[ 3 / biz / cgl / ck / diy / fa / g / ic / jp / lit / sci / tg / vr / vt ] [ index / top / reports / report a bug ] [ 4plebs / archived.moe / rbt ]

Due to resource constraints, /g/ and /tg/ will no longer be archived or available. Other archivers continue to archive these boards.Become a Patron!

/g/ - Technology

View post   

[ Toggle deleted replies ]
File: 85 KB, 734x434, amd btfo.jpg [View same] [iqdb] [saucenao] [google] [report]
58975309 No.58975309 [Reply] [Original] [archived.moe] [rbt]

NVIDIA GTX 1080 Ti Rumored To Launch In March, Reportedly Now In Production


>> No.58975315

Go spam your blog somewhere else.

>> No.58975326

>letting AMD know a month ahead of time what performance level to target
What fucking idiots.

>> No.58975350


Vega doesn't even come close to the 1080

Not sure who the idiot is here m8

>> No.58975367

That's just in time for AMD to use 1080ti's to show off their CPUs

>> No.58975384

Well you're a fucking retard then, because all the leaks we have of Vega show that it is indeed very close if not better than the 1080 in performance.

t. 1070 owner

>> No.58975397

is it even relevant? 1080 still around 700-800
who insane enough to pay even more? it goes into titan x territory

>> No.58975457
File: 124 KB, 880x960, image.png [View same] [iqdb] [saucenao] [google] [report]

I mean doom mightve honestly been the poorest choice to be showing off GPU power on, it runs very very well
I'm excited to see more vega tests but reaching 1070/1080's OpenGL performance, in a very well optimized game AND using the other well optimized API, is not really a fair or equal comparison

>> No.58975483
File: 337 KB, 867x848, 1430775192503.jpg [View same] [iqdb] [saucenao] [google] [report]

>founders edition for 1100 USD

>> No.58975512
File: 9 KB, 237x218, dat faec.jpg [View same] [iqdb] [saucenao] [google] [report]

>wake up
>realize that new GPUs and CPUs were probably released while I was asleep
>check benchmark sites and make 20 threads about the latest GPU and CPU on /g/ee
>fall asleep

>> No.58975520


>> No.58975523


>70 frames while Vega is screaming in your case like a burning jet engine with preferred Vulkan API and Doom

1070 does it better


>> No.58975529
File: 335 KB, 320x225, giphy.gif [View same] [iqdb] [saucenao] [google] [report]


>> No.58975612

>Very close if not better than the 1080
>Choosing Doom running on Vulkan where RX 480 has 30% performance gain when compared to Open GL just to show us a meaningless 10% lead against stock 1080.

Their flagship Vega is going to be behind 1080 in DX11 titles and 1440p. And that's not even close to what Titan X has to offer, or the potentially cheaper 1080 ti. And then you have Volta which will absolutely ravage both Vega and Pascal.

AMD fucked up their GPUs as always. Somehow they're still playing catch up but at least with Ryzen the 5% IPC difference doesn't matter because the performance per dollar is absolutely insane. That's not the case with their GPU marketing policy of releasing same GPU for 5 years while giving you a $50 discount each time.

>> No.58975655

Not a full gp102

>> No.58975685

Doom wasn't the only game they demoed. The Battlefront demo shows pretty conclusively that it's very close to the performance of the 1080. It was able to do 60+ fps at 4K on a map that the 1080 struggles to achieve with lower graphical settings.

>> No.58975698

so it's basically Titan

>> No.58975713

Except nobody knows what the settings for Battlefront were. BF was always optimized towards AMD.

>> No.58975725
File: 30 KB, 864x745, image.png [View same] [iqdb] [saucenao] [google] [report]

You mean the 980ti?

>> No.58975749

Yes we do. The settings were maxed out with FXAA on. And Battlefront doesn't favor AMD nearly to the same degree that Doom does.

>> No.58975760
File: 119 KB, 728x622, image.jpg [View same] [iqdb] [saucenao] [google] [report]

Yes he meant two of nvidias top-of-the line cards from last year, that performs basically identical to their 1070 now
Yes he's saying it takes 2 $300 cards to match vega or the $600 card

>> No.58975777

Only the SLI 980tis get more than 60 fps so I'm not sure what your point is. That also doesn't say anything about what map it was done on. The performance difference from map to map is really big in Battlefront.

>> No.58975783


>> No.58975794

There's also the OC 980ti that's more relevant than the stock founders edition

>> No.58975811

Launch 20-23

So the PAX East presentation on 10-12th march is not fake. Good.

>> No.58975813

>reference: 44fps
>aftermarket card: 60+fps
Why is this allowed?

>> No.58975820

>Nvidia has Pascal refresh
>AMD has brand new architecture
Unless Nvidia can put out Volta a year earlier than expected my money is on AMD

>> No.58975858

>980ti = 1080ti

K...and poolaris was not even faster than the 390x...

>> No.58975860

Again, no idea what map that's benchmarked on.

Here's the map the Vega demo was done on: https://www.youtube.com/watch?v=eNTAMApSlz4

He's using an overclocked 1080 and getting frequent drops below 60 fps, and he doesn't even have FXAA turned on. Even if you take into account the extra action compared to the Vega demo, it's pretty good evidence that it's on par with a stock 1080 at worst.

>> No.58975881

I mean I would trust that the guys benchmarking it aren't just looking down at the ground
Does battlefront not have a benchmark scene or something like that?

>> No.58975896

The 1080 doesn't even dip down to 60fps on max kek

>> No.58975908

>half a year old video
>150 of them are from you

>> No.58975934

I don't think we watched to same video. The 1080 there is dipping below 60 fps constantly. The Vega demo was capped to 60 fps, so the actual framerate could have been higher. Not to mention the higher graphical settings it was using.

>> No.58975964

>Incapable of reading
>Still thinking Polaris was supposed to be high end
It's literally half the size of Hawaii and costs half as much.

>> No.58976200


>> No.58976251

Every benchmark expect that video with 3 views a day shows the 1080 well above 60fps the whole time

>> No.58976323

That tells me nothing if I don't know anything about what maps they were tested on

That video I posted is only relevant because it's the same map

>> No.58976328

Are we still discussing this? AMD taped the Vega exhaust on purpose so you don't know the performance 6 months away from release, since we have absolutely no idea how the card responds to throttling.

tl;dr you're literally arguing about hearsay

>> No.58976389



>> No.58976416

I'm strapped for cash and didn't want to wait, so I picked up a 1050 2GB for $110, is it a good purchase?

>> No.58976431
File: 27 KB, 500x365, image.jpg [View same] [iqdb] [saucenao] [google] [report]

>all benchmarks are faked because it doesn't support AMD's RYZE to the top!!!

>> No.58976443

>Show Zen doing good in benchmarks
It's actually performing better than expected in leaks.
>Show Vega doing bad in benchmarks

What is Lisa Su's master plan?

>> No.58976488

More like it's complete retardation to use different maps that have huge differences in performance on the same system to compare performance across GPUs.

>> No.58976514

keep in mind I'm only playing GTA 5.

I have an i7 6700 3.4ghz
12gb ram

>> No.58976539

Well I'm sure the benchmarks where the 1080 does the best are all just looking at the ground
And the benchmarks where the 1080 does the best are more legitimate tests

>> No.58976542

>NVIDIA GTX 1080 Ti Rumored To Launch In Every Month Until It's Launched

>> No.58976556

should have gotten a RX470 or at least a 1050ti 4gb

>> No.58976623

Only had $120 to spend, so those were out of my price range. maybe in the future

>> No.58976639

Nobody was looking at the ground you dumbass shill. You should use the same map for the same reason you use CPUs of similar power when making comparisons: to minimize the effects of the things other than what you're testing. Which in this case is GPU performance.

>> No.58976773
File: 37 KB, 678x381, 1474606862423.jpg [View same] [iqdb] [saucenao] [google] [report]


hotly anticipated newcomer reportedly with a grain of salt

cutting a few more SMs from GP102 allows NVIDIA to improve """yields""",just switch to samsung fabs already

The truth is that:

- P6000 is the real Pascal
- Titan XP is the real 1080Ti

>selling twice the same card

~$4800 US


>> No.58976838

>start combing through battlefront 4K benchmarks from different websites
>endor survival training
>endor survival training
>endor survival training
I mean one day-one release they might've been benchmarking willy nilly, but people clearly found the most consistently demanding mission to test, and the 1080 smashes in all of them.
But I mean the fury comes in second place, so since AMD is almost the best is it a legitimate benchmark now?
Im just saying that claiming the 1080 can't hold up 60fps is just wrong according to every single measured benchmark

>> No.58976862

kek did you just learn about the cluster removal?
This has been happening since the 780, and has been happening much more on AMDs side

>> No.58976871


notice P6000 destroys Titan XP and Ti only with a single 8 Pin versus 8+6 at exactly the same TDP

>> No.58976902
File: 276 KB, 1280x720, image.jpg [View same] [iqdb] [saucenao] [google] [report]

>he just now learned about this

>> No.58976933

P5000 is the real GP102.

>> No.58976944
File: 23 KB, 416x312, wambo.jpg [View same] [iqdb] [saucenao] [google] [report]

>AMD can simply choose the performance lvl of their next card as they please
>"quick Raja, nvidia has released the 1080ti and it's very powerful. what shall we do?"
>"no problem, I'll just switch the setting of our vega cards from Mini to Wambo"

>> No.58976951


This gen was particularly different for obvious reasons, DGX and self-driving; stock went thru the roof not that I complain but it's just the principle of artificial shortage and the arch is sometimes questionable on a clock to clock perf, next gen

>> No.58977066

>Im just saying that claiming the 1080 can't hold up 60fps is just wrong according to every single measured benchmark

That was never the claim to begin with. Some maps it achieves 60 fps at 4K handily, and on others like that Endor chase sequence, it struggles. Gettings 60 fps in one situation doesn't guarantee you'll get 60 fps in another.

Also, the Fury X performing that well at 4K shouldn't be a surprise, it has HBM. It beats the 980ti more often than not at that resolution.

>> No.58977089

i'm going to purchase 2 quadro P6000's and put them in SLi

>> No.58977129

Could you post that link so phone posters can open it?

>> No.58977138

>happening much more on AMD's side
I'd contest that, Nvidia used to make 4-5 derivatives of a single chip.

>> No.58977253

Remember Fiji?
Remember Hawaii?
Remember Polaris?
Remember the Titan?
Remember the Titan Z?
Remember the Titan X?
tl;dr: you can't avoid the jew

>> No.58977263

Besides the Z

>> No.58977295

>phone posters
lmao well deserved faggot

>> No.58977315


I'll admit the P6000 is sexy asf but something just doesn't feel right, it's like getting scraps at the door from light commercial monoliths specifically for 3584 CUDA from a designed 3840

Not really hitting the 10 multiplier, we're still in between revolutions but the P6000 really looks nice, I'm not sure why anyone would buy a Titan XP over a P6000, GPUs will cost the same price as the cars that they drive is quite conceivable at this point. Smart, as usual and you wouldn't expect any less from the field.

>> No.58977316
File: 57 KB, 516x380, bda.jpg [View same] [iqdb] [saucenao] [google] [report]

Well to be fair the Titan Z is TWO chips

>> No.58977345


Quadros are so fucking sexy.

>> No.58977355

Which spawned 2-4 gpus each

>> No.58977366

If Vega truly has binning rasterizers (like Maxwell/Pascal), Doom is probably the least flattering thing AMD could have been demoing.

idTech 6 does screen-space volumetric binning on the host processor, which is most of the reason that it runs so well on older architectures including GCN.

>> No.58977400

Well yeah binning is normal, why do you think it began with the 780?

I mean for fucks sake G92 had at least a dozen versions.

>> No.58977446

So why complain that the top-of the line card is using the complete top-of the line GPU?

>> No.58978517

Holy fuckballs I swear this card came out months ago. This is fucking with my mind a little bit.

>> No.58978797

Hey stranger you seem lost reddit is over yonder

>> No.58978820

Be smart and pair an athlon with a 1050ti

>> No.58978871


>> No.58978927


Can anyone who is better with hardware understanding please explain to me why the full GP102 GPU with more CUDA cores and such can have specs that mostly equal or surpass the Titan XP but somehow has the same TDP with 1 8-pin while the Titan XP has 8+6?

>> No.58978988

>I'm not sure why anyone would buy a Titan XP over a P6000

BEing within 10% of the performance for 20% of the price?

>> No.58979072

I still don't get why would anyone buy anything more powerful than a gtx1060 it's just overkill. The human eye can't even see past 480p/30fps

>> No.58979247
File: 14 KB, 264x264, 1454958664170.jpg [View same] [iqdb] [saucenao] [google] [report]

a (you) for u

>> No.58979295

Buying ATI was a mistake, I would say they should just cut their losses and sell their GPU division but I forgot they at least get console money plus Intel might be using AMD iGPUs in their next CPU release.

>> No.58979306

Thanks, I don't deserve it.

>> No.58980682

AMD killed ATI, ATI was so based

>> No.58980696

What the hell was based about no drivers?

>> No.58982584

Literally not spending 100 bucks more for the Titan ishygddt

>> No.58982894

>1080 still around 700-800
Where the hell are you shopping? You can get a 1080 for sub 600 now

>> No.58982919

In NZ its still $799 for a GTX1070

>> No.58983016

they kind of can, they can use same die they sell enterprise and sell it for less with small things disabled

all the speculation about doom/battlefront is pointless, ES usually is half the card
they may as well emulate before printing with same result, it's mostly about power delivery bug testing

>> No.58983060

$916 as of current rates here

>> No.58983100

>March 2017: Vega gets released
>November 2017: Vega gets released with a fan that does not resemble a Mig-27 takeoff.

I hate AMD for this.

>> No.58983163

Even if Vega is better than the 1080Ti, Volta will completely BTFO Vega. I'm predicting AMD's success here will be short lived. They just don't have the resources to to run neck-and-neck with Intel and their buddies anymore.

>> No.58983187

Hold up, are you guys seriously gonna waste money on a GTX1080 TI when a GTX1070 is enough for most games?

>> No.58983191

Yes because I'm not poor.

>> No.58983231

Any chance the memory bandwidth will be increased with the Ti?

What is usually the difference with the Ti editions?

>> No.58983254

GTX 1050 is 2GB
1050 Ti is 4GB

>> No.58983261

try canadacuckland

>> No.58983276

Oh I should've read the article. 480GB/s just like the TITAN X. +50% over the standard 1080.

If it really is going to have 12GB, this will be a deep learning monster where memory bandwidth is very often the bottleneck.

>> No.58983278

volta might go very wrong, nvidia plans for 10nm(remember how 20nm worked out for everyone?)
it's not coming until 2018, unless they magically solved scalability that both promise in volta and navi
and navi comes out in 2019

so either rebrand or magic. because AMD never was far behind on actual GPU tech

>> No.58983288

>this will be a deep learning monster
consumer NVIDIA cards can't DEEP learn by design, they even cut it from Titan

>> No.58983303

>$1000 card
>mfw i bought a hd 7770 for $30.
>tfw don't give a fuck

>> No.58983308

what's the retail? $1300?

>> No.58983326
File: 582 KB, 991x754, 1458273182906.gif [View same] [iqdb] [saucenao] [google] [report]

>mfw 1050 Ti coming in the mail

Gaming over 1080p and 60fps is a literal meme, so I don't give a fuck.

>> No.58983345

you do realize this is pre-drivers vega right? it's only going to get worse for you pascal early adopters. the biggest cucks are always x80 owners.

>> No.58983353

great can't wait for another 1000 dollar gpu, right alongside the 1000 dollar cpu from intel.

pc hardware definitely moving forward.

>> No.58983365

I know the sub ~$6k Pascal cards don't do half-precision, effectively cutting their memory in half. But at 1/6 the price and for 1/4 the memory, the consumer 12GB cards are still pretty respectable for hobby-level DL. Especially for those not interested in images and video DL.

>> No.58983395
File: 52 KB, 400x400, wccftech.jpg [View same] [iqdb] [saucenao] [google] [report]


>> No.58983646

>hobby-level DL
it's a thing? you can't hope to teach net anything practical with small cards in reasonable time can you?

>> No.58984140

>And then you have Volta which will absolutely ravage both Vega and Pascal.
Volta is a Pascal refresh with HBM. Nvidia has already mentioned this.

>> No.58984393

Does that mean 1070 prices might drop?

>> No.58984397




>> No.58984431


Don't crush their dreams with reality so hard.

>> No.58984631


>> No.58984934
File: 1.15 MB, 2560x1440, Ghost ReconĀ® Wildlands - Closed Beta2017-2-3-14-28-46.jpg [View same] [iqdb] [saucenao] [google] [report]

I have i5 and 1070 overkill for it. There is no point in new cards unless you have 8 threads cpu or aiming for 4k. Ether of those are expensive as fuck.

>> No.58985024

>waiting for Vega

Waitfags will never learn.

1.It will be fucking expensive thanks to HBM2 memory

2.It will be faster than a GTX1080, but I doubt it will be the case with the GTX1080Ti

>> No.58985827

>8 threads

That's where you're wrong kiddo

>> No.58985886
File: 1.62 MB, 1920x1090, 1487016796513.png [View same] [iqdb] [saucenao] [google] [report]

>>letting AMD know a month ahead of time what performance level to target


AMD will never get even close to be as good as Nvidia or Intel.

We won, bros. Can't wait to get new GPU.

>> No.58986064
File: 102 KB, 450x443, 1480271903523.png [View same] [iqdb] [saucenao] [google] [report]

>We won, bros
child detected, unless you have assets invested into Intel/nVidia you are just a dumb fuck fanboy who glorifies monopoly while getting screwed by it

>> No.58986084
File: 944 KB, 1920x1693, 1487056040095.jpg [View same] [iqdb] [saucenao] [google] [report]

>buttblasted AMD poorfag detected

Burn yourself.

Name (leave empty)
Comment (leave empty)
Password [?]Password used for file deletion.