[Boards: 3 / a / aco / adv / an / asp / b / biz / c / cgl / ck / cm / co / d / diy / e / fa / fit / g / gd / gif / h / hc / his / hm / hr / i / ic / int / jp / k / lgbt / lit / m / mlp / mu / n / news / o / out / p / po / pol / qa / r / r9k / s / s4s / sci / soc / sp / t / tg / toy / trash / trv / tv / u / v / vg / vp / vr / w / wg / wsg / wsr / x / y ] [Home]
4chanarchives logo
1070 benchmark
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.

You are currently reading a thread in /g/ - Technology

Thread replies: 255
Thread images: 39
Poo-laris BTFO

1070 confirmed for above Titan X performance!
>>
THANK YOU BASED NVIDIA
>>
>>54722049
>manufacturer A's unreleased product's performance is rumored to be "about X"
>manufacturer B is screwed

i do not understand how this makes sense to people
>>
Polaris is better. 1070 is neck and neck with titan x, polaris is far bit a head of it.
>>
>>54722117
http://videocardz.com/60253/amd-radeon-r9-480-3dmark11-benchmarks

>3DMARK11

OH LOOK, IT'S THIS DUMBASS AMDPOORFAG POO IN THE LOO AGAIN
>>
>>54722117
>CF

Hmm I wonder what that stands for??!?!!
>>
>1070 and 1080 were both cucked by nvidia putting only 8 pin
>overclocks are power limited
Lel.
>>
I find a strange sense of irony in that, people will always go to a benchmark, until that one specific benchmark happens to show a competing product performing better.

Almost have to wonder about the actual age of some people on this board.
>>
>>54722224
Isn't the 1070 rumored to have 25% of the CUDA cores of the 1080 disabled? Seems like it's punching a bit above its weight then in that Firestrike benchmark if either of those claims is accurate. Maybe the 1070 is more powerful than we've been lead to believe.
>>
>3dmark

Since when is 3dmark an accurate measure of performance? I need to see ingame fps.
>>
>>54722261

Or mental status. The brand loyalty here is straight up autistic.
>>
>>54722276
Eh it's pretty fucking accurate if the game isn't fucked with optimizations that show bias to one company or the other. Wait, you're right since that's basically no games these days lel.
>>
But is it actually 8gb?
>>
>>54722049
>1070 confirmed for above Titan X performance!

It wasn't a question whether the 1070 can beat a Titan, but a question of when did the drivers come out that gimped the Titan to 1070 levels.
>>
>>54722264
Good ol' Nvidia, gimping cards straight from the get go. Can't say I'm surprised, after all, this is the 970 2.0
>>
>>54722347
That's how different grades of GPU have always worked.

The "gimping" is actually mostly it's GDDR5 vs GDDR5X.
>>
>>54722346
>there are bakas in this sekai that think that Titan X is a consumer card or that people actually buy it for consumer use
lel kek lulz
>>
>>54722347
Yeah but AMD does that too. I mean the 380 and the 380x are technically the same GPU.
>>
>>54722347
Don't forget how badly they're gimping AMD cards into requiring 2 of them to make sure their flagship can compete with the Nvidia flagship.
Rememebr Nvidia controls all benchmarks real and imagined. Fucking taiwanese jew illuminati.
>>
>>54722343
Underrated
>>
>>54722347
>He doesn't understand product binning
290 is a defective 290x that didn't make the grade
380 is a defective 380x that didn't make the grade

970 is a 980 that didn't make the grade

960 is literally a 980 sawed in half (D/E/F sample)

>>54722343
Games are barely even using 4GB and it's not powerful to run 4K at slideshows
>inb4 futureproofing

A 970 will be obsolete in 2 weeks and drop down to sub $200 levels long before games required more than 3.5 GB of RAM.
>>
>>54722049
>>54722117
>3DMark Fire Strike GRAPHICS Scores
>3DMark GRAPHICS Scores

Are these graphs from the same benchmark or not?
>>
>>54722390
>Polaris
>flagship
Too obvious man
>>
>>54722316
Exactly.
For years now I've seen nVidia shills run straight to Furmark and 3Dmark..While AMD shills will always run to, well, anything that isn't Gameworks sponsored.
>>
>>54722449
>IT"S NOT AMD'S FLAGSHIP REEEEEEEE VEGA IS COMING OUT HONEST! IT'S NOT LIKE AMD HAS EVER DELAYED SHIT INDEFINITELY
2018 launch for Vega just in time to get BTFO by Volta.
>>
>>54722427
>drop down to sub $200 levels
SLI fags will buy these up before the prices crash that low thinking they're getting a great deal for a cuck card at $260.
>>
>>54722475
>SLI fags will buy these up
Why not get double the peformance of 2 1070's for $200 extra instead? It's literally more performance per dollar anytime. And 970s slied are terrible because of the ram issue.
>>
>>54722117
Fucking retard the 2nd result is crossfire
>>
>>54722473
Polaris 10 is the successor to Tonga (285, 380, 380x). Stop trolling me m8 REEEE.

>>54722499
I don't know man. People are stupid. I'm just saying they will create enough demand to keep 970 prices high. This shit has been going on since SLI/crossfire became "sort of" common over the past 5 years.
>>
>>54722448
nope theyre from different benchmark which is why the titan x has a much higher score on one than the other. idk why theyre using to completely different benchmarks
>>
File: 1454969778689.gif (941 KB, 627x502) Image search: [Google]
1454969778689.gif
941 KB, 627x502
>>54722536
I suspect jewery, from either Nvidia or AMD
>>
>>54722346
>t wasn't a question whether the 1070 can beat a Titan
No you're right, the question was whether it would beat the 980ti, beating the titan x is just icing on the cake.

Stay mad AMDrone
>>
>>54722533
I don't think it will, People bought the 970 because they wanted a cutprice 980. The prices have since dropped like a stone since the 3.5gb meme came out. Generally prices don't drop if demand is strong. And I know most consumers will avoid AMD like a plague even though they might be cheaper because of the heat/noise.
>inb4 my 380x doesn't even make a sound or go above 65c

Fans are noisy while gaming, heat is still produced and blows into your room requiring further ventilation. Most people dislike these things.
>>
>>54722049
If this is legit that's actualy insane, at 380$ this card will be a big hit
>>
File: 1448705572008.jpg (73 KB, 500x333) Image search: [Google]
1448705572008.jpg
73 KB, 500x333
>>54722117
>AMemeDrones not even intelligent enough to read a simple chart
>what is CF
>>
>>54722117
>same card code appears at the bottom with 75% the performance
>the card appears again with :CF in the code name

I wonder what these magical letters mean?
>>
It must suck to be a poorfag NEET right about now seeing these benchmarks BTFO'ing AMD. If you want quality, sure you buy it at a premium rate, but who cares if you make enough
>>
>>54722117
>42% scaling in CF
Amd and drivers

Pick one
>>
>>54722645
>our high end cards are better than their mid range cards, AMD FAGS BTFO

Can't wait for Vega to release and the nvidia tears it will bring.
>>
File: 8GtfmB.jpg (62 KB, 500x375) Image search: [Google]
8GtfmB.jpg
62 KB, 500x375
>Vega chips my released
>Judging midrange to high end silicon
>Literally retarded logic

Muh Nvidia hurrrrrrr
>>
>>54722676
These aren't Nvidia's high end cards silly

This is gp 104, the succesor to the 970 and the 980

GP 102, the succesor to the 980ti and the titan x will be their high end card
>>
>>54722049
Holy shit, Polaris 10 will have to be really good for people to not save up the extra 100$ for this kind of performance.
>>
$230 polaros
>equal to gtx 970

$339 1070
>better than 980ti


I dont really participate in console/gpu wars but are amd fans literally retarded?
>>
File: no corssfire.jpg (895 KB, 3729x3055) Image search: [Google]
no corssfire.jpg
895 KB, 3729x3055
>>54722117
god damnit this is getting annoying listening to people talk about the crossfire performance

we don't even know if this is real, but fuck me, look at that, crossfire scales 90-100% now, and that is barely above 50% scaling

assuming this is real, either the drivers are not worked out, and we will see a performance gain, or there are no crossfire profiles that are working correctly, so we will see a gain, but all this does is confuse idiots and give ammo to people who hate amd for some retarded reason.

so lets take the confusion out,

this is supposedly polaris 10, the 380 and 380x replacement.

we don't even know this for sure, its not vega, that's to far out to be showing silicon yet, but polaris is around the corner so we assume this based on the hitman benchmark.
>>
File: 1464110851656.png (417 KB, 890x1040) Image search: [Google]
1464110851656.png
417 KB, 890x1040
So this is where Polaris 10 (R9 480 & R9 480X ?) would be placed if the 22.6% decrease (calculated from 1080, Fury and R9 390x score decrease) in score (from 3DMark to 3DMark Fire Strike) is consistent with Polaris 10 architecture.
>>
>>54722943
In between the 980 and the fury yes, just like in the last benchmark
>>
>>54722881
If it's not real, it's laughable, If it is real, still laughable.
AMD fags here claiming that CF doesn't stand for crossfire is the most laughable thing ever and just adds fuel to the fire that AMDrones are retarded.
>>
>>54722881
>NVIDIA BTFO Threads the instant this shit comes out
>People with half a brain point out that CF is in there
>BUT IT'S NOT EVEN REAL I WANT TO BELIEVE WE DON'T KNOW FOR SURE BACKPEDALLING

sasasuge amdtards
>>
>>54723603
>>54723620
>I literally can't read words
>>
>>54723603
>>54723620
Pretty sure actual AMDfags have paid close enough attention to know that the 480x will be an upper midrange card, not better than the Titan X, lel.

DESU the benchmarks are probably legit but we still have to wait a while to see some real performance with aftermarket tweaks for all brands.
>>
>>54722881
Maybe because you know, AMD's stellar crossfire drivers are at it again
>>
>>54723683
These are good proxy's for performance though because all cards in that benchmark are probably stock
>>
>>54722049
Well I guess we know what everyone looking to upgrade from 970/390 performance will upgrade to this year
>>
>>54722049
buy it goy, at the end of the day is going to be gimped when another card comes out
>>
>>54723813
The only gimped cards are AMD's which can't even match Nvidia performance which more cores and wider memory busses

>le gimp meme

https://www.youtube.com/watch?v=DvCsT_dk-1E
>>
>>54723825
BUT YOU'RE NOT POSTING THE CHERRYPICKED MEME BENCHMARKS REEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE
>>
>>54723762
Oh yeah no doubt, I'm just saying if it turns out that Pascal overclocks as well as Maxwell did compared to the latest from AMD then it could shake things up enough to be interesting.
>>
>>54723861
Yeah should be pretty awesome
>>
File: 1262163212155.jpg (11 KB, 220x168) Image search: [Google]
1262163212155.jpg
11 KB, 220x168
>>54722049
>incoming shitload of buttcheap used 980ti's
>>
>>54723891
>There are retards on this thread that spent $600+ on 980tis knowing full well that a die shrink with literally exponential performance is coming out.
Jesus take the wheel.
>>
AMD IS FINISHED AND BANKRUPT
>>
Holy shit how cringey was Jensen's announcement of the 1070.
>>
>>54723825
nice shilling panjeet
>>
>>54723966
Awesome rebuttal Pajeet, totaly discredited that evidence
>>
>>54723976
get a trip so i can filter you nvidiot
>>
>>54723976
>evidence
All he proved was AMD gets better over time while Nvidia is stagnant. AMD once again is the smart buy.
>>
>>54724013
He proved that Keppler isn't getting gimped, but the gap between older AMD cards and it closes because 'newer' AMD cards still run on the same old shitty GCN architecture and thus benefit from optimizations for their rebrands

If they actualy gimped their cards like you claimed, there would be a performance decrease on Keppler cards, and there isn't
>>
>>54724013
That is easy to do when it is the same GPU year after year.
>>
>>54724025
That's exactly how an Nvidiot would spin that conclusion
>>
>>54724046
That is what an honest person would see if they looked at the numbers.

Where do you see a performance decrease from gimping? Show me
>>
>>54724025
>>54724037
What does it matter if it's easy to do it? What does that matter to the consumer, you idiot? Fact is AMD cards are getting supported for longer, whereas old Nvidia cards are stagnating.
>>
>>54724013
>gets better over time
Maxwell got better over time, they just gave up on kepler because maxwell actually makes them money right now and not kepler, they don't sell kepler cards anymore so they don't give a fuck.
I don't think this is scummy. Only retards keep GPUs for longer than 2 years and expect them to keep up with software updates.

7970s are still being sold by AMD today tahiti and tonga are basically the same shit, It's actually the fact that AMD recycles so many cards that they still have 'driver' updates.
>>
>>54722049
It's over guys. AMD won this round.
>>
>>54724063
Most AMD cards are literally still being sold today.
It's not the fact that they're supporting them longer. It's the fact that they're rebranding the shit and selling them today still while Nvidia has moved on to bigger better things.
>AMD are saints
Anything pre GCN is forgotten like an ugly stepchild.
>>
>>54724063
>Fact is AMD cards are getting supported for longer, whereas old Nvidia cards are stagnating.

They get supported for longer because they keep reusing old architectures. If you think this is a good thing you are tech illiterate
>>
Nvidiots literally on Iron Lung life support. Someone just pull the plug please.
>>
>>54724070
How did you arrive at that conclusion?
>>
>>54724090
So fuck anyone who bought anything before 9xx series?

>>54724095
They're keeping up well for their price, I don't see a problem
>>
>>54724090
>Make extremely powerful and versatile architecture
>Let it mature and grow over time
>End users see boosts in performance over the years
Seems like a pretty based company DESU.
>>
>>54724068
7970 is still relevant today.

unlike pre-war shit from Novodya
>>
>>54724095
>>54724090
>they keeap reusing old architectures
>AMD only has to release actual new cards every 4 years to stay competitive
So you guys are saying Nvidia is just shit at making GPUs then?
>>
>>54724095
And they have iterated upon GCN tbqfh
>>
>>54724141
They sell bigger chips with much higher power consumption at a lower price to stay competitive yes.

That's what you do when the competitor has a superior product, you cut prices and take a smaller margin.
>>
>>54724123
>So fuck anyone
It's not fuck 'them' it's 'that's all the performance you're going to get and we'll do bugfixes but we're refining the new architecture now because we're actually selling those cards right now'.

It's literally the same with AMD their architecture is just older and recycled. They're still making the same old stuff they were years and years ago and selling it.
Nvidia stopped selling Kepler cards for more than 2 years now.
>>
>>54724145
really? what features did they bring to later versions that they also did not bring to 1.0?
>>
>>54724128
the 7970 is basically still being sold by AMD as the r9 380x today. It's not surprising it's still relevant but it's severely gimped by lack of ram.
>>
>>54724174
Houseonfire 2.0
and
Nodrivers 3.1
>>
>>54724174
I'm sure you know how to use google, buddy
>>
>>54724198
You made the claim, it is up to you to defend it.
>>
Can't wait for the geforce 1070 but I have to. Waiting on MSI or EVGA to get their hands on it. Then that's all me.
>>
>>54724216
You've made that claim that AMD's been reusing the same architecture, you haven't provided any evidence to support your claim either (cause it's bullshit)
>>
>>54724216
https://en.wikipedia.org/wiki/Graphics_Core_Next

Fucking read for fucks sake.
>>
>>54724237
It's common knowledge that they are still on GCN
>>
Nvidia will win if you can actually fucking buy a 1070 for $380
Good luck with that though
>>
>>54724235
same. looking forward to the evga 1070
>>
>>54724237
>he doesn't know that AMD has been stagnant on GCN for literally 3 generations of cards now

It's like an ignorant retard that just came out of the cave yelling at you that the sun doesn't really exist because he's never seen it before.
>>
>>54724185
Spiritual successor but slightly new arch with updated GCN.

And no, those chips aren't RAM starved. 2GB and 4GB 380s perform fairly close at 1080p. The 380x only sports 4 more CUs, hardly enough to soak up the bandwidth provided by 4 GB/256 bit bus.
>>
File: 1453398618703.jpg (45 KB, 717x638) Image search: [Google]
1453398618703.jpg
45 KB, 717x638
>>54724277
You're a retarded cunt
>>
>>54724247
>>54724248
>>54724277
GCN has been through several iterations. Nvidia has done the same shit since Fermi by making minor improvements with each generation, they just don't have an overarching name for it like AMD does.
>>
>>54724298
Doesn't really matter, it doesn't have enough juice to go past 1080p anyway. Which is laughable because everyone was mad with Nvidia for the 3.5GB meme when the 970 was literally marketed as a 1080p card.
>>
File: Roadmap2.jpg (933 KB, 3000x1687) Image search: [Google]
Roadmap2.jpg
933 KB, 3000x1687
>>54723683
this
AMDfag here. AMD said Polaris will be 390x performance at a lower watt/cost. It will be AMDs mid-range card. Fury will be high end until Vega. AMD is focused on low TDP with Polaris.

AMD people you really need to read before you come on here talking shit.

>>54723806
I'm still interested in Polaris or Nano for my mini M-Case build.
>>
>>54724247

GCN 1.1
> New audio engine
lol ok

> Extra hardware in the card.
We added more items to the cards! ok

GCN 1.2
> Overclocked processor for faster tessellation
ha

> New audio video decoder
Eh, most AMD users are Linux fags and can't use that anyways.

> Delta color compression
GUYS! We created an image compression that lives on the GPU. Have to have the latest GPU to use it. So no one uses it.

and I got tired of reading. Totally ground breaking things there. Esp updating the audio engine on A GRAPHICS CARD

> inb4 amdrones defend these features.
>>
>>54724324
Are you honestly going to debate the jump from Keppler to Maxwell was just an iteration of the same architecture?

Are you tech illiterate?
>>
>>54724185
thats wrong.
the r9 380/x is not a rehash of the r9 280/x
unlike the r9 290/390
the 380/x(gcn 1.2) is actually more up to date than the 290/390(gcn 1.1)
>>
>>54723683
>>54723603

i honestly think its people who like nvidia over amd that are calling the crossfire benchmark the polaris 10 just to make amd supporters look like idiots.

i mean look at it this way, we aren't even seeing the normal scaling with that, if we were we would see it at 30-32k.

something is seriously off with the benchmark and it is at best just causing confusion.

>>54723620
take the crossfire out as there is something wrong with that,

amd has a price range for the 480x, and that is 230$-300$, if this performance falls in that price range, this is an amazing card, but nvidia lovers are claiming its going to cost as much as the 1070 founders edition... which is litteraly fukcing retarded.

we will likely know within a week the pricing of the polaris, and if it is in that 230-300 range, it will be a good price performance card especially at 1080p, and depending on how the new cards deal with vr, possibly the best option for vr as well.

with vega coming out later on in the year to deal with the 1440 and 4k segment.
>>
>>54724311
Says the retard that doesn't know that AMD is still using GCN.

>>54724324
>Doing the same shit
[citation needed]
Maxwell and Kepler are literally totally different architectures.
>>
>>54724270
Would be surprising if we could. $430 is more likely.
>>
File: victimgirls11_08.png (525 KB, 1030x1500) Image search: [Google]
victimgirls11_08.png
525 KB, 1030x1500
>>54722049
I told these dumb AMD cucks that the 1070 was slightly more powerful than the Titan X. But no, every single one of them had to go
>but you don't know that
or
>480X is faster

And now the official number is out and I'm proven right. What are the AMDrones saying now?
>480X is a 1060 competitor!
Moving the goalposts, I see. Even when the 480X will be priced in the same bracket as the 970/1070 and the 480 will start at $329.

Fucking AMDcucks
Pic related. Nvidia is Guile and Polaris 10 is the little girl
>>
>>54724346
It's not a rehash, but tonga and tahiti aren't different enough to be anything other than an update to GCN 1.2.
They're basically the same chip design/size probably done to make it cheaper to produce since tahiti was getting long in the tooth.
>>
>>54724342
Pascal is Maxwell on "16nm" though :^)
>>
>>54724353
Yeah, the alternative currently would be getting 80% of the performance from AMD for 60% of the price

It's like I'm back in 2012 again

Only this time amd has Vega up their sleeve
>>
>>54724270
What do you mean "actually?" That's the launch price. The other people who mod the cards will probably boost the price though. I'm gonna very interested to see what the difference between stock and OC is. I'd probably not go more than like... 420. At 1080p 60fps the 1070 is probably overkill as it is and i can't really even think of what big PC game i'm looking for this year after Blood and Wine.
>>
>>54723702
crossfire and sli are a crap shoot at best, nvidia puts out more profiles faster, in some cases of gameworks games only their cards work in a dual setup, but when amd gets their profiles out they usually beat sli in performance, and i refuse to believe amd wouldn't get the best profiles out for benchmarks, so the cf bench is likely off by a wide margin.
>>
>>54724373
It is, but Keppler to Maxwell was a completely different architecture while that mongoloid claimed Nvidia was using the same architecture since fermi
>>
>>54724378
This isn't an official benchmark. This is leaked shit.
>>
>>54722137
I don't know shit about shit, in what way is 3DMark11 biased toward AMD?
>>
>>54724371
Nvidia would be a competition if they offered a GPU to max out games for 300$
They don't/
>>
>>54724373
Everyone knows that Pascal is a die shrunk Maxwell, Kepler is not Maxwell in any way shape or form.
GCN is basically the same shit rehashed across 4 generations now with 1.4 basically a die shrink.
>>
>>54724371
>the 480 will start at $329
Why do you have to make up lies to support your point? You're already correct about the 1070 outperforming it. Few people doubted it, and shitposters and trolls you encounter here are not evidence to the contrary

What we do think we know is that the 480x should be $299.

We'll now have $299 fury's, $399 titan x's, and hopefully $599 titan x +20%

I don't really see how people are upset about this gen. Pretty good intro and we know we have more to look forward to

Wait fags won
>>
>>54724419
Neither does AMD, dumbass
>>
File: 1454967220281.jpg (11 KB, 197x197) Image search: [Google]
1454967220281.jpg
11 KB, 197x197
>>54724422
>now with 1.4 basically a die shrink
>>
>>54724377
No, the launch price is the founders edition
Msrp doesn't mean fucking shit if even the reference cooler costs an extra $80. There's legitimate reason to doubt you'll be able to get a decently cooled pascal card at Msrp

I mean if we're wrong then everyone wins, I just doubt it
>>
>>54724436
Pray tell how GCN 1.4 is going to be different from GCN 1.3 apart from lower power consumption and performance gains attributable to die shrinking. Other than renaming it for marketing reasons to Polaris?
>>
>>54724436
Wait it's GCN 1.3*
>>
>>54724419
Nvidia literally has the entire high end segment in their pocket.

And they will soon come with a 1060 to shit on polaris's mainstreamcards
>>
>>54724431
AMD does not have a single competitive product at any price point this generation
>1070 costs $380, yet performs significantly better than a $300+ 480X
>1080 costs $500 and will almost certainly undercut Vega 10 ($650) at price-to-performance by dropping the MSRP when it launches
>the 1080 Ti will cost the same as Vega 10, but deliver significantly better performance
>the 1060 will cost under $250 and beat out both the low-spec 480 and Polaris 11 at the same price
>the 1050 will likely do the same because Polaris 11 is a weak, underpowered piece of shit
>>
>>54724473
>Nvidia doesn't have a single competitive product at any price point this generation
FTFY.
>>
>>54724469
>October
>Soon
>>
>>54724455
Okay. Fair enough. All we can do is wait and see. I've got a bunch of gift cards lying around cause certain people decided *not* to show up for Christmas. So it depends on how soon we get legitimate benchmarks and aftermarket tuned models out for comparison.
>>
>>54722049

Will I be able to make a neural network for a reasonable price soon?
>>
>>54724455
MRSP means manufacturer recommended retail price.
Which is the price at which Nvidia will stop selling boards to you if you try to undercut the other manufacturers. The basic standard card will sell for the MRSP because that's literally the minimum price that Nvidia allows and manufacturers are free to charge whatever they like for their premium shit just like Nvidia does with their reference cooler. Which btw is full aluminium with a vapour chamber and not a piece of plastic with 3 fans.
>>
>>54724487
Delusional
Tell me why any consumer would spend $329+ on a card when there is a $380 option that not only outperforms it by a significant margin, but can be overclocked to a much higher frequency, thereby yielding much better performance on top of that?
Why would anyone buy a 480X at that price? There is no point when a few dollars more will get you a much better, faster, and user-friendly card.
>>
>>54724487
Actually they do, basically the pattern for the past year has been Nvidia release cards with a certain price and AMD cut prices to $30-40 less for similar performing cards.
>>
>>54724473
>doesn't have any performance numbers for either 1070 or 480X
>no pricing info on 480X
>other bullshit
LOL nigvidia pay2please xddd
>>
>>54724473
>implying msrp for 1070
>lying about msrp for 1080
>making up prices for the ti and 1060
>making up performance for the 1060, Vega, and ti

Why can't you just make up your opinions based on facts and not imagination?

Because I bet that Vega 10 will cost less than the 1080 and outperform it.
>>
>>54724495
>soon
Not soon enough. The 1080 Ti will have around twice the raw computing power as the 1080, which means that it will undoubtedly beat the full Vega 10 at its own game. The 1080 Ti just has to show up at the same price ($650) and the Vega 10 will become a best seller like the BetaMax player.
>>
>>54724520
They won't Polaris can't launch at those prices, they'd have to start at $300.
Nvidia is forcing AMD prices down. This is what competition is supposed to do. I don't get why everyone is yelling about how Nvidia are Jews and AMD are saints.
They do the same exact things with prices.
>>
>>54724512
The founders edition cooler is literally the same cooler they put on the reference 980ti dressed up with a new shroud. It's not $100 worth of cooling.
It will be several months before 3rd party cards actually sell for MSRP
>>
>>54724473This guy works and AMD and Nvidia and knows all the inside scoop. No need to read future tech news. It's all here.
>>
>>54724520
Because you're literally imagining the price point of the 480x, literally imagining the overclocking potential of both cards and imagining that the 1070 will be able to overclock by a higher percentage than the 480x, and implying that you'll be able to get a 1070 at Msrp with a cooler good enough to overclock

You're literally making things up
>>
>>54724529
>doesn't have any performance numbers for either 1070 or 480X
It's posted in the fucking OP, dipshit. Polaris 10 has, at most, 5 TFLOPs of compute power. The 1070 is simply more powerful and only costs a few dollars more than the $329+ 480 (not even the X)/
>>
>>54724531
Probably, but by the time Vega 10 comes out and if it outperforms it. Nvidia will just cut prices like they have done consistently throughout the life of maxwell.
>>
>>54722049
How much better is it compared to my HD 7970? It works fine but i'm thinking about upgrading my monitor...
>>
>>54724568
Nvidia literally only cut maxwell prices for their own cards
970 was $329 constantly and the 980 only dropped to $500 after the release of the ti
>>
>>54724473
1070 is is 450$ and will be dropped to 380 at some point
the 1080 is 700$ and is barely better then the 980ti a card that was 650smrp and is now, 540$ new... wow what a great lineup

meanwhile amd has a fairly large jump in their 380/x lineup, for a yet undisclosed price, we are assuming between 230 and 300$

if its 230, its the new bang for buck card, if its 300 then it still has a place with the 1070 being nearly 100$ more after the at some point price drop.
>>
>>54724564
Are we making up prices now? The 1080 will cost 1080 dollarydoos!
>>
>>54724564
>The 1070 is simply more powerful and only costs a few dollars more than the $329+ 480 (not even the X)/

Too bad the 480x will be about $300, and the 480 about $250.
>>
>>54724568
The Vega 10 CAN'T outperform GP102. That's why AMD has to resort to the Vega 11 to do that. And word is that Vega 11 is a dual GPU card, not a single wafer like its competitors.
>>
>>54724572
Well the 1080 there seems to be over twice as good as the 970. So compare your card to the 970

It's not an exaggeration to say that a 1070 will be about 3x as fast if not more than your current card
>>
>>54724583
Yeah because the card they're competing with the r9 290/390 is exactly the same price.
>>
>>54724394
which is why im saying there either are not proper drivers all around and we will see a performance jump, or there are drivers but no proper crossfire profiles and we will see a jump there.

how good these cards are has nothing to do where we have them ranked, we knew this for close to if not more than half a year, it's all about the price point.
>>
>>54724587
>450$
Nope, it's $379 for all non-reference cards. Third party Zotac 1080s are already listed for well under the $699 mark at roughly $600
>>
>>54724473
Linus visits /g/
>>
>>54724409
isnt 3dmark one of the programs that if you had amd it ran like hell because intel paid them off back in the day?
>>
>>54724601
>word is
First I've heard of that
>>54724624
Because I'm saying Nvidia doesn't cut prices to compete with AMD, even when amd clearly wins. The 970/390 was neck and neck, but the amd fury outperforms the 980 in almost every case. No price drop for the 980 then.
>>
>>54724601
>The Vega 10 CAN'T outperform GP102.

We have 0 benchmarks for vega 10 OR GP102. Shit, we barely have any SPECULATION about what the hardware of GP102 will look like, and none for vega 10. Stop making shit up and go hang yourself, faggot.
>>
>>54724649
>making excuses
No, that was another benchmark and Polaris 10 doesn't even have the same raw compute power as the cut-down GP104. Don't even try to use this bullshit of an excuse to explain the 480X's poor performance.
>>
>>54724461
so far, and to my understanding, the shaders are called gcn, everything around the card has been re done at the very least, not sure about gcn at its core due to amd not talking much on it.
>>
>>54722427
>960 is literally a 980 sawed in half
Nope, 960 and 950 have a different GPU (GM206 vs 204 for 970/980). Obviously very few actual differences but it does have fixed-function HEVC decode and a few other extras that didn't make it into 204.
>>
>>54724564
There is no credible source, and compute performance does not reflect gaming performance. Look at the 980ti against the Fury X; the Fury musters nearly 9TFLOPS, the 980Ti gets 6TFLOPS, but the 980Ti trades blows with the Fury X. Stop shitposting fucking mouthbreathing retard.

You don't have any of the cards that you just stated; the only card we know the explicit performance of is the 1080, which is merely a paper launch. When the product launch comes and the clocks are revamped or a different cooler or a new driver rolls out, these can be vastly different.

>$329+ 480
Once again, you don't know that. You're clearly delusional and trying to cream yourself to graphics card predictions and your ideal world where AMD is nonexistent and every gamer suckles on the teat of Jen-Hsun Huang.
>>
>>54724600
>will be about $300
[citation needed]
>480 will be $250
[citation needed]

>>54724636
No manufacturer is stupid enough to shoot themselves in the foot to lose market share by going way over the MRSP. These deluded AMDumbs are just grasping at straws.
>>
>>54724545
the 1080 already costs 700$ and msrp at 600$, you really fucking think a 1080ti would only be 50$ more if its double the processing power?

whats more likely is the first release, where the ti is 20% more powerful, and titan will be an extra 40% over the 1080.
>>
>>54724667
GP102 is a consumerized version of the existing GP100. That means the same die size and core configuration, but with the FP64 cores cut down or deactivated. We know what the GP100's raw power is, we just need to reduce the number by about 75-80% to get GP102's power figure.

Now multiply the lower end Polaris 10's TFLOP number by two to get Vega 10's TFLOP.
GP102 is still ahead by a margin of 10%.
>>
>>54724662
>AMD clearly wins
Fuck off, they're literally neck in neck.
>AMD fury outperforms the 980 in every case
So that's good for AMD then, I don't know why you're complaining?
>>
>>54722676
Literally the same shit said Last year before the fury dropped.
>>
>>54724549
>Nvidia is forcing AMD prices down.

the 380x was 230$ msrp at luanch
the 380 was 200$
the 280 was 250$
the 280x was 300$

they have a range of 200-300$ to launch the card at, why would anyone think it would be over 300$
>>
>>54724700
>the Fury musters nearly 9TFLOPS, the 980Ti gets 6TFLOPS, but the 980Ti trades blows with the Fury X
That's because of the drivers on AMD side is fucked. The Fury X should be doing better than the 980 Ti and the Titan X by any measure, but it can't and never will because AMD can't seem to make drivers that can squeeze every ounce of power out of the GPU.
AMD sucks at making drivers for their cards. That's a well established fact.

>>54724705
What happened to the 980 when the 980 Ti launched? $50 price drop.
Fucking dumbass AMDrones.
>>
>>54724725
Learn to read

I said that Nvidia doesn't cut prices for AMD even after clearly losing. I said that the 970/390 was not that case, but the fury/980 was

You're more interested in making your shitty retarded corporate fan club look smarter than the other retarded corporate fan club than actually having a real understanding of the new gpu market
>>
>>54724703
citation
basing the prices off prior 280 and 380 cards, they have a range of 200-300$ to play with.
>>
>>54724739
>why would anyone think it would be over 300$
Why did AMD put the tag "$329-$650" on their Capsaicin event poster?
>>
>>54724703
>lose market share by going way over the MRSP

litecoin mining, it doubled all the cards prices for 6 months.

now, imagine this on nvidias end, they paper launch hard and can't keep up with demand, so prices on non refrence go higher then founders because they are actually better versions of the card.
>>
>rebrandeon launching 'new cards at these prices'

280x is the 7970
380x is the 285

Please quote the launch prices of these cards and not the price of the exact cards ones with new stickers.
>>
>>54724800
for
>>54724739

>>54724764
The fury only 'clearly beats' the 980 on >1080p resolutions. 99% of people don't have 4K monitors. If the 980s start stalling in terms of sales Nvidia will cut prices.
Doesn't really matter now because they don't even make them anymore because the die shrink is out.
>>
>>54724753
no it wasn't drivers for the fury line, its literally they made the card for the 20nm process and got fucked hard by their fab.

granted a good portion is also driver overhead, but that's not the majority at least on the fury.
>>
File: 1405887386259.jpg (32 KB, 300x375) Image search: [Google]
1405887386259.jpg
32 KB, 300x375
I hope you all get paid to get this angry about unconfirmed synthetic benchmarks and rumored prices of consumer products.
You should seek help, if you don't
>>
>>54724753
Exactly my point, drivers are also a limiting factor the raw performance of a GPU. Most people don't realize that "TFLOPS count" is dictated by a standard formula (IPC * shader count * core clock in GHz); but most assume IPC is 2, but this is all situational dependent on workload, including gaming.
>>
>>54724800
>380x is the 285
380 is the 285. 380x is the uncut version of 285 that didn't make it into 200-series.
>>
>>54724825
>I hope you all get paid to get this angry
Earning a little side money isn't bad at all. I get to have fun triggering fanboys, too.
>>
looks like i still wont be using freesync on my monitor this year
>>
>>54724791
So it's Nvidia's fault that AMD cards are being bought for buttcoin mining?
Jesus fuck retailers sell at the highest price they can get away with. It's called capitalism, fucktard and AMD doesn't get a single dollar over what the board partners give them.
RETAIL =/= MANUFACTURING AMD didn't benefit at all from the buttcoin craze they suffered badly and lost market share because retailers pushed up prices due to fuckers buying up all the cards to mine memecoins.
>>
>>54724844
Yeah, and tonga is basically the same as tahiti with new gcn 1.2 memes.
>>
>>54724879
PROTIP: first-gen FreeSync monitors have een gobbling up obscene amounts of power because it forces AMD cards to run at a much higher clockrate at all times.
You fucked up.
>>
>>54724890
Well they did benefit but not because the prices jacked up. They benefited because they sold more cards than they could shit out in time.
The only people screwed there were the consumers by the retailers.
>>
>>54724800
550 on the 7970, 500 on the 7970 ghz edition
285 if i remember right launched at 250$

lets also not forget 7970 is around 4 and a half years old now. wish i was looking into a card at this point in time so i knew how long it stayed 550 and knew when it hit the 300$ range.
>>
>>54724771
got an image for that?
>>
>>54724961
I got a 280x for $150 this year before the 380x launched and was announced.

It's basically a brand new 7970 for 1/5th the price 3 years later minus the wear and tear of age.
I really don't understand why people crow on and on about 'it lasts really really long! AMD are gods. The fact is that the fans will probably start failing (I replaced my 6950 because the fans were rattling) the cards are not designed to go for longer than 2 years in the first place. It's nice to have a performance 'boost' because AMD is still stuck on GCN but the other components of the card like the FUCKING FANS THAT HAVE TO BLOW 24/7 TO KEEP THE THING COOL will die anyway.
And I own both Nvidia and AMD cards.
>>
>>54725050
Nice, I own a 7950 (280X) for $80. Honestly, aftermarket cards last pretty long if the upkeep is alright. My last 6850 failed after 6 years a few weeks ago, hence the new card. QC on aftermarket coolers need to be a little bit better IMO, having a BLDC fan whirring all the time should last way way longer, nothing to explicitly do with thermals. Plus, having 1 fan on a Kepler card will last a lot shorter than 3 on a GCN card.
>>
>>54724771
>Why did AMD put the tag "$329-$650" on their Capsaicin event poster?
Those are the prices for the AMD "VR ready" cards you can buy TODAY. $650 is the price of Fury X and $329 is the price of 390.
http://www.amd.com/en-us/innovations/software-technologies/technologies-gaming/vr/vr-ready

This entire thread is a retard shitfest
>>
>>54725219
Who the fuck would pat 1k for a R9 Nano?!

http://shop.amd.com/en-us/promo/ecxNCIUS125808?k=liquidvr
>>
Currently running a GTX960. What can I expect when I upgrade to a GTX1070? 110 more GTXs?
>>
>>54725249
IT'S CUTE DELETE THIS REEEEEEEEEEEEEEEEEEEEEEEEEE
>>
>>54725256
about 2.75x the performance considering the 960 is basically half a 980 and the 1070 is about 1.75x the performance of a 980.
>>
Damn. The NigVidia damage control ITT is making my dick rock fucking hard.
>>
>>54725256
>upgrading from one generation to the next
unless you want 4k or 2560x1440 there is no point upgrading from a 960 yet.

>>54725296
Your maths and your starting values are wrong.
1070/Titan X are only 30% better than a 980. 980 is about 90% better than a 960. 1070 comes out at 2.5x a 960
>>
>>54725457
>unless you want 4k or 2560x1440 there is no point upgrading from a 960 yet.

1440p or VR. I think it will be the big pascal before we get a decent single 4k card.
>>
1.7%
20%
whirran noises
introducing the nigvidia GT 300 20% underclock pro beaner-too many teats edition
with 1.7% more wood screws
>>
>>54725524
I miss wood screws but I don't miss inelite.
>>
>>54725517
If you play at reasonable settings even a 980 or 390x can do 60fps 4k and it looks a lot better than playing the same game maxed out at lower resolutions.
There is no reason to not get a 4k monitor if you can afford it.
>>
>>54725578
That is what's nice about PC gaming. Want higher resolutions? Sure. Prefer higher graphics settings? Sure. All personal preference.
>>
File: Capture.png (14 KB, 709x130) Image search: [Google]
Capture.png
14 KB, 709x130
What is this
>>
>>54722049
Great job faggot, a new product is faster than the old one!
>>
>>54725624
The Fury X rebrand to Fury 9.
>>
>>54722473
>295X2
>Pro Duo
Why is everyone ignoring their 2 best GPU'S?
>>
>>54725750
Because it is crossfire on a single card that can barely stand toe to toe with NVIDIA's top of the line.
>>
>>54725786
I knew that Pro Duo is 2 nanos in sli, but is the 295X2 also like that? Haven't actually looked much into that card
>>
>>54725894
>295X2
>X2
Simple stuff, Phil.
>>
>>54725894
Two gimped 290X in Crossfire. ASUS made a better version with air cooling, more power delivery, and two non-gimped 290Xs on a single PCB.
>>
File: 97938171.jpg (7 KB, 206x237) Image search: [Google]
97938171.jpg
7 KB, 206x237
>>54724636
>Third party Zotac 1080s are already listed for well under the $699 mark at roughly $600
Show me. Source please.
>>
>>54726067
It got taken down, but it was on a European retailer for like an hour. Check the blog sites.
It's a black plastic version of the "Founder's Edition" shroud.
>>
>>54724771
Its fake. Simple as that.
>>
File: 18514991.jpg (10 KB, 445x482) Image search: [Google]
18514991.jpg
10 KB, 445x482
>>54726086
>It got taken down,
Yeah... i thought so.
>>
>>54726114
Of course it got taken down. They're not supposed to be on sale yet. I guess all those 380X leaks on XFX's page last year were fake too last year?
Fucking AMDrone and logics don't mix.
>>
File: 44688280.jpg (7 KB, 252x240) Image search: [Google]
44688280.jpg
7 KB, 252x240
>>54726134
Guess youre the only one who seen this "leak" mate.
>>
>>54726154
Google it nigger ;^)
I'm not even supposed to be shilling, but this is too much fun
>>
>1070 doesn't even beat 1080
Is NoVideo even trying at this point?
>>
>>54726176
Nothing in google, mister shill. No leaks.
>>
>>54726200
:^)
You've been fun, but now I got to go
>>
Any news about the mobile cards?
>>
File: 50462012.png (93 KB, 216x241) Image search: [Google]
50462012.png
93 KB, 216x241
>>54726176
You prolly the same guy who lied about fake polaris price image. Hello there, getting defeated again.
>>
File: 1428011673834.png (256 KB, 390x470) Image search: [Google]
1428011673834.png
256 KB, 390x470
>>54726190
>>
>>54726226
Why do you niggers keep taking my baits?
>>
File: 1461985296337.gif (1 MB, 500x720) Image search: [Google]
1461985296337.gif
1 MB, 500x720
>>54726190
audible kek
>>
File: 05426331.jpg (23 KB, 374x353) Image search: [Google]
05426331.jpg
23 KB, 374x353
>>54726250
But i didnt? I browse currytech/videocardz so its pretty hard to bait me with just lies.
>>
AMD

IS

FINISHED

AND

BANKRUPT
>>
>>54726284
At least I got paid for it, that's all that matters
>>
>>54726301
I'm sort of jelly. How do i get a job like that?
>>
>>54724346
Please.
280x is already a rehash of of the 7970.
285 and henceforth 380x have inferior 256 bits memory interface.
My 7970 Ghz from asus is the best buy I ever made. 2nd best was radeon 9700.
>>
>>54726315
Look up internet or work from home marketing positions. Most of them are review shills for Amazon and whatnot, some are for forums and boards such as these.
No experience required, either.
>>
>>54724337
>Eh, most AMD users are Linux fags and can't use that anyways
What is MadVR
>>
>>54726063
I thought it was PowerColor? Devil 13 or something?
>>
>>54726453
Sorry, the ASUS card had a waterblock from ekwb, not aircooling.
>>
ive been on AMD for 10 years and this broke my devotion

im an nvidicuck now im sorry
>>
>>54722049

Titan X
Shading Units: 3072
TMUs: 192
ROPs: 96
SMM Count: 24
Pixel Rate: 96.0 GPixel/s
Texture Rate: 192.0 GTexel/s
Floating-point performance: 6,144 GFLOPS
Memory Size: 12288 MB
Memory Type: GDDR5
Memory Bus: 384 bit
Bandwidth: 337 GB/s

GTX 980Ti
Shading Units: 2816
TMUs: 176
ROPs: 96
SMM Count: 22
Pixel Rate: 96.0 GPixel/s
Texture Rate: 176.0 GTexel/s
Floating-point performance: 5,632 GFLOPS
Memory Size: 6144 MB
Memory Type: GDDR5
Memory Bus: 384 bit
Bandwidth: 337 GB/s

1070
Shading Units: 1920
TMUs: 120
ROPs: 64
SM Count: 15
Pixel Rate: 96.4 GPixel/s
Texture Rate: 180.7 GTexel/s
Floating-point performance: 5,783 GFLOPS
Memory Size: 8192 MB
Memory Type: GDDR5
Memory Bus: 256 bit
Bandwidth: 256.0 GB/s

Aside from the crippled memory bandwidth (but muh 4th Gen delta color compression), capability wise it's roughly around 980Ti levels
>>
>>54726329
The Tonga series was intentionally gimped in the memory bus to make it fit into a certain budget. It had so many other improvements over GCN 1.1 that it would have been stepping on Hawaii's toes if they made it any more powerful.

>>54724337
Fucking retard. Continue to be ignorant or just keep up the great work trolling the shit out of me and whoever you were replying to.
>>
>>54724326
>Doesn't really matter, it doesn't have enough juice to go past 1080p anyway.
No shit, that's why it's a $200 card. Has nothing to do with the VRAM amount though, so thanks for agreeing with me I guess.
>>
>>54726465
Right I remember that. The Devil 13 is a separate thing, custom PCB by PowerColor with two fully fledged 290x or 390x on board. The thing weights 7 pounds and comes with a prop to hold it up in your case lel.
>>
>>54726524
so the 1070 is a 980 refresh? the gtx980 has 2048 cuda cores and the gtx1070 only has 1920 but has better drivers probably
>>
http://www.geforce.com/hardware/10series/geforce-gtx-1070

THANK YOU BASED NVIDIA
>>
>>54726723
They just moved everything 1 tier For example:
1080 = 980Ti
1070 = 980
1060 = 970.
>>
>>54726730
das sum fancy text n graphics n numbers
>>
>>54726723
It has updated drivers and 3.5 GB more RAM than a 980.
>>
File: 1358740615013.png (7 KB, 400x400) Image search: [Google]
1358740615013.png
7 KB, 400x400
>>54726730
Until multiple independent reviewers confirm that the 1070 has the full bandwidth across all 8gb I'm not even a little bit interested

I'm not going to get 3.5 meme'd again
>>
Huh, Polaris is doing way better than i expected a "mid range" card to do. Wonder what the 480 will retail at.
>>
File: 1454475190674.png (230 KB, 640x360) Image search: [Google]
1454475190674.png
230 KB, 640x360
>>54726861
>FULL 256BIT MEMORY BUS
Oh wow, this is supposed to be impressive?
>>
http://www.anandtech.com/show/10336/nvidia-posts-full-geforce-gtx-1070-specs

>The GTX 1070 is not shipping with any ROPs or memory controller channels disabled – GTX 970 style or otherwise – and as a result it retains GP104’s full 64 ROP backend.

FULL 256BIT MEMORY BUS

STAY MAD, AMDPOORFAGS AS YOUR SHITTY MEME CAN'T BE USED ANYMORE
>>
>>54722049
Wasn't this obvious?

I mean every man on this planet who knows SOMETHING about GPUs should have expected that the 1070 is ~15-20% slower than the 1080.
SOO Titan X/980TI
>>
>>54726883
8GB MEANS 8GB BASED NVIDIA DOES IT AGAIN
>>
>>54722943
>>54722881
>>54722049
>>54722117
I wonder if the CPU (by extension higher end ram) plays an important part in this benchmark.

1070 was benchmarked with the latest 6700K
480/X was benchmarked with the older 4770

These benchmark shouldn't be conclusive at this moment, but we should see maybe a tier or two difference with final benchmarks.
>>
>>54726894
AMDumbs were crowing about how the 1070 was going to be slower than a fury x which is the main thing this is a win over
>>
File: newking11.jpg (149 KB, 1016x849) Image search: [Google]
newking11.jpg
149 KB, 1016x849
>>54726894
Except 1080 is 980Ti levels.
>>
>>54726949
Well its equivalent to reference 980ti.

kek, I wonder if the 1070 OC as well as 980ti. Then the real results will shows. We all saw the other benchmarks regarding OC Titan X/980ti vs 1080.
>>
It's basically over at this point. Nvidia is finished.
>>
Quick question, trying to learn more about gpus.

Why is it not based upon price per FPS instead of a benchmark test? Or are the two close enough related they're equal?

For instance, even though nvidia seems to get higher benchmark scores, is the price difference between AMD also equivalent to more real world measurements, like fps in games and rendering speed while doing something like video editing? Are those kinds of numbers public somewhere?
>>
>>54726982
>equivalent
It's faster, probably even faster in real terms considering the firestrike extreme and performance scores
>>
>>54727028
There is no real numbers, only speculations.
>>
>>54722117

Fuck, I might as well get a Nano or keep my 380.
Thread replies: 255
Thread images: 39

banner
banner
[Boards: 3 / a / aco / adv / an / asp / b / biz / c / cgl / ck / cm / co / d / diy / e / fa / fit / g / gd / gif / h / hc / his / hm / hr / i / ic / int / jp / k / lgbt / lit / m / mlp / mu / n / news / o / out / p / po / pol / qa / r / r9k / s / s4s / sci / soc / sp / t / tg / toy / trash / trv / tv / u / v / vg / vp / vr / w / wg / wsg / wsr / x / y] [Home]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.
If a post contains personal/copyrighted/illegal content you can contact me at [email protected] with that post and thread number and it will be removed as soon as possible.
DMCA Content Takedown via dmca.com
All images are hosted on imgur.com, send takedown notices to them.
This is a 4chan archive - all of the content originated from them. If you need IP information for a Poster - you need to contact them. This website shows only archived content.