https://blogs.nvidia.com/blog/2016/ 07/21/titan-x/

>>55692957
AMD ON SUICIDE WATCH

File: xd.gif (230KB, 165x115px) Image search: [Google]

230KB, 165x115px

>TITAN X will be available Aug. 2 for $1,200 direct from nvidia.com in North America and Europe, and select system builders. It is coming soon to Asia.
>$1,200
>Did we go too far?

File: 2 jew 4 u.png (359KB, 538x385px) Image search: [Google]

359KB, 538x385px

>>55692957
>$1200

>>55692957
>1200
1080ti is probably going to be 800. Ill stay with my 1080 until the 1280ti or the equivalent.

AMDPOORS JELLY ON SUICIDE WATCH

AMD IN FLAMES

>>55692957
>$1200

Wow, another absurdly high end card for the small handful of idiots out there who actually are willing to drop more money on a single card than most people spend on their whole rig.

File: 1453168371408.png (18KB, 448x434px) Image search: [Google]

18KB, 448x434px

>>55693031
>Being a waitfag
>Not upgrading to every architecture's Titan

File: remi2.jpg (58KB, 244x269px) Image search: [Google]

58KB, 244x269px

>>55693046
>being poor

>>55692957
>paying $1200 for a meme card when in 2 years the 1170 will outperform it for $350

File: 1436035676956.gif (2MB, 500x500px) Image search: [Google]

2MB, 500x500px

>>55692957
>$1200

Unless it can outperform two (2) 1080s I don't want to bother

>>55693031
Are you so poor you can't buy the best card?
LMAO at your life

>>55693048
>not buying the x80 Ti for 66% of the price and 95% of the performance

>>55693062
>using a picture of the smiley with a carat nose

>>55693062
It's going to have at least 150% of the performance of a 1080 according to some guy that knows his shit

File: 1458447846571.webm (230KB, 538x300px) Image search: [Google]

230KB, 538x300px

rx490 when?

>>55693096
Not until Q4. AMD is kill. Whatever they put out will immediately be followed by the 1080ti which will btfo every offering.

File: lolpng.png (149KB, 1120x977px) Image search: [Google]

149KB, 1120x977px

>Vega is DOA

>>55693096
Never, the 480 was the biggest polaris chip. Wait vega

>>55693096
Doesn't even matter.
It'll go something like this
Titan X in August
490 sometime at the end of the year
1080 Ti immediately following the 490

>>55693106
>AMD is kill

That may be so, but you're entirely delusional if you think a $1200 card has anything to do with it. Around 0.00001% of people will be buying a Titan X.

Why did they put SLI on this thing?

I mean who the hell is gonna spend 2,500$ on GPU power unless they do heavy 3d imaging..

>12 GB of GDDR5X memory (480 GB/s)

AHAHAHAHA

>STILL NO HBM2

File: 1451577237987.png (16KB, 362x324px) Image search: [Google]

16KB, 362x324px

>>55693180
>implying HBM2 is even good

File: fag.png (56KB, 1225x412px) Image search: [Google]

56KB, 1225x412px

>>55693165
Richfags who buy hardware just to jerk off to benchmarks, rather than play childrens' gaymes.

>>55693189
>>>/v/

File: amd-radeon-r9-fury-x-4-go_9cc2eb05b1c90291.jpg (249KB, 2025x1440px) Image search: [Google]

249KB, 2025x1440px

>>55693180
>HBMeme

>>55693165
>unless they do heavy 3d imaging..
Well there you go. These are for people who have actual important things to do with them, not just play on their computer.

>>55692957
Not bad, but the story is kind of retarded. 1080 is pushing 9 teraflops, why wouldn't the titan push at least 11 teraflops?

>>55693214
#RadeonRebellion

File: HBM.png (649KB, 1631x1571px) Image search: [Google]

649KB, 1631x1571px

>>55693214

File: 1458892321084.png (85KB, 361x333px) Image search: [Google]

85KB, 361x333px

>>55693214
>pump sounds
>coilwhine
>housefires
>underperforming

>>55692957
>$1200
Jesus Fucking Christ.
>GDDR5X
Fuck. They really put HBM2 only on their enterprise gear. That's a shame.

>>55692957
>GDDR5X
wew

It's just a cash grab until they release the 1080ti in a couple months. Just like last time. People will still buy it, regardless.

File: 1467921300729.jpg (29KB, 255x244px) Image search: [Google]

29KB, 255x244px

>>55693165
>puts SLI on a titan
>won't put SLI on a 1060
OY VEY

>>55693226
end this meme
The only reason people considered the Titan a workstation card is because the original titan had excellent double precision floating point (1:3 TCC mode vs the 780 Ti's 1:24) ; something you really only do with workstation cards.

Titan X didn't have that. Likely, this card won't either. It compromises their cornerstone of the market with the Quadro line.

File: perfrel_3840.gif (48KB, 400x751px) Image search: [Google]

48KB, 400x751px

>>55693214
>implying HBM1 is the same as HBM2
>implying the Fury X was bad
>implying it doesn't match the 980 Ti at 4K despite 2GB less VRAM
>implying it doesn't almost match the Titan X at 4K despite 8GB less VRAM
>implying that isn't because of HBM
>implying you're not tech-illiterate spillover from /v/

>>55693276
now be a good goi and spend your life savings on a card that will be replaced within a year.

>>55693286
look the 1070... HBM2 or 1 is useless, just use GDDR5x with a bigger bus

>>55693287
There are people who wipe their ass with that amount of money and they're not pajeets.

1080 ti whennnnn

I don't want to wait a year to upgrade my build

>>55693309
>bruteforcing is the way to go
Were you involved on the Pentium4 design, by any chance?

>>55693069
more like 120% the performance with aib partner cards

reminder that AMD's top line cards are going to blow both the 1080 ti and this dumb thing completely out of the water for less money

wait for the 490 if you have even two brain cells to knock together

Well fuck, I didn't think big pascal was coming until the end of the year.

I bought a new guitar with the money I was going to use with a gpu.

Time to offload some stocks I guess

>>55693058
But then you can't write
>Being poor
On 9gag

File: madbulance.png (24KB, 695x567px) Image search: [Google]

24KB, 695x567px

>>55693337
>wait for it

>>55693320
God, the GTX 1080 use GDDR5x@256bit a destroy all card at high res like 4K. Memory is overrated

File: 1274581975691.gif (3MB, 445x247px) Image search: [Google]

3MB, 445x247px

>>55692957
>GDDR5X

>>55693356
>I don't know how GPUs and hardware in general works: the post

I'm not even poor and I'm also on my computer for hours every single day and probably consider it my most valuable possession but I still don't get how people justify spending that much on a single GPU.

>>55693382
It's a status symbol. No different to buying a fancy car or such things. The kind of person who'd buy a Titan X has more money than they know what to do with. Some epeen points are worth the investment to them.

>>55693380
The GTX 1080 don't have memory bottleneck, If Nvidia can save money, why they need put HBM2?

>>55692957
>US$1200
That means it will be close to AUD$3000

And then in 6 months time the 1080Ti will launch at less than half the price.

File: perfrel_3840_2160.png (39KB, 500x1050px) Image search: [Google]

39KB, 500x1050px

>>55693380
But he is 100% right, can you elaborate pls?

i think it's gonna be faster than 1080 35-40 %

>>55693430
>>55693410
Just because something performs well without it, doesn't mean it can't be even better. Not having HBM effectively makes the 1080 worse than it would be with it, because it is objectively better than GDDR5X, even if GDDR5X is fine as it is. Saying otherwise is asinine.

>>55693437
lol no they already say on the site it's 40% faster than the titan x so ~40% faster than the 1070, 20% faster than 1080

>>55693096

RX490 is most likely a Polaris 10 with HBM memory. That's it. Not a dual GPU, or Vega or a Polaris 9/12.

>>55693452
>fury x has hbm
>still gets btfo by 980ti

>>55693469
>in APIs from 2009

>>55693452
HBM is a gimmick

>>55693481
retard

>>55693486
you mean like gddr5x?

hbm is at least power efficient unlike gddr5, only reason nvidia have low thermals is because the last time they tried having hardware async the housefire that is fermi was created, so they gimped their hardware.

why do you think gtx 1060 pcbs look like a $100 amd gpu?

>>55693499
Nice comeback, nvidiot

>>55693499
nice meme /v/

i remember the jump from 680 to titan are almost 40 %

>>55693469
>still gets btfo by 980ti
Is that so? Then explain >>55693286 friend. Memory bandwidth matters more the higher you push resolutions, and the Fury X matches a 980 Ti at 4K, despite getting BTFO at 1080p. Explain your reasoning for this if HBM and memory bandwidth are a meme.

>>55693503
>>55693506
>>55693516
dumb amdrones

File: hbmp.png (21KB, 621x371px) Image search: [Google]

21KB, 621x371px

>>55693452
>power efficient
it's a gimmick

>>55692957
>and then the 1080ti blows it out the water
history repeating itself

File: laughing baboon.jpg (39KB, 372x341px) Image search: [Google]

39KB, 372x341px

Don't worry guys, as soon as the first Nvidia card with HBM comes out, the Nvidiots will be willing and able to acknowledge its benefits. It's only brand wars that render it a "gimmick" and a "meme" for now.

>>55693535
Why do you even have a 2TB HDD? 50GB HDDs are enough, you can fit Windows and Office on it, why would you ever want more? That's just dumb

>>55693503
HBM2 is fucking expensive and useless right now

>>55693214
Nvidiots before announcement:
>Just wait untill we crush your pathetic gddr5 chip with our superior HBM2 chip
>HBM is the future
>AAYYYYMMMMDD poorfags can't afford HBM2.

Now:
>B-b-but HBM is useless
>HBMeme
>Who needs HBM anyways when we have GDDR5X

>>55693554
what are you, poor?

File: 1460063672167.gif (897KB, 800x430px) Image search: [Google]

897KB, 800x430px

>>55693503
>only reason nvidia have low thermals is because the last time they tried having hardware async

>>55693556
kys

>>55693567
sorry not thermals, tdp

>>55693551
>power efficient
Like I said, save 15w is useless, agian HBM2 in 2016 is a gimmick

File: 1461273313117.png (1MB, 909x960px) Image search: [Google]

1MB, 909x960px

>>55693556
>GP100 = GP104

Nvidia are taunting amd with all their releases. It's basically a "come at me" directed towards amd since the pressure is on amd to try and beat these new nvidia cards.

>>55693653
It's more like beating a dead horse.

>>55693602
>GP100 relevant to any of you manchild neckbeards playing videogames
True Pascal is that, the rest is scraps. Volta is the thing you niggas want.

>>55692957
gp102

U wot

File: DGX1Parts.jpg (236KB, 2232x840px) Image search: [Google]

236KB, 2232x840px

>>55693748
The GP100 is not the same market

>>55693777
Did anyone else get a massive erection? Holy shit all that power.

>and it still won't be able to do 4K on ultra

>2 august

meanwhile amd not able to release custom 480 after a month. I wanted to buy a 480 but fuck amd not gonna wait another month

>new pascal gpu every month since June

What the fuck they used to wait at least 6 months.

They're probably finalizing Volta now from the looks of it.

File: anigif_original-grid-image-30208-1385149291-3.gif (805KB, 245x281px) Image search: [Google]

805KB, 245x281px

>2 1060 if they kept the sli bridge would destroy a 1080 at 1/6 the price of a 1080
>if you could set 3 1060 in sli they'd demolish the Titan for 1/2 the price

AHAHAHAHAHAA WELL PLAYED NVIDIA
REALLY TRYING TO SECURE THOSE SHEKELS NOW?

>>55693927
2x 960 destroys a 980 too and cost less. This applies to every single *60 series. No exceptions.

Only retards go for 2 weaker cards over a single strong one.

>>55693144
August 2nd.
Am I the only one with the impression that the usual time tables have been shortened? The only leak I saw for the GP 102 is not even 1 month old yet, from those shipping manifests.

We might see Ti cards by October at this pace.

>>55694027
correcting: gp102 for ti's and gp 100 for titans.
(that manifest had them all)

>>55693748

File: _81762508_risitaspromo.png (417KB, 640x360px) Image search: [Google]

417KB, 640x360px

>>55693587
Knowing nVidia, using GDDR5X gives them way better profit margins. They know their audience trusts whatever they do to be optimal and will gladly pay up without second thoughts.

File: nyaruko.png (261KB, 500x500px) Image search: [Google]

261KB, 500x500px

>>55693927
>not going for SLI Titans
Are you poor?

>>55693016
They need to clear the checks for all of the 10xx series shilling.

>>55694084
Why AMD can't make similar card and sell it for $1k+?

11 TFLOPS

>AMD on buttfragged watch

They're literally trying to climb a shit rope.

>>55693556
They've got a big shilling army. It's hard to organize, they spend a lot of non-parallel cuda cores on that task 24/7.

If on one hand they get a good flexibility and display good on the fly first responders memetic reconstructions, it's also hard to coordinate and suffer from a bit of overall inconsistency, on the other.

>1080 keks on suicide watch

>>55692974
LOL AMD IS DEAD BURIED CREMATED.

File: 1468885943207.jpg (143KB, 1024x1024px) Image search: [Google]

143KB, 1024x1024px

>>55694109
Do you really want the prices set on a trend to go in that direction?

I know you just playing for the awesome bantz, but that's really pathetic.

>>55692957
gdi i just found the gigabyte gtx 1080 xtreme in stock. Anyone want it at retail price?

kek, AMD is basically dead now. this card is going to be 40-50% faster than a maxlel titan x and AMD won't even be releasing a response to it until q1 of next year.

we're in for a wild ride, nvidia only ever released 780ti and 980ti when AMD released a competitive product, so the $1200 titan x is here to stay.

>>55693046
>finally a single card solution for 4K
>too expensive
enjoy your "RX 480 optimized for 1080p gaming"

>>55694142
Fury X was 8.6 TFLOPS this time last year. Titan X was 7 TFLOPS.

Don't even try to imply NVIDIA will ever win in raw compute power.

>>55694227
>Don't even try to imply NVIDIA will ever win in raw compute power.

they don't need to, AMD doesn't even compete with NVIDIA in the GPGPU space. the socketed variant of P100 has been uncontested for over 3 months now...

>>55694198
god dammit, amada is cute as fuck

>>55693466
Polaris with HBM would mean a complete new die, you can't tackle HBM to a die like you cam tackle GDDR to a PCB (provided the memory controller was able to manage it)
The RX490 is either Vega, or a dual GPU card, which is almost impossible considering they could had launched that not long after the RX480

>>55693227
It's listed as 11 TFLOPS at 1531 MHz.
http://www.geforce.com/hardware/10series/titan-x-pascal

>>55694239
She better be. Her body is the only asset amd can currently sell.

>>55692957
>$1200
>no HBM2

NVIDIA FAGS BURRIED AND DEAD. DONT CALL SUICIDE WATCH. ITS TOO LATE.

It better fucking deliver 120fps on 1440p and 60fps on UHD, if not, then it's waiting for the next gen again. Pascal has been a huge disappointment, it would have been fantastic if they were at least close to MSRP.

>>55693287
>$1200
>life savings
you should have paid attention in school poorbro

>>55694194
Why would you cremate a buried guy?
Cremate that shit before burial moron

>GTX 1080
>2560 Cores

>Titan X Pascal
>3584 Cores

Brutal

>>55693550

what benefits? HBM is unwieldy as fuck, has shit yields because of the fact that you need twice the dram chips, can't have a single bad TSV, etc. also requires an extra assembly and test step, completely redesigned PCB and the like.

meanwhile gddr5x has the exact same pinout as normal gddr5 and be a drop in replacement for gddr5.

the only real advantage i can see for HBM is that you get a much denser package, so you can build a much smaller and simpler PCB (because all the traces for the memory bus are now gone), which is great for SFF and water cooling but pretty meh for anything else.

>>55694273

it has 4 of the SMs disabled like P100 as well. yields are probably still shit for huge 600mm2 dies, 16FF has only been in mass production for about 6 months now whereas with 28nm nvidia and amd waited 2 years to being making huge dies.

File: 1468217862861.gif (215KB, 400x524px) Image search: [Google]

215KB, 400x524px

Where's the reply, AMD, WHERE'S YOUR NEXT HIGH END CARD, AMD

It keeps getting harder and harder to wait

>>55693056
Never met rich person that was willing to throw their money away on a $1200 video card, they either worked to hard for it or received an education that prevents them from making stupid decisions.

>>55694273
should have be 4096 cores

>>55693380
>ridicule new GPU for not using HBM 2
>new GPU has 50% more memory bandwidth than best GPU on market
>implying this is somehow insufficient

>>55692957
>No HBM2 and still over $1k
Looks like I'll be giving my money to AMD this year kikes

File: satensmug2.jpg (157KB, 560x570px) Image search: [Google]

157KB, 560x570px

>>55694299
>Just Wait™
This should be amd new motto.

>>55693380

do you? because everyone with even a little bit of knowledge knows the extra bandwidth is useless for gaymen without enough fillrate/rop throughput to saturate it. it's why cards like the fury x sucked so bad.

>>55693554
and in extremely tight supply (which leads to expensive, which leads to not worth it)

>>55693056
Ken
Some of us have $4000 a month in expendable income and still aren't wiling to shell out that much on a card without HBM2

File: 75fbf6c4fb.gif (191KB, 500x375px) Image search: [Google]

191KB, 500x375px

>>55694338
well, it was supposed to be 200% more bandwidth.
Or whatever stock freq HBM2 has * 4096 bit bus amounts to.

You can't say that you never wanted it in full honesty just to win a memetic argument over the internet.

$1200.... fucking lol

>>55694374
1.2k should cover that cost aplenty.
The fud over hbm2 supplies must've been confirmed though. If vega is depending on it to launch to market... the wait will intensify by a lot.

>$1200

>>55694387
>well, it was supposed to be 200% more bandwidth.

only if you use 8 high stacks. hbm2 with 4 high stacks (which nvidia is already using for p100) is the same bandwidth as hbm1 but each stack is 4GB instead of 1GB.

disable one of the stacks leaves you with only 384GB/s bandwidth, less than the 480GB/s that gddr5x gives you.

>>55694378
gib monis or i repor u

You guise should see the Titan X (invisible #2) Black that will be launching 6 hours after the 10xxTi's. For just $1450.

>>55694347

>HBM2 (Just Wait™ for Vega)
>Vulkan
>DX12

>still slower than nVidia

>>55693587
Why would you even want an SSD, raid 1 with HDD isn't good enough?

>>55694387
Of course we want it, the question to ask is does it give sufficient benefit to be worth it. With the latest GDDR5x and 50% wider bus, I don't think the lack of HBM 2 is going to significantly hurt the new Titan X's performance.

NVidiots denouncing HBM 2. AMDrones saying the Titan X is a failure/joke because it doesn't have HBM 2. Both sides are idiots.

>>55694347
buy 4x480 and burn your cuckshed

>>55694430
your post was one of the most moderated ones, and my criticism may have been a bit too harsh. I should've quoted a different one.

But aren't hbm2 binned for higher freqs as well? I thought we would be seeing the likes of 1tb/s bandwidth whence they started to ship.

Fury hbm1 is 500mhz, iirc.

File: Mrbeale.jpg (7KB, 300x168px) Image search: [Google]

7KB, 300x168px

>tfw new Titan X FE for 2000usd.
>1200 is AIB's msrp

>>55694485
>But aren't hbm2 binned for higher freqs as well? I thought we would be seeing the likes of 1tb/s bandwidth whence they started to ship.

you can clock it higher but you would need perfect chips and it would ruin the power efficiency. hbm2 will probably be clocked at 500mhz like hbm1 was.

>>55694438
Or you could get a job.
Looks like I'll be buying a SCAR 17 to hold me over

>>55694536
Titans are only ever made as reference model
That $1200 price IS for the FE

File: 01-PCB_w_600.jpg (67KB, 600x453px) Image search: [Google]

67KB, 600x453px

>>55694593
Boggles me why they still went with a blower with all the throttling Maxwell Titan X had. Maybe this one won't be such a housefire but with how the 1080 FE had issues I fucking doubt it.

>>55694462
Kek no, I'll be buying a i7-6700k and single rx 480 to hold me over till LGA-2066 comes out next summer and I'll be doing an mATX build in a white NZXT Manta case with Skylake-X and dual GPU setup with, at this rate, a pair of Vega GPUs for the simple fact that nVidia are being Korean kikes, and water cooling the entire cereal box with a pretty custom loop because I'll have $4-5k eating a hole in my PC budget by then. At the moment, I'll likely be blowing some cash on a new gun safe and rifle

Havent followed gpu stuff for a while, will the 1080ti be a 4k proof card? How is the 1080 doing? How is AMD doing?

>>55694621

shouldn't be an issue unless you use the retardedly conservative stock fan profile. any vapor chamber blowrer cooler should be enough for a 250w card, it will just be noisy.

>>55694621
EVGA made a waterblock version of it the Titan X as well as an ACX cooler for it.

>>55694648
>it will just be noisy.

Well that's the problem. You spend $1200 you'd think you could get something that doesn't have that issue. Especially if it's marketed more toward more professionals.

>>55694621
If you spend $1200 on a GPU and don't water cool it you shouldn't be wasting the money one it.
Most enterprise server racks I've fucked with had custom loop water cooling solutions applied as AIO units due to space constraints for airflow to the GPU drawers

naming convention should've been
Titan X
Titan XX
Titan Black XXX

>>55694657
>spending the money on a Titan X and not upgrading the cooler
maybe you can't really afford a Titan X

>>55694657
>Especially if it's marketed more toward more professionals.

blower coolers are the universal standard for 'professional' cards. only gaymer stuff has good coolers.

>>55694670
>Titan X
>Titan XX
>Titan Blacked
Ftfy

File: amd2.jpg (52KB, 568x612px) Image search: [Google]

52KB, 568x612px

>>55694639
>i7-6700k
>rx 480
>next summer a pair of Vega GPUs

>>55694669
>and don't water cool it you shouldn't be wasting the money one it.

Exactly my point. Why offer the blower solution? Should have gone Fury X style.

>>55694670
>not Titan BBC

>>55694683
>blower coolers are the universal standard for 'professional' cards.

Shouldn't be for cutting edge hybrids like this. Nvidia's blower coolers for the Titans are more gaming oriented as well. Not like their Quadros. These are meant for a specific professional market where I'd argue that custom cooling isn't as common.

>>55694688
>Exactly my point. Why offer the blower solution? Should have gone Fury X style.

would be better if it was a full cover block instead of the junky AIO that AMD used.

>>55694683
>professionals

aren't there Quadros at lower price points than Titans?

What's the benefit on chosing Titans over Quadros? Clock speeds (and a driver that'll allow you to play vidya during your down time)?

I'm not bashing this segment. I understand the folks that go for this for e-peen alone and i'd do the same if had the spare change.

>>55694751
If you're doing a custom loop you can get your own block on it so it's besides the point.

>>55694687
I have expendable income but I'm not stupid. I'm not spending >$300 on a GPU I'll be replacing in 6 months or less, and same for a CPU that I'll be replacing in under a year, when an entire new socket, chipset, and series of CPUs will be available.
Same reason I'm not replacing my 2005 GTi, it still runs fine, my wife has a 2013 jeep she's happy with and I'm paying off her student loans while saving for a down payment on a house and for a CNC mill and lathe. Why would I throw $700 or $1200 on a GPU that I'll be tossing to a needy Ohio aspiring tech kid in a few months. That would be like those idiots who bought 980 Ti cards for $600+ a month before the 1080 was released.
Damn nigga
>>55694688
Seriously, I have no clue, nVidia is proving to be extra jewy this year with the drastically increased release prices from the last gen

>>55694027
Probably increased/timelier competition this time around. Perhaps Nvidia is trying to dominate the news and portray AMD as a has-been that can't even compete in the same league.

>>55692957
I haven't followed much of the GPU news, especially the GP100. Seeing as the 10** series has had awful stock, how many dies will they even have given that they're going to be monstrously big?

Any real specs?
did they made a real compute card this time or is it gaming card for 1200 again?

>>55694846
only directly from nvidia and select vendors => no stock they paper launch and paper re launch it

>>55694846
Yields gonna be atrociously low, undeniably. But at this price point doubt it'll be an issue. Maybe some constraints during launch week, but nothing like the 10xx cards (which is debatable if they even have supply problems, restocking seems to be coming fairly low but the demand is also unarguably high).

>>55694027
>We might see Ti cards by October at this pace.

Pretty much will go like this

Vega matches 1080 performance? Adjust prices. Release GP102 as 1180 or whatever they call it later in year. Basically Kepler refresh again.

Vega beats 1080? Release 1080 Ti as performance king again.

Vega beats GP102? Well that's probably not going to happen.

>>55694878
>We packed the most raw horsepower we possibly could into this GPU. Driven by 3584 NVIDIA CUDA® cores running at 1.5GHz, TITAN X packs 11 TFLOPs of brute force. Plus it’s armed with 12 GB of GDDR5X memory – one of the fastest memory technologies in the world.

>Up to 60% faster performance than previous TITAN X

>12B Transistors
>GP102 will feature 40% more CUDA cores than the GP104
>Memory Bandwidth 480 GB/s
>250 Walts. 1 8-pin and 1 6-pin
>Price $1,200 USD


>The card will be available on August 2nd in the US and Europe.

quoting from OCN OP.

>>55694894
>demand is also unarguably high
Oh, I've seen plenty argue (more like whine/throw tantrum) the lack of stock isn't due to demand (I'll let you guess who)

>>55693077
Wasnt 1080 supposed to be like 4x 900s and it turned out it was just VR performance?

>>55694903
If vega does manage to they'll just release another Titan Z

>>55694906
hm, lower clock but more cores
if it's exactly like smaller pascal it is going to be terrible at video games, meaning it is probably not bad at compute

>>55694936
>Titan Z

Oh god, that failure? They'd get tromped by refusing to go the AIO route like AMD does. And you know they're going to make a dual Vega card for sure.

>>55693925
Pascal wasn't in the roadmap at all, they been working on Volta for at least 4 years now

>>55694936
they suck at it, amd sucks at it but nvidia is even worse at dual gpus

>>55694931
Yes, 3x 980s to be accurate in VR
Basically it's more nVidia marketing that idiots will pander to.
I'm waiting to see actual performance and price between vega and this.
If nVidia wins for both I'll get 2x Ti GPUs hands down, if they keep up this shekeling price trend then I'll be buying 2x of vega. Especially if vega has HBM2 and anywhere near the performance of the 1080

>>55694757
>aren't there Quadros at lower price points than Titans?
There are, but most of them aren't even comparable to the *60's, they're quite slower
Quadros equivalents to Titan's are at least 5x more expensive
>What's the benefit on chosing Titans over Quadros? Clock speeds (and a driver that'll allow you to play vidya during your down time)?
On the first Titan you could get a Titan card that was identical in performance to a Quadro for a fraction of the price, since the using a Titan instead of a Quadro is pants on head retarded, you lose certfied drivers and DP performance for some vidya advantage

>>55694953
>>55694958
>8224 CUDA CORES!!!!!!!!
>14 TFLOPS
>WOOOOWWWWWW
This is the shit going on in the tech industry now.
This is like the tech version of the fucking Clinton foundation

>>55694961
Well, if 480 anything to go by AMD keeps close to last decade standards meaning: 200-250-300-400-500-600 price ranges. So 490x(or whatever they will call it) going to be 400.

>>55694953
>And you know they're going to make a dual Vega card for sure.
That's probably reserved for Navi

>>55694995
I'm talking Vega though, which will have HBM2, 490 will likely be either Polaris 9/12 or dual Polaris 10

>>55695012
no, vega 10 is 490-490x
vega 11 is new fury or ~495

no word on memory any of this will use though

>>55695012
490 won't be Polaris based, Polaris sole objective was to be as cheap as possible, a scaled up Polaris would be brutally limited by ROP's, and it wouldn't clock high enough to offset that like Polaris 10

>>55695005
I'm definitely curious as to wtf "scalable architecture" even is, doubt it's dual vega, but honestly the description is so vague it could be two dicks strapped to a leaf blower powered pinwheel

>>55694910
I'm an a AMDrone and wrote that post.
I think that the launch was a bit rushed (let's say a hard paper launch) AND the demand was a lot higher than they had expected.

If this situation go for longer than the launch wiindow, 4-6 weeks, it'll corroborate that they're facing supply issues. But I don't think that's the case, my theory is that they just pulled the release dates a couple weeks earlier than advisable for a healthy first day stock and they're simply refilling unexpected high amount of back orders now.

>>55695005
They wouldn't wait that long. They make a dual GPU for every gen.

>>55695020
I think you have those numbers backwards. 10 is the big card.

>>55695020
I didn't realize it was vega # I thought it was just vega. The more you know.

>>55695025
Agh I thought it was specifically Polaris 10/11 that was supposed to be cheap as shit and that other Polaris #s would be larger more pricey and better performers.
Makes sense that wag would be the performance sector of it tho

>>55695025
It's possible they'll make a dual Polaris card. I think some leaks came out that the 490 was listed at AMD with a manufacture number that they usually list their dual GPUs under.

>>55695046
that's with polaris 11 small 10 big
it's just order in what they developed it
vega has normal cycle 10 to 11

File: 726959187.jpg (138KB, 522x700px) Image search: [Google]

138KB, 522x700px

>>55694910
>>55695029
but I wrote lotsa of

p a p e r l a u n c h

a

p

e

r

l

a

u

n

c

h

just to make you guise mad and shit on your release day breds.

>>55695071
Paperlaunch or NDA up until the day it's in the stores? Guess which one is worst.

File: nvidia.jpg (920KB, 1920x1200px) Image search: [Google]

920KB, 1920x1200px

Why the fuck didn't they put the full GP100 chip on the Titan X and save the GP102 for the 1080ti?

Are they going to cut down the GP102 chip for the 1080ti?

>>55695100
Paper launch with overhype for capability.
AMD has been forthcoming with their plan for this gen of GPUs and it's capability being focused on low-mid range market and performance.
Pascal has been underwhelmingly the same and overwhelmingly overpriced by comparison.

>>55695012
490 as dual Polaris is Wccftech + a couple retarded youtubers nonsense.

They pulled this out of nowhere simply based on wild rumors (pushing the release to october vs delaying to 2017 q1). This theory came out of their collective asses.

>>55695020
>vega 1*
Why is this still being parroted? Where is the fucking sauce that there are two different Vega chips that aren't just undercuts?

>>55695104

GP102 is just GP100 with 1/32 FP64 instead of 1/3. it's still the same size and number of SMs.

>>55695121
>poorlaris was not overhyped

>>55695104
One - GP100 has more FP64 units per core than GP104 because it was designed purely for heavy compute tasks
Two - GP100 is massive for a fresh node shrink and won't get cheap or bountiful anytime soon
GP102 combines the core architecture of the GP104 with the size of the GP100 (without the FP64 units). There is no point in selling the GP100 cards to consumers who will complain about how weak it is for its size.

>>55695046
Polaris 10 is the Rx 480. The full chip.

>>55695104
>full GP100 chip on the Titan X
You want this shit to cost at least two thousand american shekels?

>>55695163
I still think that the overhype for the Rx 480 was carried out by Nvidia shills. Next level reverse shilling.

>>55695204
haha, wow, the paranoia runs so deep

>>55695167
That post was talking about Vega. Also all the leaks I've read has Vega 10 mentioned as the big card so not sure what that guy knows that I don't.

>>55695163
Polaris was proclaimed to be for 1080p and entry into VR to make VR affordable for the masses.
So far it's doing exactly what pajeet claimed.

>>55692957
A 1080 outperforms a Titan X by like double I really think they need to re-evaluate their pricing.

>>55695210
You want paranoid?
Start reading about Operation Gladio and COINTELPRO

>>55695212
RX 480 will get BTFO by 1060 in VR when SMP begins to be used.

>>55695210
I agree with the other anon. I was always saying it was going to be at least 390/970 and people got way out of hand. The latter part of the hype was funny, but prior to that was worrying.

>>55695174
Well the Maxwell Titan X was the full Maxwell chip wasn't it?

>>55693567
Thanks for the epilepsy

>>55695232
Do they not already have it enabled in their Funhouse demo? Doesn't seem to me like there's that big a difference between Maxwell and Pascal with it.

>>55695232
And then nobody will care, because fuck VR until it's 4k120Hz minimum with far better rendering and controls

File: 1463946919748.jpg (31KB, 601x508px) Image search: [Google]

31KB, 601x508px

>>55695204
>blame Nvidia for the polaris overhype

File: 1468909002441.jpg (40KB, 600x600px) Image search: [Google]

40KB, 600x600px

>>55695210
But it is a really good idea.

Let's try to shit on intels next mainstream launch:
people are already mad at them for negligible performance increase over generations, if we (AMDrones *wink*) fake some benches for 20-30% IPC increase and roll all out with it on saturation attacks on every forum and youtube channel out there; when they release another 3% mainstream broadwell (not E) like chip - there's gonna be riots.

>>55695271
>noVRfag detected
Why are noVRfags so butthurt about VR? On some level I understand the AMD v. nVidia bs (though I think it's irrational) but the VR hate is just asinine,

>>55695254
could you link a review/benchmark of maxwell v. pascal using VR Funhouse?

>>55695303
I don't have any benches. Bunch of people saying they were running it fine on 980 Ti's where the 1070 was recommended. Performance wise they're pretty much the same.

>>55695276
Didn't you know? Nvidia shills are responsible for all ills in the world.

AMD releases new chip? Nvidia shills hype the performance.
New AMD chip doesn't meet expectations? Nvidia shills denigrate its performance.

>>55695303
I'm just not interested in VR until it gives me flashbacks of falluja
The tech is nowhere near that point yet so I'm not interested. I'll rock the 4k120Hz displays that hit the stores next summer tho

>>55695303
pcper last podcast said that both the 1060 and the Rx480 run VR games just fine. They went ahead and said that Polaris didn't seem to be affected even on nvidia's funhouse.

File: 556777.jpg (114KB, 1280x720px) Image search: [Google]

114KB, 1280x720px

>>55695339
AMD TALK ABOUT POLARIS SINCE FEBRUARY

>>55695339
No son, they overhype it before the release.
After the fact that would be tantamount to giving them free advertising and shill man-hours.

>>55695339
Bullshit

>>55695386
>>55695364
>>55695378

I see sarcasm is a lost art.

>>55695353
>Polaris didn't seem to be affected even on nvidia's funhouse.

Wait, so can I play it on my 290x?

>>55695339
AMD shills:

Conspiracy theories.

File: 1468975007670.jpg (121KB, 850x846px) Image search: [Google]

121KB, 850x846px

I think I succesfully shifted goal posts.

No one cares about Titans anymore.

>>55695349
Seems like you have quite the closed mind; your loss, brother. Do you rock 4k120Hz displays now?

There would be no video game industry if consumers refused to play at 640x480 resolution or until computer graphics matched a certain acceptable level of realism. 8-bit Nintendo still has some of the best games of all time.

>>55695353
There hasn't been any proper benchmarking performed on these cards regarding VR yet. I know HardOCP is working on putting the various cards through a selection of VR games and am eager to see what their testing methodology is.

File: blablablabla.webm (251KB, 533x360px) Image search: [Google]

251KB, 533x360px

>>55695339

>>55695390
well, you're really bad at it.

>>55695439
Did the Nvidia jews poison your well or something?

File: e5qatOe.jpg.png (57KB, 864x576px) Image search: [Google]

57KB, 864x576px

>>55693106
Won't really be an issue, desu

The 1070 and 1080 are overkill for most games on the market right now, and more expensive than most people are willing to pay, while the RX 480's are beating nvidia out at the lower 970 / 1060 / 980 tier pricing and performance/dollar. On top of nvidia having bigger supply issues for the 10xx series than AMD is.

This isn't going to change much between now and December - once supply stabilizes for both AMD and Nvidia, AMD will start price cutting below what nvidia can match without a loss on each sale. Once we hit around Oct-Dec the RX490 will be released, competing with the 1070 and 1080 at a lower price point again, possibly even competing with the 1080ti, but I expect that'll happen with the RX 495 instead.

Everything above the 480/1060 is going to be overkill for all current games, and there's nothing particularly great coming out anytime soon to drive sales of the 1080's and 490's, so the higher end cards will languish for a good year or more, even if nvidia maintains an advantage, which is unlikely, particularly since everything new is going to be vulkan/dx12, and more of the older but still supported games will get vulkan updates, further incentivizing an AMD purchase.

I'd honestly skip this gen for nvidia: They don't have the vulkan/dx12 API/hardware figured out yet, and AMD has insanely good yields combined with very good performance that is guaranteed to improve over time with driver improvements and greater adoption of the low level API's - and all for cheaper prices than Nvidia's offerings.

>>55695426
are there benching tools for VR games? Apart from VR ready checkers.
I didn't watch many VR reviews, but all of them were only subjective experiences. And pretty basic, like yeah, this card gave me motion sickness that one didn't. Nothing too substancial.

>>55695457
>AMD has insanely good yields
citation, please

AMDEAD

File: 480 best.png (56KB, 668x384px) Image search: [Google]

56KB, 668x384px

>>55695457
4GB 480 flashed to 8GB master race

File: dr-evil.jpg (34KB, 355x397px) Image search: [Google]

34KB, 355x397px

>>55695447
What does nvidia or amd has to do with you being bad at sarcasm?

Who's being paranoid now?

>>55695466
Yeah, there's no real tests yet.

SMP supposedly gives almost double performance as the second eye's view is """free""". I'm very interested in learning if that's true, because that would be a massive performance advantage for nVidia.

The VR funhouse demo is extremely physics heavy and from what I've read the settings affect the physics more than anything else.

>>55695480
>AMD so desperate for marketing momentum they give away 4GB of RAM for free
>they do it for free
enough said

>>55695519
No doubt in my mind the 480 will drop to $200 sooner than people think considering they could afford to do it.

>>55695469
I'm not him but since the Polaris chips are quite small they probably have very good yields. Some people gonna exaggerate and claim this as a fact.

I'd say that since no one in the media managed to dig some fud for this, than it's probably true.
>>55695339


(see anon, that's how you sarcasm)

>>55695480
This is an important point, because the very fact that the 4GB 480's are actually just 8GB 480's with disabled memory very strongly hints that AMD can undercut Nvidia by at least $50 and still make a profit.

Within the year it's going to be $200 8GB RX480's vs $250+ 3GB 1060's. That's assuming Nvidia manages to get enough production to even meet demand. It's entirely possible the 1060's stay at $300+ for the rest of the year.

>>55695519
It's more along the lines that the 480 was never to have 4gb except on paper and the manufacturers merely flashed some of them.

Oldest trick in the book.

>>55695426
I know VR is going to be out there, it's just not my personal interest until it reaches a certain goalpost that fits my own concept of what is playable and enjoyable.
Will VR develop to that point? Yes.
Will VR carry on without my participation until then and succeed? Yes.
Will I spend for the hardware to use VR instead of traditional monitors before that point or even likely after then?
No and likely not.

File: kaiji.jpg (140KB, 1280x720px) Image search: [Google]

140KB, 1280x720px

>>55693653
Please stop this, can the underdog not win once

>>55695530
It'll happen after the AIB's get their money's worth at the higher launch prices for a couple months.

I'd be willing to bet a price drop happens at least by November, before all the christmas sales and deals start. AMD is going to kill it this winter, especially if Nvidia can't get their supply issues under control.

>>55695530
>considering they could afford it
but can they? There are virtually no 4GB models sold and they could be selling the card at or below cost for all we know just to have that $200 MSRP. Anyone with half a brain knew the 4GB card wasn't a real product, just a marketing ploy.

I think the RX 480 is a very low margin part and the 4GB is even worse than the 8GB. AMD cannot afford to give up margin by slashing prices (which also makes their "going for market share" excuse confusing, to put it politely).

Question to all the goys in here: EVGA 1070 SC or FTW?

>>55695570
>RX 480 is a very low margin part and the 4GB is even worse than the 8GB.

If you mean low binned, no. Also the memory is exactly the same as well.

The lack of 4GB is probably because they could actually cut costs and ramp up more production by only having one line. It's too new and they didn't have enough time to bin anyway.

>>55695234
the hype was this, at worst it would be 390/970 and if the ocs were true up to just below a 980ti

ocs were either bullshit, or they had golden silicon.

we are now seeing from people who fucking hate amd (hardocp) that 3rd party venders will launch 1500-1600 but the silicon lottery is really for those speeds.

what these do is still up in the air, but if benches are right, its over a 1060 which is over a 980 so right below a 980ti, and these were on 1420 mhz.

the hype was largely correct, however, we got a kick in the teeth from the power delivery they used, the crap stock cooler, and the wait for good third party.

>tfw AMD cut everything to win in the mid-range
>Nvidia just releases a better mid-range card after they release their high end

>>55693165
People with money to spare on such a thing.

>>55695586
SC is the stock pcb with evga's cooler. What do you think.

>>55695586
>EVGA
no

>>55695588
low margin = low profit relative to the cost of the product

>>55695570
>I think the RX 480 is a very low margin
Not really, the fact that die sizes are small gives credence to the contrary. Nvidia must be suffering a lot more on those margins, hence the higher prices.

File: heatpipe.jpg (78KB, 790x378px) Image search: [Google]

78KB, 790x378px

>EVGA

ayy lamo

>>55695618
Why not? Am I supposed to buy some shitty Gigabyte card?

>>55695601
>people who hate amd (hardocp)
Guess you didn't read their 1060 review - they were sucking AMD's balls over the RX 480's Vulkan performance (you know, in that one game that doesn't have nVidia Vulkan support yet?) and DX12 (which is questionable as many games were shown to run better on the 1060). Most other sites clearly stated 1060 as the overall winner. It really threw me for a loop, seemed bias as fuck which is odd as everyone is saying they're biased in the other direction.

>>55695636
Nah, nvidia's margins are always large because AMD is so pathetic.
Nvidia hasn't sold a low margin card since the 200 series

>>55695586
This thread has already been succesfully highjacked by AMDrones. So your answer is neither: get 2 Rx 480 in Crossfire™

>>55695636
It's not just the cost of the bill of materials, R&D costs and a new architecture research and development on a $250 product? That can't be good; at least that's the way I'm looking at it. AMD beat revenue and earnings last quarter (of which RX 480 wasn't a part but for 1-2 days) but still lost money and will most likely lose money next quarter as well (even with """tremendous""" RX 480 sales) - that's a telling sign.

>>55695418
i honesty find it hard to believe some of that overhype was amd shills, le the 3.2ghz version of the image using pink, some of those images were fucking retarded, and obviously to make fun of amd, and those made up most of the threads around the launch window.

File: mfwletty.jpg (60KB, 332x352px) Image search: [Google]

60KB, 332x352px

>>55695638
At least the die is not touching the cut pipe.

>>55695659
>2x 480
Please no jokes

>>55695636
nVidia is printing money right now. They raised prices because they can. AMD lowered prices because they have to. Simple concept but not one that AMD fans care to admit.

>>55695501
What SMP looks like is when you render a scene in blender and project the rendering on the scene in viewport so you get the 60+fps to move around in it.

If nvidia is doing something like that, rendering out the entire screen then moving the camera a bit, that is honestly amazing no one else thought of it, as we have been baking fake ao for years before real ao was a thing.

>>55695680
Ok, but what does it say when AMDrones believed and defended it? Don't you think some of the Vega nonsense being thrown around is similar?

>>55695641
>gigabyte
>literally any better than EVGA
Anon wtf is wrong with you?
Go buy an ASUS or MSI or Zotac or even PNY
Why the fuck... Nvm I don't wanna know

>>55695646
Kyle either fell in love with Amada or had a nasty concussion and forgot who he was shilling for.

That review was weird as fuck.

>>55695703
Where did AMD lower their price? If anything the 480 rushed Nvidia to launch the 1060 to market sooner instead of trying to get poeple to buy the 1070

File: RadeonProDuo1.jpg (239KB, 1200x746px) Image search: [Google]

239KB, 1200x746px

>>55694109
Wanna guess the price of pic related ?

You guys complaining about the price are all missing the fact that cards like the Titan or Radeon Pro Duo are not targeted for average consumers or gaymers. They're meant as a poor man's Quadro or, in Radeon Pro Duo's case, Firepro, targeted towards content creators that want a workstation that can both play video games as well as having a kickass graphics card, that supports features that normal gaymer cards don't, for the applications they use to make money.

If you go and check out the prices on high-end Quadro/Firepro cards, you'll see that Titan X and Pro Duo really are poor man's entry level content creation/compute cards.

>$1,200
how can any sane person take this card seriously?

that kind of money you can buy dual 480's and a nice freesync 1440p 144hz monitor or a 1080 and a jewsync 144hz 144p monitor and max out all the games and enjoy a great gaming experience.

>>55695544
Try out a vr demo if any are close to you, is amazing how good even shitty 320x180 looks when its 3d and your head is a control axis.

But then I love racing games, and if even 1 roller coaster game gets a good cockpit view... god damn that may be the last game I ever need to buy.

>>55695722
I miss the days where you could just BIOS flash to a Quadro. They killed it all with Kepler and the perf/watt meme. No fucking compute.

>>55695646
the only games shown to run better where all dx11 titles.

the reason why nvidia had stronger dx11 drivers was because they were multi-threaded which helped lower driver overhead since it could be spread out across multiple threads.

but another reason was because of their use of a software scheduler.

one of the reasons why fermi ran so hot was because it utilized a hardware scheduler, just like all amd gcn based cards do. hardware scheduling draws a lot of power and more power means more heat. why did they use a hardware scheduler? a hardware scheduler will always be faster than a software one. less overhead, and the gpu can do it much faster than software.

the problem with a hardware scheduler? once built, you cannot modify it. you have to build a whole new card if you update the hardware scheduler.

but nvidia wanting to move on from their house fire fermi's decided to remove hardware based scheduling with keplar and beyond. this is the main reason why keplar used far less power and ran cooler than fermi. nvidia realized with dx11, you didn't need a complex hardware scheduler. most of the scheduler went under utilized and was overkill. with dx11 multi-threading capabilities, and making their drivers multi-threaded, it help alleviate a lot of the driver overhead one would endure with utilizing a software scheduler. in turn this gave them the opportunity to now have more control over scheduling. able to fine tune the drivers for individual games. well, they had to. this caused a lot of work on nvidia's driver team, but it helped them max out every ounce of juice they can get from their cards and lower power and reduce heat.

maxwell continued this by removing more hardware based scheduling.

the problem? dx12 and vulkan need a hardware scheduler to be taken full advantage of. you need it for the complex computations of async and to manage compute + graphic operations at the same time. they're complex, and you need the performance.

>>55695218
>A 1080 outperforms a Titan X by like double I really think they need to re-evaluate their pricing.

titan x outperforms a 1080 by a few % give or take depending on the game.

>>55695731

it's not a price/perf card. it's a prosumer/enthusiast gaymer card.

>>55695731
It's called having a job

>>55695747
and this is why nvidia cards cannot do async properly. not only do they not have the hardware needed to run compute + graphics at the same exact time, but they lack the complex, high performance hardware scheduler to run them. their hardware can only do compute or graphics one at a time. with pascal nvidia did some tweaks to help speed up the switching between compute and graphics, but it still isn't optimal. its a bandaid. pascal still comes to a crawl if it recieves to many compute + graphic operations. it cannot swith fast enough.

whats funny is nvidia knew what they were doing. they just didn't think compute was ever going to be useful in graphics and games.

here's a nice article from keplar's launch done by anandtech:
>http://www.anandtech.com/show/5699/nvidia-geforce-gtx-680-review/3

>GF114, owing to its heritage as a compute GPU, had a rather complex scheduler. Fermi GPUs not only did basic scheduling in hardware such as register scoreboarding (keeping track of warps waiting on memory accesses and other long latency operations) and choosing the next warp from the pool to execute, but Fermi was also responsible for scheduling instructions within the warps themselves. While hardware scheduling of this nature is not difficult, it is relatively expensive on both a power and area efficiency basis as it requires implementing a complex hardware block to do dependency checking and prevent other types of data hazards. And since GK104 was to have 32 of these complex hardware schedulers, the scheduling system was reevaluated based on area and power efficiency, and eventually stripped down.

>>55695714
>ASUS or MSI or Zotac or even PNY
Thanks for the laughs

>>55695757
>The end result is an interesting one, if only because by conventional standards it’s going in reverse. With GK104 NVIDIA is going back to static scheduling. Traditionally, processors have started with static scheduling and then moved to hardware scheduling as both software and hardware complexity has increased. Hardware instruction scheduling allows the processor to schedule instructions in the most efficient manner in real time as conditions permit, as opposed to strictly following the order of the code itself regardless of the code’s efficiency. This in turn improves the performance of the processor.
>Ultimately it remains to be seen just what the impact of this move will be. Hardware scheduling makes all the sense in the world for complex compute applications, which is a big reason why Fermi had hardware scheduling in the first place, and for that matter why AMD moved to hardware scheduling with GCN. At the same time however when it comes to graphics workloads even complex shader programs are simple relative to complex compute applications, so it’s not at all clear that this will have a significant impact on graphics performance, and indeed if it did have a significant impact on graphics performance we can’t imagine NVIDIA would go this way.
>What is clear at this time though is that NVIDIA is pitching GTX 680 specifically for consumer graphics while downplaying compute, which says a lot right there. Given their call for efficiency and how some of Fermi’s compute capabilities were already stripped for GF114, this does read like an attempt to further strip compute capabilities from their consumer GPUs in order to boost efficiency. Amusingly, whereas AMD seems to have moved closer to Fermi with GCN by adding compute performance, NVIDIA seems to have moved closer to Cayman with Kepler by taking it away.

>>55695712
It uses a single geometry setup stage for multiple projections:

http://www.tomshardware.com/reviews/nvidia-geforce-gtx-1080-pascal,4572-3.html

>Nvidia calls the first its Simultaneous Multi-Projection Engine. This feature is enabled through a hardware block added to GP104’s PolyMorph Engines. This piece of logic takes the geometry data and processes it through as many as 16 projections from one viewpoint. Or it can offset the viewpoint for stereo applications, replicating geometry as many as 32 times in hardware and without the expensive performance overhead you’d incur if you tried to achieve the same effect without SMP.

>Take VR as an example. You already need one projection for each eye. Today, games simply render to the two screens separately, incurring all of the aforementioned inefficiencies. But because SMP supports a pair of projection centers, they can both be rendered in one pass using a feature Nvidia calls Single Pass Stereo. Vertex processing happens once, and SMP kicks back two positions for each vertex corresponding to your left and right eyes. From there, SMP can apply additional projections to enable a feature referred to as Lens Matched Shading.

>>55695764
important part here:
>NVIDIA is pitching GTX 680 specifically for consumer graphics while downplaying compute
>downplaying compute

it's also why in nvidia's "dx12, does and don'ts" they state not to run to many compute + graphic operations at the same time.
>https://developer.nvidia.com/dx12-dos-and-donts

their hardware cannot handle it. while amd's gcn not only can, but shines brighter when its under heavy async load.

here's some more interesting reads on nvidia's async debacle:
>http://www.overclock.net/t/1606224/various-futuremarks-time-spy-directx-12-benchmark-compromised-less-compute-parallelism-than-doom-aots-also

yes its mostly focused on the time spy issue regarding their usage of async, but it does dwell into nvidia's architecture limitations.

also the use of the hardware scheduler is why amd gpu's used more power and ran hotter than nvidia's since the keplar and gcn 1 days. if nvidia slapped a hardware scheduler on pascal, their gpu's would not just use as much, but most likely use more than amd's since nvidia is on 16nm instead of 14nm like amd.

>>55695675
You are corrrect, nonetheless it is also a lot easier to spread the R&D costs over a high yielding total amount of chips/waffer than over low yielding big dies (where the fab costs starts to hurt much more), especially when both are going to high demand products and selling like literal hotcakes.

>>55695716
Ok, at least I'm not the only one who thought so. It was Brent Justice who did the review and he was defending AMD hard the few days the PEG power draw issue was around (and, frankly, did not have a solid understanding of the issue).

>>55695646
Watch Joker Productions video on the 1060 vs RX 480 DX12 review.
It's strictly DX12 And he's historically got a bit of a hardon for nVidia and he says the 1060 is garbage and to either get a 480 or a 1070 but not to waste your money on a 1060

File: 1469136985372.png (498KB, 1920x1080px) Image search: [Google]

498KB, 1920x1080px

with the way gcn works, it scales more with a stronger processor than a weaker one.

in doom - vulkan, async is enabled on amd cards and async is used HEAVILY in doom - vulkan. the older cpu's cannot feed the ace's and cu's fast enough. you still get a boost, but not as big. slap in a 6700k and, well in doom, it turns that $200 card into a $400 one. its able to keep up with the 480 and feed it plenty.

nvidia on the other hand doesn't have async enabled. id disabled it since it gives nvidia cards a regression and they're waiting for nvidia to release a driver to reenable async for nvidia cards (nvidia lacks hardware schedulers and their cards cannot do graphics + compute at the same time on the hardware level, so they depend on software scheduling and preemption). so the only benefit nvidia is getting is the general less driver overhead. which is why nvidia gets a bigger boost with older cpus and not newer, stronger cpus. the older ones cannot keep up with the driver overhead, so switching over to dx12 frees up a lot of resources for older cpu's while the 6700k is strong enough that it doesn't matter so nvidia see's less of a boost.

thats why you'll notice the stronger the processor becomes, the less of a boost the 1060 receives, and the higher the boost the 480 starts to receive.

gcn is built to be fed, and utilize async. the more you feed it, the more powerful it becomes. give it a ton of things to do and it shine. vulkan / dx12 will always give amd a boost but the stronger the cpu, the more boost you'll get.

if you're building a pc now a simple 6100 is more than enough for a 480. if you're on a first generation i7, it be best to upgrade. regardless if its amd or nvidia.

>tldr
processors like the x4 955 & i5 750 bottleneck the 480 in dx12 / vulkan. you still get a boost, but not as great. they cannot keep up.

nvidia receives a weaker boost with stronger processors because driver overhead is less of an issue with stronger processors, like haswell and skylel.

>>55695721
You don't really believe AMD spent years and billions in R&D on a $250 GPU, do you? I believe Kyle Bennett's info on how things went down and that AMD had to scramble to rebrand RX 480 as a lower end GPU than what it was meant to be. Even if you don't, based on GTX 1070 pricing, AMD could have gotten away with $300 for RX 480.

>>55695218
>re-evaluate their pricing.
Why? This is aimed for the extreme end of enthusiasts who don't care about price to performance ratio. In other words not a card for you.

File: 1464101560343.jpg (3MB, 9999x9999px) Image search: [Google]

3MB, 9999x9999px

>>55692957
>https://blogs.nvidia.com/blog/2016/07/21/titan-x/
will all you faggots stop looking at that page as it won't load for me

desu i would have bought that card as i have a 21" analogue monitor and it and the 980 are the last to have analogue outputs

not a poor fag, but i don't want a crappy lcd (crt rules (NO LAG)

File: timespy-3.png (270KB, 602x869px) Image search: [Google]

270KB, 602x869px

even in a benchmark that performs async similar to nvidia's dx12 "do's and don'ts" list, where it uses little async (under 30) and mostly compute with little graphics, amd still showed a stronger boost than nvidia.

yes nvidia came on top with the 1070 and 1080, but their boost with async on isn't as great as amd's. even the 480 scored higher gains than the 1080.

>>55695822
>why 480 score less than fury?!
fiji contains more compute cores. fury x contains 64 vs the 40 of the 480. the fact that the 480 still scored ~4% less really shows how more efficient it is though.

File: 1461749604019.jpg (256KB, 539x559px) Image search: [Google]

256KB, 539x559px

>>55695636
>AMD today announced revenue for the second quarter of 2016 of $1,027 million, operating loss of $8 million

it was 70m loss q1, 137m q4 2015

and since we know only things they sell is gpus and cpus for server access(what's the thing called?) i'd say you have a point

>>55695738
Wasn't impossible to accomplish this on the 6/700 series, but certainly more difficult than a simple BIOS flash: http://www.guztech.nl/wordpress/index.php/2013/11/researching-nvidia-gpus-geforce-gtx780-and-gtx-titan-to-tesla-k20-and-tesla-k20x/
No idea if this has been done on any 900 series cards as of yet, though.

>>55695822
also the reason why there is no boost at all with maxwell is because futuremark stated async was disabled regardless for maxwell cards due to nvidia.

>http://www.pcper.com/reviews/Graphics-Cards/3DMark-Time-Spy-Looking-DX12-Asynchronous-Compute-Performance

>>55695847
wow 70 million loss down to 8 million.

>>55695845
>still scored ~4% less
The card scored 20% less, that percentages are the difference in score depending on async being en- or disabled.

>>55695780
A high yield, low cost design that is selling like hotcakes even at its launch price is just about the best thing AMD could hope for to recoup R&D costs. That's 60-70% of the market and most of the money involved.

>>55695758
Literally who else aside from those six manufacturers are nVidia AIB Partners in N.A.?
Oh... Nobody.
Well then

>>55695753
That's not having a job, that's being an idiot.

>>55694903
>>55694936
lets look at it this way, the current leaks for the cip vega will be, and any idiot watching, suggests it will have at least 64 rops, this points to potentially double the card, remember crossfire 480 already match a 1080 overall some games resolutions it beats it, some it loses but it always keeps pace if corssfire works...

now imagine the card that is 2 480's in one, without the crossfire scaling issue (sure its 85-95% but its still performance on the table)

will a card halfway between 9 and 11 tflop beat it? thats an honest question right now will it?

Please nvidia just finish AMD quick, this is brutal to watch

>>55695885
EVGA has good returns rep, they have to because of high number of returns. If you buy anything else you'd at least wouldn't have to RMA it.

>>55695863
Team in the Red till the bitter end.

>>55695847
was their recent one time IP sale to China included in that quater's figures?

>>55695931
licensing deal, not sale

>>55693284
What is a double point workload? I have honestly been wondering this for a long time and trying to figure out what this workload is so I know if nvidia is really screwing people or if its a niche that was worth killing.

Also, may pro workloads are now ram bottlenecked, but the programs till they hit that ram limit respond better to higher clocks, there are very few programs that demand pro drivers to work.

File: 1450598134207.jpg (74KB, 483x381px) Image search: [Google]

74KB, 483x381px

>>55695920
>2021
>AMD is dead
>"GET YOUR BRAND NEW TITAN XL FOR 5999.99 US DOLLARS TODAY"
>JUST

>>55695808
>I believe Kyle Bennett's info on how things went down and that AMD had to scramble to rebrand RX 480 as a lower end GPU than what it was meant to be.

There's not a single fact to substanciate this theory other than the voices inside Kyle's head.

>>55695952
Not quite, they still have to compete against their previous generation.
They'll just end up like intel is now, having to work only because their previous generation of products still exists

File: matron.jpg (28KB, 367x348px) Image search: [Google]

28KB, 367x348px

>>55694621
the only problem i have with that is the heat
why?
well in my room it gets fecking hot and i mean hot, i have a set of window fans and it still gets hot, excruciatingly hot

it's 2016 when the hell are they gonig to make a card that idles like my 5870 at about 31°C it never goes over 68°C

>>55694264
fun thing is, college diplomas are getting devalued as fuck when to be a secretary you need a degree.

big companies don't want to do on the job training something required regardless of if you went to college or not, and instead just want you heavily in debt you can't bankruptcy get rid of so you are desperate to keep the shit job for shit pay while nepotism is the only way to get a non dead end job.

>>55693244
pump sounds, still quieter then air
coil whine, yea, you hear that when the cooler is basically silent
housefire... no
underperforming... just because nvidia slit its 1000$ gpus wrists to bring something to that price bracket doesn't mean it underperformed.

>>55693203
>Mountain View
>Bay Area
Hmm, wonder if this guy has money...

File: 1452157380144.gif (472KB, 250x269px) Image search: [Google]

472KB, 250x269px

>>55695964
>actually wanting 3% perf boost per year

>>55695457
To be honest I was hoping for something that sat between the 1060/480 and 1070 for around the $300 mark. Something that lifts me to around 80-100 fps in most games and never (or rarely) dips below 60. Then I could lock off at 60 on my 60Hz monitor and not worry. My issue is games tend to sit in the 30-90 area and I don't want to go out and buy a Freesync monitor because I would need to change to 16:9 instead of my nice 16:10 ratio and it would have to be 1440p IPS which means driving that resolution would require a better GPU anyhow and also IPS panels are all over the place right now. I can't win no matter which way I turn.

>>55696040
>IPS panels are all over the place with uniformity right now.
Fixed.

>>55695923
Exactly my point. Fuck GB and EVGA
>>55695909
What's the question?
>>55695961
Who the fuck is this fucking nerd, Kyle Bennett?
>>55695978
Too hot? Pony up bitch, time for phase change cooling where it'll never exceed 0°C
>>55695979
Pro-tip
Work for government for 10 years as your college debt is forgiven as long as you make minimum payments on it.
Then contract for ten years and work on continuing to build 401k etc, then go back to government work for 10 years and have govt pension and 401k etc

>>55696005
Nope it all goes to his $8k monthly 600sq-ft studio flat rent

File: async.jpg (360KB, 712x637px) Image search: [Google]

360KB, 712x637px

>>55695822
Async's performance gains don't come out of thin air. Nvidia has saturated the GPU to a greater extent than AMD has without async, so it's only natural that they'd see smaller benefits from it.

>>55695931
they didn't get the money yet, so probably not

>>55695713
I honestly don't think any of them did, they were so obviously bullshit I find it hard to believe. However on an anon image board, you can be anon, blend into the crowd and be a dumbass for a few weeks saying it would oc higher then cpus and kill nvidias volta cards this gen, then still be anon and point to yourself, non of us the wiser, and say 'they' were fucking retards and an objectively fantastic card is shit and underwhelming.

personally i expected a bit more form a 14nm die shrink from 28nm, but third parties seem to be able to clock it to where I was hoping it would preform.

Oh, one more thing, I'm on a 280x, before that a 5770, and before that a 6800ultra before that a rage 128mb (i think, possibly 64) and before that viper diamond 32mb (cant remember more than that) the nvidia was an outlier, i usually go for mid range as best performance per dollar, and i keep cards usually till they fucking cant be used, the 280x was so video editing would be better, but the card cant render video for some reason, apparently it can, but i cant get it to.

the 480 at 970~ performance is a meh, ill wait, at 980+ its more i want it now, especially considering the hardware decode that my cpu struggles a bit with.

>>55695767
No idea what any of that means, not that smart, I just know blender and something similar where you take a long render can project it onto the geometry in the scene and move around in it with next to no processor use.

>>55696052
>Work for government for 10 years as your college debt is forgiven as long as you make minimum payments on it.

its not forgiven, it's seen on tax as an income, so lets say you had 80grand in debt, on a 40k a year job, then that 80 grand is now seen as income so you have to file 120,000$ income that year and you are completely fucked as far as taxes go.

File: goin underground.gif (3MB, 224x328px) Image search: [Google]

3MB, 224x328px

>>55696052
lol that would be better

i keep meaning to get a acu but being a britfag our summers usually only last a few weeks, and the heatwaves a few days,

my biggest problem is my room is only 7' by 13' and i have a shitload of figurines, so space is at a premium, acu's range from £300 (398.32 US Dollars) to seriously silly money, i could afford one but they cost an arm and a leg to run (i might buy myself one for my birthday)

>>55695802
>all this FUD
The point of Vulkan and DX12 was to reduce driver overhead by massive amounts, you worthless idiot cocksucking shill.

>>55696206
damn, stating facts seems to make the nvidia PR team salty as fuck

>>55696080

this. GCN is inherently underutilized because it always has more memory bandwidth and SPs than it does ROPs/TMUs. if the ROPs are bottlenecking the GPU then the SPs can continue crunching with async, on NVIDIA's side the GPU is never front end bottlenecked like that because NVIDIA always focuses on improving pixel fillrate/throughput first.

>>55695802

where are you getting the idea that people are going to buy a $400 6700k and a $150 Z170 motherboard to go with their $240 GPU?

most people buying low end GPUs are doing so to upgrade ancient sandy bridge systems or to build new low end pentium/i3/i5 6400 systems.

>>55696314
i don't know what those words mean but NVIDIOT SHILLS ON SUICIDE WATCH

File: 1360368220235.jpg (80KB, 720x720px) Image search: [Google]

80KB, 720x720px

>>55696351

>>55694303
clearly you haven't met rich people.

I went to a casino with my boss and his boss one night. He lost something like $20,000 that even and was laughing about it.

also the only time I have ever smoked a cigar, apparently it was a $500 cigar. It was pretty gross desu.

even state welfare scum can afford to drop thousands on a shitty 100" tv they don't need too.

Spending $1200 on a graphics card doesn't make you rich, nor is it much money at all.

>>55694378
that's the problem here. getting a titan with 12gb of vram and it's not hbm2. these niggas retarded

I'm definitely in the market for the performance the titan X pascal offers, but I am in no way prepared to pay Nvidia's jew tax (even if the titan has proper DP again). So assuming AMD's equivalent is the 490 or fury x 2 or whatever the fuck i'm really, really going to be keeping an eye on it. A 4k screen is my next major upgrade and I need a gpu to drive it.