one word, laptop tier gpu
>>54036331
>designated amd streets
it's a shame to see AMD making every possible mistake they could, this was supposed to be a generation where AMD could've realistically caught up to NVIDIA and become a true competitor again.
>>54038054
>slower that a GTX 950
So Polaris 11 confirmed both more performance and less TDP than my trusty HD 5670?
Seems like I'm sold. Just gotta wait until 2018, when it becomes affordable.
>>54038216
but amd are the better competitor with the better priced products, they just dont have the marketshare to reflect this, your wording is wrong
>>54038216
AMD was never behind in GPUs.
>>54038451
>>54038569
nice try raja, but AMD actually has more expensive and less performant GPUs than NVIDIA's current offerings.shilling won't save your job.
>>54038241
>slower
>>54038778
http://videocardz.com/58955/amd-polaris-11-sku-spotted-has-16-compute-units
>>54038689
t. paid nvidia employee
>>54038826
nvidia shills do it for free
>>54035831
>If all AMD have to offer this year is mid-range trash, they're in big trouble.
It's midrange trash that is supposed to max out current games in 1440p at full 60hz using 100W.
I dunno, it sounds like a winner to me, especially since Nvidia STILL doesn't support DX12.
>>54038689
>AMD actually has more expensive and less performant GPUs than NVIDIA
I want eurotrash like this to just fuck the fuck off this site already.
>>54038857
The R9 480, R9 470 will NOT
>>54038857
>I dunno, it sounds like a winner to me, especially since Nvidia STILL doesn't support DX12.
wat? nvidia is the only one with full dx12, 12.1 and vulkan 1.0.8 support.
>The full uncut GP104 die (GP104-400-A1) will constitute the GTX 980 Ti successor.
Why the fuck would nvidia replace the Ti line this early when they can do what they did with the 980 to 980 Ti transition and milk people to double dip, *especially* considering that AMD don't have Vega until 2017 and GP104 will likely take the performance crown for a single GPU by beating the 980 Ti by a hair, just like the 980 did to the 780 Ti when it was released?
No Fury X / 980 Ti successor until 2017 guys. At most we'll get a Titan X refresh for a bazillion $ to make up for the godawful yields of a large 16nm die.
>>54038913
Nvidia Maxwell don't support Asynchronous Shaders
it will be just like last time:
first year:
small cut down cores sold as high end to keep the % performance gains as if there was no die shrink
next year:
lets overclock the small cores, do some architecture improvements
next next year:
here, have this full chip for €999 without VAT.
we are all getting jewd by both nvidia and amd
>>54038942
>implying I'm not waiting for new cards to get released, to buy older ones for less shekels
but yeah, duopolies are fucked up.. I wish there was a third company to compete with them.
>>54038940
>muh async
only gives you any benefit if your hardware is underutilized to begin with
http://www.phoronix.com/scan.php?page=article&item=amd-nv-glvk&num=2
>>54038216
The only thing AMD is behind in nowadays is marketing and amount of fanboys shilling for your shit telling people to buy because it's the "safe choice" even if it's technically not faster.
>>54036177
>Maybe they are simply going back to the old naming scheme with 490x being the highest end model with no "Fury" thing on top of the numbers
The intent was this
>R9 X90 = GTX X80
>R9 X80 = GTX X70
>R9 X70 = GTX X60
R9 are the enthusiast cards that cost over $200.
>R7 X60 = GTX X50
>R7 X50 = GT X40
>R7 X40 = GT X30
Basically, Radeon numbered 10 higher than their Nvidia market equivalent, so they are naming their cards directly after Nvidia's naming scheme.
The Fury/X should have been the R9 390 and the 390s should have been the 380s
>>54039108
except AMD isn't faster, that's why people are buying NVIDIA GPUs. 'MUH MARKETING' is a pretty dumb excuse, as if it absolves AMD of all the dumb shit they did like rebranding and sweeping driver issues under the rug with no fix.
>>54039100
(You)
>>54039164
Except AMD is faster on just about every price tier. Just about every single game that's come out this year has performed better on AMD hardware and that trend is going to continue, Pascal won't change that.
>>54039100
Dude, if Volta do not support async, they are fucked.
BTW
https://forums.overclockers.co.uk/showthread.php?t=18724552
>Async Compute is super hard & too much works for devs
>>54039196
you mean 50% behind unless you have a $500 CPU to go with your $300 rebrand?
http://forums.guru3d.com/showthread.php?t=398858
AMD needs to make a compelling product before people will be interested in buying them. Lying won't change that, raja.
>>54039243
>Dude, if Volta do not support async, they are fucked.
kepler and maxwell already support it, so will pascal and volta.
>>54035815
>AMVIDIA
>not AMVIDYA
>>54039244
This is precisely why I laugh at all the AMDtards that think that DX12 will change the game. Simple fact is that it won't. All the early DX12 titles that have shown AMD pulling ahead is because AMD took the time to work with those developers to get their drivers working FOR ONE VERSION of the game and leave the updates entirely up to the developer. We will see those same titles show Nvidia cards pulling ahead in a few months time when all that AMD-assisted legwork goes neglected because of a lack of developer support.
DX12 is not a game changer, and nor will GCN 4.0
Nvidia will continue to be the better choice for all systems because of their superior driver support and controlled environment under the Gameworks umbrella. Nvidia has their shit together for the long term. AMD will run hard and burn fast.
>>54039264
They do support it on software level only. Pascal will probably be the same since they use the same architecture as maxwell.
>>54039485
>Pascal will probably be the same since they use the same architecture as maxwell.
where did this meme come from? pascal is not anywhere near the same as maxwell.
>They do support it on software level only.
you clearly have no idea what you're talking about and are just regurgitating memes, there's been support for running multiple cuda kernels asynchronously since fermi, newer architectures are no different in this regard.
>>54039587
Thanks of needful doing, more 420 rupees in account sir pajeet! Bye ok and continuing!
>>54039485
And it will STILL be superior to whatever AMD can hobble together because of their superior support. AMD's async advantage doesn't even add up more than a 10% performance improvement, if at all. The gains from async is entirely dependent on the developers and most developers have already stated how difficult it is to work with async computing. Hence, most DX12 games (if there will even be that many) will forgo async computing and other hardware dependent features in order to make better use of their time. Nvidia will retain most of their marketshare as a result of DX12 games running better on their hardware IN THE LONG RUN.
AMD's "advantage" (if you can even call it that) is short-lived and flawed from any standpoint. Not only does it involve more work for the developers to do (which costs them more money and precious development time), but it will also show how little AMD cards have to gain from supporting an obscure featureset and drive consumers to making a better choice with Nvidia's superior product and support because of the all important question: "Why should I buy AMD?"
There is no answer. Nvidia users have nothing to gain and more to lose by switching over to AMD cards. AMD's support for developers is limited, which will hurt them in the long run when they stop allocating time to work on DX12 features that benefits AMD cards. Meanwhile, Gameworks will continue to ensure that more games are better supported by Nvidia hardware.
>>54039618
5rupie been sent to your account ranjeet
Thinking of getting a laptop with a 14" w/ 970M or a 15.6" w/ 980M in it, to replace my desktop which crapped out. I can get a good deal on clevo laptops for both around 1400.
Is it a good idea to hold out until the next line of mobile processors or is now still a good time to get a laptop?
>>54039618
Do the new cards normally come in at the same price as the old ones?
>>54038796
>I find it unreasonable to compare those results to anything
25¢ deposited
>>54039349
>superior driver support
>nothing but unstable WHQL's for the last two months
>>54039587
Fermi had a hardware scheduler for that. With Kepler they offloaded that bit to software. Even with Pascal having preemption, it still requires a costly flush before performing an async operation. AMD needs no such flush.
>>54040209
>https://compubench.com/result.jsp?benchmark=compu15d&test=585&text-filter=&order=median&ff-desktop=true&os-OS_X_cl=true&os-Windows_cl=true&pu-CPU=true&pu-dGPU=true&pu-mGPU=true&pu-ACC=true&arch-x86=true&base=device
The T-Rex bench is and was VERY accurate GPU bench retarded...
>480 has 2316 cores
Coupled with the architectural and process enhancements, this card should be quite a bit faster than the 380, no?
>>54040347
between 380x and 390 performance, probably reaching 390/390x perf with an OC.
>>54040347
Comparing core counts between architectures is meaningless.
GP100 has only 16% more cores than big Maxwell and 25% more than big Kepler, but it performs 70% and 110% better floating point performance respectively.
>>54040569
AMD has already demoed it to perform at least on par with a 390x
>>54040824
>AMD has already demoed it to perform at least on par with a 390x
>believing amd marketing after bulldozer and rebrands
i have seen nothing from poolaris so far that shows it to be anything but shitty midrange offerings, especially considering the 200mm2 die size and leaked benchmarks.
>>54040303
And we have no clue what clock that engineering sample was at. Could've been in 2D mode even
This is literally what Windforce GPUs will look like in the Pascal generation
>>54040956
The Polaris 11 is a new 14nm R9 370 with a 128bit bus.
>>54040880
Live demos are pretty hard to fake. You can't just woodscrew more performance out of a GPU.
>>54041038
Why would Polaris 11 use higher memory clocks than Polaris 10?
>>54041129
>Live demos are pretty hard to fake. You can't just woodscrew more performance out of a GPU.
they never showed the GPU and the demo was locked to 60 fps, any card currently can achieve 60 fps in battlefront, even a 370 and 950.
>>54041129
>Why would Polaris 11 use higher memory clocks than Polaris 10?
Probably because the memory bus is very small
>>54041198
That was Polaris 11.
At the other conference they demoed Polaris 10 (probably R9 480) with Hitman performing like a 390x or better. That demo also ran on a machine where journalists could take a closer look.
>>54041311
>At the other conference they demoed Polaris 10 (probably R9 480) with Hitman performing like a 390x or better.
no, all they showed was hitman gameplay
>That demo also ran on a machine where journalists could take a closer look.
they only thing they were showed was a backplate (pic related), the card still could have easily just been a normal 28nm card. AMD has a history of lying and producing fake mockups (like the fury x2 they showed last summer, which we now know was completely fake as the final product required 3x pcie power to even be functional, which the mockup didn't have).
>>54041424
forgot pic
>>54038384
It's time to get a job.
>>54041424
Holy shit you are desperate. Nano is 175W, so you could run dual Fiji off dual 8 pins just fine. tripple 8 pins on the final version are just there for margin of error.
>no 490
JUST
>>54041645
>Nano is 175W, so you could run dual Fiji off dual 8 pins just fine.
nano has to throttle to half it's stock speed when gaming to achieve that though, and AMD never claimed that the fury x2 would be a nano x2. they straight up lied and deceived potential customers.
>>54041671
Everyone knew this. nvidia isn't releasing 1080ti until next year either.
It will probably be 1070/480 with GDDR5 in June. 480x/1080 with GDDR5X in August or September
Who waiting for Vega/Volta here?
>>54041690
>This much desperation
They called it dual Fiji. "Fury x2" is a name journalists used
>>54041764
Volta is 2018.
Vega is big Pascal (490x)
>>54038384
If youre that poor or budget minded you shouldnt even concern yourself with new technology.
DX12 won't change anything. DX11 is just starting to get used. The industry has barely moved away form DX9.
AMD's compute throughput is starting to look pretty serious. They need something to compete with CUDA tho. Hopefully open source. Pascal is gonna destroy Polaris. Vega will be on par with Pascal.
>>54041645
You're a fucking lying retard
The reason why AMDead needed THREE PCIe 8-pins (THREE FUCKING 8-PINS) is because the Fiji drinks electricity like a motherfucker and two 8-pins couldn't cut it. They didn't want a PSU-limiting R9 295x2 again with its fucking retarded current requirements that blacklisted 80% of all PSUs on the market. And yes, they did straight up lie to the consumers by pretending that it would only need 2 8-pin connectors just to show off "how energy efficient two Fijis could be". They straight up lied in front of the whole world.
>>54038919
Who cares? Mass Effect Andromeda won't release before 2017.
The 1080Ti should release just in time for it, making it perfect to go with my old 780Ti then.
>>54042154
>350W TDP
>150W less than 295x2
>Not possible to run off two 8-pins
That's well within the PCI-E specs, Rajesh.
>>54042314
the bus itself provides 75w and each 8 pin power connector provides 75w, IHVs actually had to add 2 extra connectors to the 295x2 to stop it from throttling.
>>54042314
Quit lying Pajeet
Two Fijis would eat up way more juice than 350W, and AMD knows that. That's why they did a bait and switch with the "prototype" board showing two 8-pins and added a third 8-pin for the running prototypes and production model.
Now use that AMD-shill money you've just earned and go buy a toilet so you can learn how to shit properly Pajeet
I'm tempted to upgrade to Polaris, my 270 has served me pretty well, but modern games are getting more memory hungry and 2gb is starting not to cut it.
Might go for whatever the equivalent of the 380X is though.
>>54042409
A GT 1040 then?
>>54042346
6-pin supports 75W, 8-pin supports 150W or so.
>>54042346
6-pins provide 75W. 8-pins provide 150W. Stop pretending to be retarded
>>54042365
If 295x2 can run off dual 8-pins, so can dual Fiji, but nice try
>>54042439
>6-pin supports 75W, 8-pin supports 150W or so.
>>54042440
>6-pins provide 75W. 8-pins provide 150W. Stop pretending to be retarde
if that was true then the 295x2 wouldn't need 4 8 pin connectors to prevent throttling
>>54042440
>If 295x2 can run off dual 8-pins, so can dual Fiji, but nice try
fiji is a larger die and draws more power than hawaii, and the vanilla 295x2 throttles heavily.
>>54042440
>If 295x2 can run off dual 8-pins, so can dual Fiji, but nice try
That's because the current (as in Amps) requirement for those 8-pins are way above what you'd normally find on PSU 12V rails, which means that you can only use a specific number of PSUs or else you have a literal housefire on your hands. Even if the PSU can provide the wattage, it needs to provide the right amount of current to get the card to work without destroying the PSU.
>have 290 tri-x
>only use computer literally ~4 weeks a year due to school 5 hours away from home
>play mainly BF4, Witcher3, etc
>290 plays them more than fine, especially after OCing
>yet I want to upgrade
considering selling my 290 and buying the Polaris X80/X80x, or potentially a 390x if they drop. am I an idiot? I don't game much right now, but it'd be nice to not have to worry that my card will be out of date by the time I get home, since I come home once every ~4 months
>>54042438
I'm kind of banking on DX12/Vulkan support making AMD the better option.
>>54042554
>I'm kind of banking on DX12/Vulkan support making AMD the better option.
it doesn't make AMD the better choice, just brings them to parity with nvidia performance wise
You can pull 170watts from either 8pins or 6pins.
Guess it relates to Vulcan, but has SteamOS, etc. got native support for it yet?
>>54042544
Have an R9 290 here too that I'm mostly content with, but I think it is far too loud and hot, so I'm thinking about sidegrading to a newer card, if the 200 dollar ones performs as well as an R9 290.
Any release dates yet of 1070 or 490?
>>54042544
yes you are, just wait till the 1080/ti/490x drops and see if the performance jump is big enough.
A 290 should last you at least 1-2 years more
>>54042597
This. People don't seem to realise that the two extra pins on an 8-pin are just ground wires. There are no extra 12V lines.
>>54042479
>Even if the PSU can provide the wattage
Electrical power is literally voltage * current. Since voltage on a rail is fixed (12V rail on the PCI-E connectors) it's impossible for a PSU to "provide the wattage" but not the current.
It's possible to buy a PSU that cannot supply sufficient power on its 12V rail(s), but then you just bought a PSU that's insufficient for your system, there's nothing special about it. Also have a look at 8 pin connectors. They have 2 extra ground wires and that's it. A 6 pin connector is in no way less capable of carrying current than an 8 pin. Having 8 and 6 pin connectors is more of a convention to make sure sufficient current can be delivered.
Even with a weak PSU you won't have any housefires on your hands, unless you maybe bought the absolutely shittiest, cheapest PSU you could find and it doesn't even come with OCP.
>>54042704
There's nothing concrete for their announcement, let alone release
Gee, i mean i admit, i do lean towards AMD. But these Nvidia guys are either shills or a cult.
>>54042792
>it's impossible for a PSU to "provide the wattage" but not the current
>It's possible to buy a PSU that cannot supply sufficient power on its 12V rail(s)
https://www.amd.com/Documents/Selecting-a-System-Power-Supply-for-the-AMD-Radeon-R9-295X2-Graphics-Card.pdf
Buy a toliet for Peet's sake, Pajeet.
>>54042704
There's absolutely nothing official at all. There are some rumors about May and June. Don't get your hopes up though, it's unlikely we'll get something really faster than Fury X/980 Ti.
>amd is fucking incompetent
>games rarely work, profiles comes out two years later
>high power consumption and heat
>cancerous "amd can do no wrong" userbase
>little glitches and bugs that requires a shit ton of time and effort
>Nvidia is fucking flawless for the first few years
>experienced my first actual non-meme driver gimping this week
>had to roll back to old drivers, gave me deja vu of my amd using days of looking for best drivers and ghetto fixes
I just don't know anymore.
>>54042819
What is that supposed to mean? Are you retarded? You do realize that delivered power is literally directly proportional to current and voltage and in this case, voltage is a fixed value, right?
All that says is that a 295X2 can pull up to 600W from the PCI-E connectors combined and up to 336W from each one individually. It's impossible, literally, for your PSU to be able to deliver that power via PCI-E connectors but somehow not deliver the current.
>>54035815
I just want to upgrade from my 2gb 7850 since launch. Been a loyal card though.
>>54042869
It means, Pajeet, that even if the PSU can provide the right amount of wattage, IT MUST SUPPLY NO LESS THAN 28 AMPS OF CURRENT PER RAIL TO EACH 8-PIN CONNECTOR, OR NO LESS THAN 50 AMPS ON A SINGLE-RAIL PSU.
That knocks out at least half of the PSUs rated between 650W-800W.
>>54042833
>>cancerous "amd can do no wrong" userbase
this applies to both user bases, just read this thread
what I've been doing/have done is
>buy 770 new
>use for ~2 years
>sell it for $250
>buy 290x
>gain 30% performance for free
which results in
>nvidia cards before they get gimped
>AMD cards after they get proper drivers
best of both worlds
as far as heat and power goes, if you can't afford the extra few dollars a month, why are you spending $400 on a video card?
>>54035815
fuck off
>>54042897
But it cannot provide 600W at 12V if it can't provide 50A. It's actually impossible, because DC electrical power is P = I * U, i. e. P = 50A * 12V = 600W.
If current (I) is <50 you no longer have a 600W power output.
Also, dear retard, individual PCI-E connectors don't necessarily have anything to do with power supply rails. There's absolutely nothing saying that each PCI-E connector has its own power rail. Lots (most) high-end PSUs don't even have multiple 12V rails in the first place, just a single one capable of providing very high current.
>>54042905
My gpu cost me $750 + $120 waterblock.
I'm fine with power consumption as long as it's within reason. AMD products makes me turn down the thermostat because it makes my room a sauna and in turn makes the rest of my house a freezer. Oh and it's not even summer yet.
I mean the 390 consumes more power than a 980 ti. Fucking really?
>>54042977
kek, an AMDrone trying to teach me about electricity
Go back to high school and bribe your physics teacher again, Pajeet.
>>54043011
>kek, an AMDrone trying to teach me about electricity
You should take all the help you can get.
Anyway, I'm going to bow out of this one, as it doesn't actually interest me to attempt to educate you, since multiplication and 6th grade Physics are apparently way beyond your comprehension already. Enjoy hammering in nails with your forehead or whatever other job your limited mental capacity allows you to perform.
>>54038689
Says who? Got any proof fuck boi? I have quite a few sources that say otherwise.
Inb4 falling for this bait
i dont really like my gtx 770 t b h
i plan on buying whatever amd replaces the r9 390 with
>>54038796
I like how you youve chosen to ingore the part where it was running battlefront with the same performance of a 960 at nearly half the tdp.
>but these benchmarks say...
They dont mean shit... like the article says they cant really compare it to anything since its a brand new arch and they have no idea how efficient it is
>>54043249
what went wrong with your 770 anyway?
>>54043272
thats was polaris 10 or 11?
>>54043000
you have no idea how bad power consumption used to be.
>>54038913
How can it support vulkan or dx12 with no async?
>>54043276
it works for what i play right now, i'm just disappointed with how it aged. it deteriorated pretty quickly.
Currently sitting on a GTX 670 2GB, do I wait for Pascal or do I take advantage of all the sales for a 980ti going on? I probably won't be doing 4k gaming or VR, just 144hz 1080 gaming really.
>>54035815
Single slot cards are BACK
>>54039164
Yeah at least nvida tells you when they stick their 3.5 in dick in you. Oh wait they just apologized and told you to buy a 980 like a good goy
>>54043356
i guess it's a matter of if they'res any graphically intensive games you wanna play between now and the new cards release.
>>54039164
>doesnt know pascal is a rebrand
>>54039164
>>54039244
>>54039349
So many shills. Did synagogue just get out or something?
>>54043289
Polaris 11, 11 is the smaller chip while Polaris 10 is the one that was running of Hitman at 1440p 60fps at capsaicin
>>54040303
>kung fury will always be top
>>54043379
Hm, my main games are generally CS:GO, LoL, and SC2. But I'm thinking about DS3 and Witcher 3 which are so ready out. I'm sure Total War games are possible buys when they get released.
>>54039100
And since AMD does have async and they greatly outperform Nvidia's cards while using it it proves AMD has the better hardware
>>54041014
I dont think itll be cool enough..
>>54043517
>they greatly outperform Nvidia's cards
do your research. AMD doesn't outperform NVIDIA in any dx12 or vulkan games/benchmarks currently.
>>54043538
MAXIMUM B8
>>54043544
kek, you do realize your coworkers shifted from claiming superior performance in dx12 and vulkan weeks ago right?
http://www.phoronix.com/scan.php?page=article&item=amd-nv-glvk&num=2
>>54038796
Nothing wrong with the results if you compare it with the right graphics card.
>>54043600
>GT 710
wew lad
>>54043559
>>54043538
I just purchased a 980ti. I hope I won't regret it. Please god..
>>54043538
>>54043559
>muh Njewdia
>>54043637
>those frametimes
top kek, amd drivers can't even stop stuttering on a card with 8gb of vram whilst the 970 maintains it's frames just fine.
>Want AMD's cards with Nvidia's drivers
>thiswillneverhappen
>Nvidia will keep overcharging for shit hardware with no h/w async and no real dx12.x+ support
>AMD will continue to make great cards with horrible drivers.
>Even nvidia have started putting out shit drivers again 3+ in a row now
i havent been able to update drivers all year wtf is going on.
Selling my 770 and getting a 490 or 590 next year
>>54043680
are you just pretending to be retarded or can you not read how that chart fills out? Nvidia is the one with the frametime spikes
>>54043701
>wanting Nvidia drivers
>>54043674
yeah my old 770 did the same thing it only gets 30-40fps max
Funny how the game also runs and looks like shit too.
>>54043695
>>54043637
Quantum break is unoptimized trash it runs at 720p internal resolution and you cant change it.
>>54043674
muh Vulkan
>>54043680
I don't know what the fuck I'm talking about: The Post
>>54043715
>are you just pretending to be retarded or can you not read how that chart fills out? Nvidia is the one with the frametime spikes
are you? the red line is the 390, it's very cleatly stuttering every other frame whilst the 970 is maintaining the same time across all frames.
go watch the video you pulled this cherry picked image from, https://www.youtube.com/watch?v=zK2BUeYqLVI the 390 is very clearly unplayable despite the higher framerate because of the stuttering
>>54043728
i got a 770 4gb oc so i dont have to worry about the 3.5 meme but yes the drivers are terrible atm.
i am sticking with 362.00 until i sell my card.
>>54043637
>>54043695
>Quantum Break
>>54043695
>>54043728
>>54043715
>>54043637
>shilling one game benchmark this hard
AMD is getting desperate it seems, trying to sell off the rebrand stock before you push out another hyped pile of shit, raja?
https://youtu.be/frNjT5R5XI4?t=9m50s
>>54042460
i run my 295x2 off a single 8+8 connector
it's fine, not sure if the cable is getting hot from electron flow or the fact that 295x2 is a toaster
>>54043753
Do you have down syndrome, a piece of steel rebar lodged in your brain stem or both? The 970 is clearly stuttering to 45+ ms constantly while the 390 stays far below.
Which card would be best for 3d rendering (mid range prices).
>>54043907
yeah, you're obviously pretending to be stupid or are color blind.
>>54043820
>Meme Souls 3
>tfw you fell for the 4GiB of HBM meme
>Check AMD forums
>every post is about weird hardware and memory clock issues
fuck sake the 290 has having these problems 3 years ago fix your shit amd
>Check Nvidiot forums
>Drivers unusuable.
What the fuck is going on.
Bad time to be a pc gamer right now.
>>54043980
Or be insane like me and go Intel.
>>54043952
>dark souls
>meme
>>54043939
>cant even read the chart correctly to notice Nvidia has the huge frame time spikes
>thinks this reflects well on Nvidia fanboys
>>54041424
>AMD has a history of lying and producing fake mockups
AMD never faked a product. I think you're confusing them with Nvidia.
>Muh woodscrews
>Muh 3.5
>(like the fury x2 they showed last summer, which we now know was completely fake as the final product required 3x pcie power to even be functional, which the mockup didn't have).
Are you aware of what a prototype is?
>>54044020
>when nvidia does it it's woodscrews
>when amd does it it's just a prototype, goy!
>>54043728
Nvidia drivers are shit, at least if you still run Kepler.
AMD is shit because of the rebrands.
At all, the best way is a used graphics card for cheap if its not a 970.
>>54044020
What about all the fake pascal mockups they kept showing off? They didn't just parade one of these around but two between the original showing and the newer Drive PX 2.
>>54044121
390 is superior to 970
>>54044188
Possibly, but its overpriced. You could get a 290 8 GB for cheap, and you won't need 8 GB unless you plan to 4K CfX.
>>54038689
Except they don't by any stretch of the imagination.
A 7970/280X/380X is now more powerful than a 970...
A 7870 is faster than a 960 and is 4 years old :)
nvidia is trash.
>>54044287
>amd shills actually believe this
top fuken kek
>7870/270X/370X faster than 960!
>7850/270/370 is too...
Cheapest 960 - $205
http://www.newegg.com/Product/Product.aspx?Item=N82E16814487091
Cheapest 370 - $140
http://www.newegg.com/Product/Product.aspx?Item=N82E16814127882
nVidia is literal shit.
It's so bad, Jen can't even find his pork shoulders.
>>54044346
>benchmarks from 3 years ago.
Looks like we found the nVidia shill...
Find me some benchmarks that show the 680 beating the 7970, then cry over >>54044352 once you realise nVidia is shit and supports their products for about 4-5 seconds before dropping them.
Okay be serious with me /g/, how important are a-sync and DX12? I expect console ports to make greater use of both, so would a current mid-low end Nvidia GPU be a horrible investment? Like if I was poor and NEEDED a GPU right this second.
>>54044371
Very important, see >>54044352
Expect a few posts after this from nVidia shills saying it's not important and nVidia does it better anyway.
>>54044432
>970 owned by a 960 in every bench
Did somebody forget to plug it in or something?
>>54035880
>to wait
>wait
>waiting forever and ever and ever and ever and ever
You might as well not have a pc get a fucking console.
>>54044485
That's the memory gimping for you
>people actually paid more for a 970 when a 960 performs better
>>54044485
it's photoshop m8
http://www.tomshardware.com/reviews/amd-radeon-r9-390x-r9-380-r7-370,4178-6.html
>>54044188
390 is currently cheaper/same price as 970 depending on the OEM you choose
Niggas do I buy this cheap ass 980ti or do I wait for pascal?
>>54044815
Wait for pascal
Literally no reason to buy a 28nm gpu in 2016
>>54044815
How cheap is it?
Pascal isn't going to improve on much over Maxwell, its biggest drawcard will be 16nm.
No async compute on Pascal (have to wait for Volta)
If it's cheap as shit, get it, if not, don't.
>>54043990
I am getting close to just selling my 770 and waiting a couple of months take a break from gaming and go back to Inthell intergrated m8
>>54044858
Won't Pascals be more expensive with probably equal or maybe a little better performance. The 980ti probably won't be replaced until like December when the x80TI model comes out, right?
>>54044909
no, the 1070 will launch at $300-350 and outperform the 980ti as well as have more vram. the 1080ti will be massively better than the 980ti but won't be out until q1 2017.
>>54044909
Hardware wise, yes.
But nVidia has already been at work gimping maxwell in recent drivers in readiness for Pascal, once Pascal hits, expect Maxwell to drop in performance like Kepler did.
980Ti/TitanX replacement isn't until Q2 2017 (same as AMD Vega 10) as HBM2 isn't shipping in volume until late Q1'17
>>54044947
This is some pretty wild speculation here, got any sources to back that up?
>>54045009
>This is some pretty wild speculation here, got any sources to back that up?
benchmarks got leaked ages ago on the 3dmark website, http://wccftech.com/nvidia-pascal-3dmark-11-entries-spotted/
>>54045009
>But nVidia has already been at work gimping maxwell in recent drivers in readiness for Pascal, once Pascal hits, expect Maxwell to drop in performance like Kepler did.
>tyool 2016
>still believing the gimping meme
>>54035815
>wccftech
>>54045044
Either the gimping meme is true or AMD is just the best with hardware period.
If the latter is true, nvidia should just give up.
>>54045078
how can you even quantify how much the hardware has been 'gimped'? no reviewers actually benchmark the 600/700 series gpus against the old 7000/200 series GPUs anymore, and the few that do are using reference cards or some other dumb shit and comparing them to AMD's OCed rebrands.
everything i've seen anecdotally has shown that these cards still hold their own and perform the same as they always have
>>54040030
get the MSI 14" gs
it has a 970m with a full voltage skylake processor and thunderbolt 3 so you can even get an eGPU down the line
I just bought a 980ti for $550. Was this a good choice or should I have waited?
>>54045243
graphics cards are pretty good right now desu. a 980ti will last you many years. there's always a shiny new jew card on the verge of being released, it's best not to think about them too much. unless, of course, you're a good little goy who needs the shiniest new toy every year even though you're playing at 1080p/
>>54045283
What if the shiny new jew card is released next month? Do you think it is alright for the good little goy to wait instead of getting something now just to get abandoned?
>>54045243
>mfw these cost $1000aud here
JUST fucking kill me
>>54045334
if you're okay with the hassle of returning your card in the hopes that they release the new one next month, yes. you also have to factor in higher price for new models + low stock of new items
>>54045367
>low stock of new items
Oh yes. That is the main terribad danger waitfags like me have to face.
>>54044432
0/10
Thank you AMD! What I really wanted was a poor performing card that doesn't use much power because electricity is hard to come by these days.
>>54043730
But amd still beat nvidia fukboi
I was considering going into the first iteration of AMD HBM2 GPUs.
But considering that I have a Fury X, and that I'm pretty happy with it, then I'll reconsider waiting for a bit since I do not think that the performance increase from both Nvidia and AMD won't be that much in comparison to the Fury X or the 980Ti.
I'm more hyped about 3D X-Point desu, I hope they come up with a port that can actually support the speed that it delivers, since the first 3D X-Point Storage Mediums are going to be bottlenecked like fuck because we currently do not have any slot in nowadays MOBOs that can deliver their speeds.
>3D X-Point Storage
>3D X-Point RAM
>3D X-Point GPU Memory
>3D X-Point dies on CPU eliminating the need for RAM
The future is fucking bright my friends, how long do you guys think it will take to have the techs I listed above?
Storage is coming out this year, so that's one of them.
>inb4 t.Intel Employee
what cards are right for me if all I want is max on 1080 at 60fps ?
>>54047266
What gaem?
a 280x/7970 will max anything you throw at it
nvidia well uhh ummm uhhh.
Yeah nah just dont go there get a 380 or 390x if you can afford it/is cheap
>>54047266
Get a 3.5.
AMD cards consume more power and thus are louder plus you get shit FPS in all GameWorks titles like the Total War games.
>>54047824
all
>>54047948
i played most newer games on my 770 4gb shitboard and with older drivers its fine.
only struggled on FO4 and Ashes of the Singlearity but their shit gamez with crap opimization.
Personally id recommend getting a 490 and then a two 590's next year should be future proof for years to come.
>>54047978
>recommending dual GPU
>futureproof when future of the company is debatable
>AMDs track record of crappy CFX support
I've owned a 7770, 7850, and now a 380X, I'm going to buy whatever 490 comes out, but god damn that's dumb.
How cheap could they sell Nano or Fury, if they decide to rebrand it? Is it economical to keep producing big 28nm chips?
>>54048054
>Is it economical to keep producing big [...] chips?
never has been, never will be.
rebranding them would be kind of stupid, they could literally port the design and it would be cheaper that way, instead they'll have designs that replace them for cheaper dies and lower power consumption.
>>54047992
ARM or Quallcom will buy tthem
just like amd bought ati
>>54048104
>Qualcomm
Rev up those housefires
>>54035815
What GPU should i get to speed up the rendering in blender? On some forums it was mentioned that CUDA would be faster, but on others it said that the R9 390X would be much faster than nvidia gpus and sometimes i read that a fast CPU would be even better than a GPU for some reason...
I have a R9 290 Tri-X OC and i don't want to spend thousands of EuroShekels for a new gpu, i just wonder if a nvidia gpu with cuda would be much faster, or if it's worth waiting for the new gpus.
>>54048190
last time i checked Blender could only use CUDA for gpu so for lowering render times definitely get an nvidia gpu. make sure you are using optimal render settings too.
>>54048436
With the new blender version it's also possible to render with amd gpus if you enable CYCLES_OPENCL_TEST=all, but i'm guessing that it isn't as fast, because it isn't fully supported yet.
And for some reason i can only render 1 tile at the same time with my gpu but 8 tiles with my cpu
>>54038071
Nvidiafags really have a hard time counting integers.
>>54038071
>one word
I'm not sure if you're just stupid or trolling
>>54039349
>superior driver support
>sets cards on fire
>>54048190
There's CYCLES for opengl rendering, but as far as I know they can't render everything as some features aren't supported.
>>54039349
Do you realize that dx12 is a low level api and the importance of driver optimization is nil compared to dx9-11 ?
>>54044858
Do we know if pascal support async? Any link?
the low end Polaris cards are coming out this year right? I don't want to get boned for buying a Nano.
When are we getting benchmarks?
>>54035923
Why would they do this?
I want a 750Ti/950 because I need something low power but decent. How likely is it that DX12 will absolutely kick Maxwell's ass and make it worse for games than even a console? I just want something about console-tier for a few years.
Pure speculation at this point, but anyway thoughts?
>>54046963
3D X-Point cannot be used as GPU memory, it's not fast enough for that. Hell, it's actually slower than DRAM.
>>54049429
To make you spend 4 times as much on a workstation card with the exact same chip, but a few 'extra' features enabled.
>>54036331
Poolooris
>>54039164
Nvidia is all marketing bro.
>>54044947
Ofc 980ti will be gimped as soon as 1070 will hit the streets
>That 960 trumping a 780ti lol
>Laughing at those who bought 780ti
>tfw I still own two GTX 480's (in SLI)
>tfw now when I tell people I have two "480's in SLI" they're gonna say stuff like:
>SLI? Don't you mean crossfire?
>you must be able to play everything in 4k, anon
>480's? Check your privilege you cis white scum.
>mfw
>>54049443
Wait for Pascal, then get the 1050. 950 sucks dik
Have a question kind of related to Polaris/Pascal thing.
I want a new GPU for myself, but don't know if I should wait for the new generation or just get something from the current one. I only plan on playing 1080.
Also, what are the chances that older generation will get a price drop soon after new generation will be released?
>>54050525
Most likely the new generation will top anything from the previous one (at least on Nvidia, Vega is the real flagship on AMD) for less price. If you like to invest the money properly, then wait.
>>54049423
New ones are coming next month. I expect nano and fury to drop in price >20% soon.
Will the price-drop of older generation cards be that significant? This jump in architecture seems to be a milestone but will it really effect prices significantly? Can we get 390X for half the price in 6 months time or will the prices for the higher gen cards hover much higher than anticipated rendering such notions moot? It would need some time to make these cards available in the first place and to keep up with demand if expectations are met. I mean a 780ti is still more expensive than a 970 or 390x and it performs worse than them in many cases. Effectively meaning that even though newer cards are better older cards can still be more expensive. If they price the new cards right then they can get away with keeping older cards as expensive as they are (or at least very slightly drop their prices) and make the new cards more expensive or at least as or around as expensive as old cards. They just can't have people not buying old cards with diminishing returns.
>>54050869
that never happens
you have to buy second hand to get old stuff for cheap
>>54050887
So why are people so happy that older cards will see a massive price-drop? Are they deluded or am I missing something? Is a $20 drop that significant? Even $50 is quite meh on the high-end stuff price around ~$350.
>>54050929
they probably mean second hand
>>54043753
are you genuinely retarded?
Dear people.
I was thinking about upgrading my whole machine lately but I heard about those new cards.
My current GPu is GTX660. Should I get 960 around June or wait for 1060 or how it would be called? Just wondering.
Im not sure about the price, too. I would probably import or something, but wont want to pay more than 250$.
>>54051371
The 1060 will not be out for a while. Your best bet is to buy a used 960 and resell it once the 1060 releases. It should have better performance than a 970 and use less electricity.
>>54046963
Wow we found someone that actually bought one.
>>54050465
That will be annoying actually.
>>54042792
are you saying i can plug a 6-pin into an 8-pin slot and have it work as intended?
Thank God. I always wanted a GPU that uses less power than a light bulb and isn't very fast.
>>54046963
>AMDrone bought a CoilWhine X
kek
Do you actually think that the consumer GP100s won't blow the Fury X and Vega 10 out of the fucking water? You're fucking delusional
>muh 3D X-Point faster than RAM
No, retard. X-Point is SLOWER than DDR3 RAM. Intel even said that themselves. Fucking AMDrones are this retarded. This is why they fall for the DX12 marketing hype and Zen.
>>54051485
What are the chances for 960 price drop in June?
>>54051565
You can find a used 4GB 960 for $150 on ebay, or a new one for just $30 more. You really can't get any lower than that in price point.
>>54042977
>tfw hd7750 popped
>mfw g3258 graphics barely run madvr
>Tfw waiting for poolaris 11
