Nvidia Volta Q3 2017!!!!

>>59960283
rip AMDead

>>59960283
AMD is tearing up Intel but Nvidia is totally decimating them. How can they ever recover

File: amdvega.jpg (156KB, 633x758px) Image search: [Google]

156KB, 633x758px

>>59960283
THIS CANT BE HAPPENING

>>59960314
Vega finds a way.

>>59960283

Sad thing is this tells us more or equal information as to what we know about RX Vega which is supposed to be released anytime now

Was gonna buy a 1080ti, is it better to wait?

>>59960283
>>59960303
>>59960323

Memes aside does anyone here actually have loyalty to either manufacturer or just buys the best one? (I am a nvidicuck if it matters, but I hope ayyymd gives them some competition or even btfos them with their next release)

Mfw MSI makes the best looking gpus

>>59960283
So does Nvidia basically keep their new hotness under wraps until amd even tickles the idea of coming out with their new architecture so they can drop a bomb on the market and steal the wind out of amd's sails? I mean since they have such a secure hold on the market they don't need to do anything until amd tries something.

The chip hasn't even taped out, this is wccftech tier clickbait.

File: 1492425424471.gif (628KB, 516x402px) Image search: [Google]

628KB, 516x402px

And Vega still waiting™

For fuck's sake AMD!

>>59960423

pretty much. in the few cases where amd has had a competitive product it made nvidia drop a new card early, like they did with the 980ti.

>>59960399
I've had both amd and intel cpus in the past but always nvidia gpus because I need proper OGL drivers.

>>59960470
>proper OGL drivers
>exit a "full screen" application in linux, one third of the panel looks corrupted
Proper OGL drivers, they should get their 2D ones working first.

>>59960432
Yeah, it would be a manufacturing miracle to have good Volta silicon by the Q3 of this year.

Fake news.

>>59960432

>my father is an employee of Nvidia

>>59960432
>>59960504

Isnt it late to still be working at AMD offices?

>>59960283
>Chinese rumor mill
>GTX 2080
>GTX 1030

You fags will believe anything if it shits on AMD.

>>59960283
Oh, another GP100-tier paper launch? Go fuck yourself novidea.

>>59960528
Nvidia already confirmed volta will be 20x0

>>59960520
>thinks regular mooks at Nvidia know what tapeout means
This isn't Intel or Samsung, who got their own fabs and can keep a tapeout secret, when AMD and Nvidia's chips come from the oven as its from a third party fab.
Not only did we hear fucking nothing, there's also no zauba information yet.

>>59960546
>said the amdrone while nervoulsy waiting for volta

>>59960575
You mean Vega? And no, i have novideo card. But yes, i will buy Navi or whatever AMD will release after Vega, at least they have drivers. Like, actual drivers.

About a valid a rumor as Vega being 30% faster than the Titan XP

>>59960559
That's pretty sad.
Eleven eighty or twelve eighty sound much better that twenty eighty.

>>59960559
Source me, no sane company announces marketing names a year before launch.

मैं आप सभी को भारतीयों का मजाक बनाने के लिए नफरत करता हूं हम दोनों कंपनियों का प्रमुख होगा

I doubt it, the source if finky at best and even if they did, it would be the lowend shit like the 750Ti was.

Furthermore the TxP and 1080ti are too fresh.

>>59960684
>too fresh
GP102 is pretty old.

>>59960689
And we only got a full GP102.. a month ago?

>>59960283
This explains why King Poojeet wants to sell RTG to intel.

>>59960704
Yields are ass. That's also probably the reason Vega is so late. Big die, and they refined the process on poolaris.

>>59960718
What do Kyle's wild fantasies say about Kabylake-G using AMD's GPU as a MCM that's been floating around the last few weeks?

>>59960718
Kyle pls.

>>59960762
Sounds like a custom apple chip.

>>59960784
Sounds weird considering they can just opt for Raven Ridge.

>>59960802
Not if they want a powerful GPU, that's iMac and MBP territory.
Raven Ridge is still a small die, won't fit over 800 shaders there, even if they are NCU shaders

>>59960821
They can just order some custom design from AMD then. Remeber it's AMD who designed current console hardware.

>>59960827
Not enough time for custom designs, maybe next year.

>>59960838
Who knows, it's all Vega-related and ayymd is surprisingly silent on it's technical details. Like, what the fuck is NCU? What did they change? What is HBCC? I wish they showed us anything.

>>59960855
AMD's only custom design this year is Scorpio, there's simply not enough time to get a custom design for Apple when F4 Zen stepping came out literally in the middle of February. not to mention Vega which is still in clockspeed and driver hell.

>AMD releases Ryzen

>intel proceeds to shit itself and suddenly moves up Coffelake by 3+ months

>AMD is about to release Vega

>nvidia shits itself and starts hyping Volta release

Thanks AMD, if not for you we would be paying 1500$ for a gtx 1050 and another 2k for an i3

>>59960878
Scorpio is designed by MS themselves, and is fabbed on 16nm TSMC. AMD only helped with its design. But yeah, this year is surprisingly busy for AMD, so many big releases.

You'll probably get to see pictures of GV100 at GTC, but availability no sooner than Q1 2018 and that's starting with GV102

>>59960910
I wouldn't be surprised seeing GV106 come out sooner.

>>59960934
So Maxvell 2.0: electric boogaloo? I doubt it.

The rumor before this was that Volta will be 12nm, very reliable, before that, it was to release in May.

>>59960967
12nm is a meme anyway, almost no density increase. I hope Volta will have some proper hardware scheduling.

>>59960372
Yes

Are AMD and Nvidia competing on who releases their next gen chip first?
We had the RX480 released 8 months ago, AMD is refreshing it with the RX500 series and later this year Vega will launch. Polaris will get in total 1 year on the market then it's out the window.
Pascal doesn't even get a new series, GTX2000 will be Volta. not even 2 years on the market.

As a reference Fermi was released in April 2010 (GTX400), refreshed in November 2010 (GTX500) and replaced by Kepler in March 2012 (GTX600). 2 years on the market.
On AMD side the RX200 was released in October 2013 and replaced by the RX300 in June 2015, 2 years on the market.

Albeit, it is the best time to push for GPU development since games are the number 1 reason people are buying powerful GPUs, we can always create games with better graphics and higher resolution and the CPU, RAM and storage don't really bottleneck the GPU in games.
It's literally a gold rush right now.

>>59960945
Why do you think so?

>>59960984
ML is growing. That's another reason for both competitors to make more and more powerful GPU's.

>>59960998
Volta is supposed to be more akin to Vega, i.e. somewhat major architecture change. They'll paperlaunch it as usual with GV100 that you'll never see.

>>59961006
yea, GPU compute has been booming since 2010

>>59961006
Yeah, but the P6000 was released like literally a month or so ago, that's enterprise hardware and it's not replaced fast, so that makes no sense.
Same with the Vega Mi25, was revealed in December but released a few weeks ago as well.
Both are ML cards.

>>59960586
There is a sourced benchmark putting Vega at 35% faster than a 1080, so near 1080ti/XP range

It's going to be the fury for the next two years

>>59961032
Isn't P6000 a Quadro and Tesla is NVIDIA's pure compute line?

>>59961048
I know what you mean, but anyone can make pretty graphs.

>>59960902
>AMD only helped in design
All Ms did was I want x for x watts to run x resolution can you do it?

>>59961054
What did they gimp this time? Weren't Quadros FP16 or did they move that to Tesla?

Tesla have historically been FP64 and the P100 is FP64

>>59961079
I wonder if consumer Vega will have non-gimped compute.

>>59961109
Everyone would just whine about TDP if it does.

>>59961123
225w for 12.5tflops fp32 is too much? /g/ is weird.

>>59961146
Enable FP64 and that goes straight over 275W

>>59961170
And that's bad considering it could do 1/2 FP64?

>>59960902
>MS designed an APU and chipset to accompany it

yeah, no they didn't. They may not have even designed the mainboard, although they do have some experience in that area. But even if they did design the mainboard, that's comparable to being handed a jet engine and strapping it to a chassis and saying "LOOK I BUILT THIS"

>>59961183
There's no reasoning here, higher TDP == bad, even if the ASIC power is 20-30% smaller outside of FP64.
Faggots don't know, Tahiti actually suffered for this, it was 1/2dpfp

>>59961210
They should just use the kike Intel version called SDP, that can be anything you want.
Kinda like Core M's rated at 4W SDP pulling over 22W under boost, or the 15W ULVs going over 40W

When you got money you can get away with everything

>>59960283
>new cards already
It is like they enjoy pissing off their customers.

>>59961236
>when you got the money
And AMD has no money. Well, at least yet.

>>59961240
Since when did nvidia care about their customers

Is this real? I was thinking of getting a 2nd 1070 somewhen this summer but a Volta launch would come handy. I would even consider the 80 depending on the price. Also my budget depends on the capability to max out ME Andromeda.

>>59961283
Good point. Many things including the new titan X being crippleware show they give no fucks.

>>59961290
About as real as Intel's 6 cores at 5.0GHz

>>59961318
It'll probably be possible but that shit will pull over 300W alone, like only 3 motherboards won't spontaneously combust with it

>>59961318
I want to get a 4K TV first, hoping to get a 1070 FTW for around 300 € then. I'll decide when I have the money, probably September. 1080Ti might be an option if the prices drop, but I think my card will be worthless until then and it is only three months old...
>Bought a 1070 for 470 € weeks before 1080s for 519 € were avaiable
>New and we're talking about FTWs

>>59961358
unlikely. The instantaneous heat generated by a chip like that would make it horrendously unstable. That is to say, even if you could dissipate 300W and the chip never exceeded that range, the spikes would cause short lived voltage fluctuations and crash the system.

That's not even mentioning how difficult it would be to maintain a constant voltage across the chip in the first place.

>>59960984
> It's literally a gold rush right now.

Whoever puts out the first card for stable 4K 60FPS gaming at a non-enthusiast price point is going to make a shitton of money.

i was going to get a gtx 1070 in about 10-11 days, is it now a bad purchase?got a 660 atm

>>59961612
Don't forget that if your GPU can drive 4k, it can drive 90fps VR at 1920x1080 per eye as well. There are multiple markets to consider right now, although the VR market is still small. If the total cost for VR drops below about $1200, I think we'll see widespread adoption, especially since compelling software is beginning to be announced now.

>>59961408
you would need a platuim certified PSU aand those are expendive but its do-able

the Raedeon Pro duo uses 500W and that is just fine.
But that and a OC'd CPU would be questionable.
X370 MOBO's on the highend have extra Power motherboard for exterme set ups

the downside being you only get 2 dimm slots because of heat output is HUGE

>>59961671
If your on a 660 then any Opition is good heck even a 480 will BTFO off the GTX660

you are really far behind anon.

>tfw just bought a 1080 a couple weeks ago
guess im waiting for a 3080 then

>>59961855
ahh unfortunately i cant deny being behind, i suppose the difference between a 1070 and 2070 jumping from a 660 will be negligible anyway, im also unfortunately running an fx 6300, i plan on upgrading to probably the 7600k though, reckon a 1070 paired with a 6300 is a bad idea? ill be playing at 1440p

>>59961812
The issue is not power delivery. The issue is that voltage does not scale linearly with frequency. As you increase the frequency that you're running a processor, the voltage required increases exponentially and it becomes increasingly difficult to maintain the same voltage across the chip. This isn't related to power consumption, it's related to frequency and voltage. AMD's Fiji chips have no issue drawing that much power because they're designed to do it; they keep relatively low clocks in the 1GHz range so that voltage is easier to control. For reference, the highest clocked GCN chips we've seen are the current RX 580s, and those don't really go much farther than 1500Mhz. Since desktop CPUs can't benefit from parallelism the way GPUs do, they don't sacrifice clockspeed in order to operate thousands of cores. A desktop CPU is designed from the ground up to operate at the highest clockspeed it possibly can while generating a manageable amount of heat and staying very stable. The 7700k is tapped out, and since a quadcore is barely stable at 5GHz, not because of power usage, but because of voltage management, we know that adding two more cores into the mix at the same frequency would be impossible. Things might change with a node shrink, but based on all the issues we're hearing about from the Intel camp regarding node shrinks, I don't expect much.

>>59962027
Don't become a corelet, get a r5.

Do you think my GTX 1070 will last me another 4 to 5 years assuming I stick with 1080P?

>>59962103
which r5? ill be honest i havent got the biggest budget and wasnt very interested by ryzen but heres your chance to change my mind, why would it be a good decision to choose an r5 over say the 7600k

>>59960399
To an extent? I buy AMD products because I consistently have fewer issues with them, because I want to support competition and because AMD legitimately seem like the less shitty company with all their open initiatives (CUDA vs. OpenCL, FreeSync vs. G-Sync, GPUOpen vs. GameWorks, Mantle, etc.).
I still buy NVIDIA and Intel products from time to time when it makes sense, but I try not to.

>>59961784
>VR
>widespread adoption

We'll only see that when celebrities start releasing virtual sex slave games.

>>59962131
Nvidias are not made to last

>>59962194
Well it is my first. I had all AMD cards before this one.

>>59962137
The R5 1600 is the best price/performance AFAIK. The benefits are that it has far more cores and threads than anything Intel offers in the same price bracket, meaning it's significantly better at multi-tasking while not being significantly worse in single threaded workloads.

The AM4 platform (the motherboard basically) it's also scheduled to last at least until 2020, unlike Intel platforms, which only last two years usually.

>>59962217
on my home pc i dont really often do anything to do with 3d work, the most stress i usually put on my cpu is a bit of video editing and photoshop, a 6600k was probably a bit overkill for this but i hear its a solid choice that definately wont bottleneck a 1070, i wont run into those issues with an r5 1600 right? also is the 1600 still a good choice considering my light cpu loads?

>>59962248
I can't see any reason why the R5 1600 would bottleneck that. I mean sure, compared to a 7600K or something similar it will, but it's only visible in the ultra high frame rates. It isn't going to inhibit your ability to comfortably play anything. To use an example, in DX11 Far Cry Primal 1080p, the R5 1600 loses 14 fps compared to the 7600K, but the drop is from 108 to 94.

Also it's widely believed that games will continue the trend we've seen over the last number of years and continue to use more and more cores (particularly with the introduction of the Vulkan/DirectX 12 APIs that make it easier than ever before) making the Ryzen chip the smarter long term purchase.

As for Photoshop and video editing. Video editing should benefit from the increased core count. Photoshop however underperforms on Ryzen because it's pretty single threaded, I believe the same goes for similar Adobe products.

If your CPU loads really are very light, you might want to go for something even cheaper than either the 7600K or 1600. There's the R5 1400 but I haven't heard glowing things about it. Tons of options on Intel's side of course, right down to Pentiums. All in all, don't worry about CPU too much for frame rates, if you need any proof that it really doesn't matter that much, here's some Intel Skylake Pentiums benchmarked and managing to run games pretty competently when paired with a powerful GPU.

http://www.eurogamer.net/articles/digitalfoundry-2017-intel-pentium-g4560-budget-cpu-king-review

I hope that the Vega card will be the new 290X. I have a FreeSync monitor and I like AMD. I think it's impressive how the 290x still holds up. If the Vega is the new 290x it will be fantastic

So the important questions stands,


How do you pronounce '2080'?

Twenty-Eight? or Twenty hundred-Eighty?

>>59962621
Since when does 20 come immediately after 10? This is going to be the 11xx series, dummy.

>>59962621
Twenty Eighty

>>59962645
10 and 20, Cancel out 0. 1 and 2. Nignog.

>>59962675
But I was absolutely ready for the 'Crank it up to 11' marketing campaigns.

>>59960399
Owned shit from intel, amd and nvidia, aside from shit drivers from old amd (ati) never had an issue with either. I don't give a shit about brand loyalty, that would be fucked up. At the time I want/need to get hardware I will get whatever I feel it will last me, will perform best for my use case and won't shit the bed.
All that said, it's healthy all the companies tear each other apart , almost to the point of killing eavh other.

File: asian_boobs.jpg (235KB, 681x1024px) Image search: [Google]

235KB, 681x1024px

sage Intel/Nvidia shills threads, they're desperate because the mid range market is 100% AMD right now, see:
>>59960090
>>59960090
>>59960090

> By launching sooner than expected they will be able to increase prices of single cards and increase profits
>able to increase prices
>increase prices

This will be the only truth of that "news"

>>59962711
kill yourself poojeet

>>59960559
citation needed
that's fucking stupid and 11x0 is better

File: 1484927825055.png (440KB, 645x1260px) Image search: [Google]

440KB, 645x1260px

rip AMD

>>59960504
>implying it's not going to be a repeat of GP100 paper launch

File: paper-airplane[1].png (498KB, 638x651px) Image search: [Google]

498KB, 638x651px

>>59963077
paper launch THIS

>>59961048
>>59961062

It wasn't a graph but a live demo of an engineering sample on Doom DX12. Happened back in January or December I think. The thing was taped up to hide info on it, most likely throttled the card. Had debugging silicon on it too. I dont think they would have clocked it high either.
I think Vega will be good, in theory. AMD is going to have the same problem as always, pushing new stuff no one is using yet. I mean, its great for innovation but waiting for months to get top performance out of your card is frustrating.

>>59960283
>volta this soon
>'''''changed plans'''''
You faggots will believe anything.

File: 1490302558198.jpg (54KB, 305x309px) Image search: [Google]

54KB, 305x309px

>>59960283
Anyone that believes this rumor should kill themselves.
It's way too unbelievable.

>16GB of HBM2
lmfao.
Just like how the 1080Ti was rumored to come out in January 2017 with 8GB of HBM2.

Volta won't come until Q2 2018.
If it comes sooner, it's because Nvidia will gut lots of features they were working on to get it out sooner like AMD did with Fiji. And there won't be a Titan Xv with 16GB of HBM2

>>59960283
what a timing! such precise!

>>59960407
Too bad their cooling sucks.

>>59960283
so any leads on the naming schema?

>>59964617
it will be paper launch same way instinct is selling now and is vega

or tesla with HBM2 which you can't fucking buy, only order, even now YEAR later

Honestly I think Vega will be really good even if Volta is coming out this year (which I doubt). Too bad I already paid the g-sync tax on both my monitors so I'm stuck with this company.

>>59960283
should I start saving up cash for 2080, it is gonna be expensive r-right

>>59963151
It also was used Fiji drivers senpai

>GTX 2070 with 11GB of GDDR5X @ 12Gbps
>GTX 2080 with 11GB of GDDR5X @ 14Gbps
It's Pascal rebrands. Only the Titan will be new.
Stop falling for clickbait

>>59964647
>or tesla with HBM2 which you can't fucking buy, only order, even now YEAR later

huh? you can buy both the tesla p100 or quadro p100 right now provided you have thousands of dollars to waste. they've been available for ~6 months now.

https://www.google.com/shopping/product/12809297875429637377

>>59965897
Can I remind you that the p100 was announced in March 2016?

oh boy, its time to replace the old Titan Xp again. i´m super hyped!

>>59960372
>>59960979
I'm not telling you to go buy it, but you need to realize "just wait for X" is a meme. There will always be something just on the horizon that is just enough of an upgrade to make you debate about it. The only time where you should definitely wait is when they have the shit dated.

>>59965099
More rebrands? I guess that's how it usually goes. One of them does their rebrands and then the other follows.

Is it too much to ask for a mini price war before next gen comes?

File: avatar54079_1.gif (55KB, 150x150px) Image search: [Google]

55KB, 150x150px

>>59960407
I've owned it all. Intel cpu and amd gpu has been my main thing for about 6 years. But to be honest I'm looking at getting a ryzen 5 16x with a nvidia gpu for my next build. Which is fucking bizarre to me. Though I won't make a new complete build till CyberPunk 2077 comes out.

And to be honest I mostly just use my i5-25k linux machine, since all I mostly do is read and watch videos online nowadays.

File: pairings.jpg (2MB, 3840x1032px) Image search: [Google]

2MB, 3840x1032px

>>59967238
Nvidia drivers are a fucking mess on Ryzen right now.

Just wait for Vega. Worst case scenario is that Nvidia gets their drivers fixed by then (unlikely since their DX12 is still shit), and the 1080 or 1080Ti also drop in price.

>>59967443
>Nvidia drivers are a fucking mess on Ryzen right now.

no they're not. only the titan seems to still be affected. the 1070 and 1080 are fine.

>>59960399
Not really, I am pretty sure everyone checks out all offers when they have to make a purchase.

I did when I bought my 1060. It ate less, performed better, was cheaper. And had Gameworks + Physx. So I went with that. (Dx12/Vulkan is there, yeah, but no games use them.)

Then there is the CPU upgrade now. Ryzen is a good bang for bucks if your game uses multiple cores. But do they use multiple cores? Hell fucking no.
Once again, i5. I was hyped though.

>>59968099
Unless you're using a 1080ti or a Titan, Ryzen will perform just as well as any Intel chip in games. Congratulations, you played yourself.

>>59960283
Finally, a good affordable* Nvidia card

*for only $1999

>>59967595
>>59967443
>on ryzen
on everything

>>59968099
>how dare you criticize my cuckolding fantasies, i love my i5 and 1060

>>59968099
>Once again, i5.
That's not true though. i5 is objectively the worst CPU purchase right now. You either buy one of those unlocked Pentiums, an R5/R7 or a 7700k if all you do is gaming. Everything else in between is objectively worse.

>>59960283
So there plan is to flood the market with high performance cards at high cost since amd cant compete? Thats actually briliant.

>>59960608
no it doesnt so good 1180 is far to much numbers
2080 is easier on the eyes and easier to understand, not to mention 1180 looks like slighty upgrade on 1080, while infact its major architecture change
>tldr 2080 looks better and makes more sense than 1180

So when can I expect Vega and when can I expect Volta. I plan to build around June

>>59960984
pascal and polaris are basically just minor architecture changes, the real changes comes with volta and vega, originaly nvidia didnt even plan to release pascal, but they probably improved maxwell efficienty to the point that instead of calling it maxwell refresh, they renamed it pascal, just amazing nvidia engineers work

File: 1468363153137.jpg (20KB, 552x460px) Image search: [Google]

20KB, 552x460px

>>59960283
ryzen AMD fallen

>>59963633
who cares about hbm2, its just another gimmick, just like hbc. Is amd srs? using hdd or ssd as cache or even ddr? are they srs? its like 50times slower and they pretend it wont slow down your game? Are they serious?
>Hey we put only 4gb of hbm2 on your chip but dont worry, you can use full power of "VEGA" chip by caching your hdd and using it for vram, its only like 100x slower, but that 1-10sec lagspikes will be hardly noticible

>>59970718
r u srs? mad btfo srs

>>59970718
you have to be 18 or older to post here

>>59960283
Vega-mite btfo befor it even had a chance for release.

File: image.jpg (183KB, 1680x935px) Image search: [Google]

183KB, 1680x935px

>>59960882
Read the last point on the article
>THE LATEST AND GREATEST GOYIM
>WE BROUGHT YOU THIS CARD FROM THE FUTURE, AND IT WASNT CHEAP!
Fuck competitive purchasing

>>59960432
It will release before July-september. So basicly build in may with Vega or wait till fall for Volta.

>>59970812
>volta
>2017
you know this isnt going to happen pajeet

>>59960984
Because their current stuff cannot reliably push 4k at over 60hz on max settings. They needed to shit out incremental udates to recoup costs and fund rnd.

>>59962027
Why would you even conifer buying such an old cpu when 1600 and 1700 are a thing?

>>59962131
Basicly if you are not upgrading to 4k there is no real reason to upgrade your gpu this gen, or likely the next couple. Most of the gpu changes will be about bringing down the cost of 4k and upping 4k rps.

>>59963633
this

File: 1491062981332.png (130KB, 1597x634px) Image search: [Google]

130KB, 1597x634px

>pajeets shilling pajeets shilling pajeets shilling rumour bait
gg plebs, eternally retarded

>>59960283
amd: release superior products at reasonable prices
(((intel/nvidia))): screech autistically at goyim while rushing out inferior products at ever increasing prices

itll be funny when the 480/580 are benching higher than a 1070 in a year from now since finewine. lmao

>>59960283
I doubt it, it would kill XP sales

>>59971401
The 480 and 480v2 aren't going to outbench a midrange card since they can barely keep up with the 1060...

The only card that goes toe-to-toe with the 1080 is the Fury X. AMD literally has no answer to the 1080 Ti, Titan X Pascal, and Titan XP.

>>59960399
I've owned GPUs and CPUs from Intel, AMD, and Nvidia. Got lucky and haven't had any serious issues so far with any of their products because every time I bought a new component it was already released for a good while with plenty of reviews.
Tempted to upgrade from a 980 to Vega for 4K 60fps and squeeze more quality out of the HTC Vive through supersampling + AA.
The 1080Ti is already out there but considering how Nvidia is reacting by releasing the TitanXP and rushing to get volta out the door, I'd rather wait and see what becomes of Vega which is around the corner.

What I am mostly worried about potential hardware faults aside is the cost and availability of after-market models since it will be using HBM2 and likely be produced in small quantities raising the price significantly.
If that is the case I would wait even longer on reviews of Nvidia products especially after the 970 debacle, or wait on AMD to release a similarly performing 90 series card the following year.
Not like I am dying to upgrade 980 still holding fine, otherwise I would have bought a 1080Ti by now.

>>59971722
1060 is mid range you dumbass

>>59967595
I like how you can't clearly see the framegraphs posted and you just lie.

It's on all Nvidia cards. The Titan just makes the problem more clear.

>>59962098
>voltage required increases exponentially
That hasnt been true for years now.
Due to modern litho processes needing extremely tight tolerances, and due to practically requiring ICs to have rigorous power control mechanisms to extract as much efficiency as possible, voltage levels very nearly follow linear curves with clock speed. What changes (quadratically not exponentially) is only the power draw. Voltage required only falls out of the nearly linear curve when reaching design limits, for instance an 1800x needing 1.3v at 3.8Ghz but 1.425 at 4.1, or the 7700k getting away with 1.25 at stock boost but needing 1.5 to break past 5.2Ghz

>>59972188
The 1070 has been mid range since the 1080 Ti came out.

Keep up with the times.

The 1060 is NVIDIA's low tier card for literal toaster PCs.

File: enough.png (122KB, 398x387px) Image search: [Google]

122KB, 398x387px

>this much bullying from nvidia
cut amd some slack

>>59973605
"Midrange" is a price, not a performance level.

Midrange is what >50% of people are willing to buy. $350-400$ is not midrange.

>>59973624
amdogs must be put down

>>59960283
Well FUCK, if Chinese website "MyDrivers" has confirmed it...

>>59960399
I basically switched them with each new card. Last one I had was a GTS 450 which served me well for almost 5 years. However, I got annoyed and irritated by the stupid GAMER bullshit they pulled with their control center. It's just one of those things you just know are there that annoy you whenever you think of it.
Right now I have a R9 390 and it works well. And since I rarely play games now anyways and will probably give up new games in a few years altogether, this might well be my last graphics card ever.

But to answer your question: I have no brand loyalty, usually go for the best one, but nvidia is on my boycott list for that horrible pandering.

File: 1492491618074.png (94KB, 1302x1301px) Image search: [Google]

94KB, 1302x1301px

>>59960283
"I think"
Discarded

File: nvidia milk.png (226KB, 600x905px) Image search: [Google]

226KB, 600x905px

>>59960399
I used to just buy the best.
I had the 9700pro, then the 7800gt, then a 9600gtx after that abruptly died.

But Nvidia's horrible drivers and horrible anti-consumer practices have made me boycott them since 2010ish.
I won't pay more for an inferior AMD if that was the option, but I'd wait for AMD to be offering something competitive or better like hopefully Vega before buying from them.

One of the biggest problems with them is that they sell $150-$350 GPUs with planned obsolescence built into them. I can understand that on $75 GPUs or some shit, but not $350 ones to not be able to play games on decent settings just 1-3 years later.
And they price gouge at any opportunity. And their rebranding was far worse than what AMD does. They would rebrand 5+ year old cards that are missing so many features.

>>59973846
>770 is that low compared to 280x which is 7970
JUST

>>59973846
>used to buy the best
>7800
>9600
??
not even best in class

File: perfrel_2560_1440(2).png (43KB, 500x1210px) Image search: [Google]

43KB, 500x1210px

>>59973876
The real shocker is the 780

>>59973924
>keklers this low
JUST

>>59973924
>(((1060))) 106%

totally non-skewed due to a careful selection of goyworks(TM) titles!

>>59964553
? I always thought their cooling was good. Who has the best cooling then?

File: perfrel_2560_1440(3).png (35KB, 500x730px) Image search: [Google]

35KB, 500x730px

>>59974026
That was a review from July last year. After that they cleared the older cards out of the graph.

>>59973903
They were the best value at the time that's still very good performance.
Actually I'm not totally sure about the 7800. But when I bought it I was a bit late to the 6600GT and 6800 party. And I thought "hey this 7800 is probably good too now that it's out".
I was young then, but I imagine it was still a decent purchase at the time, unlike those that bought 9800s, 270s, 280s, etc.

>>59960372
"Is it better to wait for a 1080ti?"
I'm sorry but you fuckin piece of shit. I play motherfuckin GTAV on modded lower-than normal settings at like, 480p, just so that I can play it with high/consistent framerates on an AMD piledriver-era APU. Shit, there's people that try to do stuff on Intel HD Mobile processors....

Take your high end graphics card and just be happy. It'll last you a decade, guaranteed. All you need is like, 720p, at 40 fps. No way a 1080 fails to do that for at least a decade.

>>59975389
>he doesn't own a nice monitor

>>59975417
Dude man fuck that shit. Rich ass people cant appreciate jack fuckin shit. Complaining about their monitor...

I put my entire piece of shit build together for practically nothing, and I remember seeing rich ass people complain about how their 970's only having 3.5 gigs of vram. I just.....I was stunned. Like, I actually have to activate the feature on the mobo that uses normal ram as backup VRAM, because my apu is actual dogshit. I'd love 3.5 gigs of vram.

God bless those poor souls on like, Intel Atom processors with Intel HD graphics, not even a number. That shit must fuckin suck.

File: toaster.png (90KB, 721x530px) Image search: [Google]

90KB, 721x530px

>>59975389
>>59975482
Mah nigga

>tfw toaster master race

>>59960399
I have owned both, but my current big GPU is AMD and I'd like my next one to be because I need hyper-v enlightenments for my games to work properly in my virtual machine setup. NVidia GeForce drivers detect these and self-destruct to coerce you into buying Quadro.

File: Screen Shot 2017-04-19 at 6.18.51 PM.png (906KB, 3360x2100px) Image search: [Google]

906KB, 3360x2100px

>>59975993
>NVidia GeForce drivers detect these and self-destruct to coerce you into buying Quadro.
It can be easily worked around

>>59976097
With hyper-V enlightenments? Post method.

File: 1484686763369.gif (36KB, 265x200px) Image search: [Google]

36KB, 265x200px

>>59976097
>proprietary drivers

>proprietary because they don't want you see the dirty spaghetti code tricks they use to make their graphics cards run shit on AMD CPUs.

>>59976110
>With hyper-V enlightenments? Post method.
It clearly says VMware for the 2nd video adapter. Add hypervisor.cpuid.v0 = false to the VMX file.

>>59976116
>>proprietary because they don't want you see the dirty spaghetti code tricks they use to make their graphics cards run shit on AMD CPUs.
Pajeet, you grammar is horrible and this statement makes no sense.

>>59976165
>Add hypervisor.cpuid.v0 = false to the VMX file.

Hyper-V won't do its thing if it can't detect it's in a VM. I've tried a similar workaround with QEMU, measurable loss in performance.

>>59976165
>leaft out a single word
>omg terrible grammar must be pajeet! xd
Go away, Ramesh.

>>59976165
>>59976216
>Hyper-V won't do its thing if it can't detect it's in a VM
wut? Why would you run a VM inside a hypervisor inside a hypervisor?

>I've tried a similar workaround with QEMU, measurable loss in performance.
Works fine for me. If you're going to try it with ESXi set latency sensitivity to high and 100% CPU reservations. The only problems i've had has been with SLI. In full screen the screen would have a purple strobe, changing it to full screen borderless window fixed it though.

>>59976220
Keep on pretending it is somehow Nvidia's fault that AMD's cpus are shit.

>>59976258
>wut? Why would you run a VM inside a hypervisor inside a hypervisor?

You don't, windows has some optimizations for paravirtualization which it only enables when it detects it is a guest in a VM. These are called enlightenments and involve things like fixing the high precision event timer, which improves GPU performance. Obviously this is designed for when windows is being run on top of HyperV, but support for them can also be enabled on other hypervisors.

If you hide the use of a hypervisor to satisfy the GeForce drivers, these optimizations are not enabled on the guest and there is performance drops. It varies depending on the application however due to what hardware timers they use, ect.

>>59960283

Fake. Now GTFO!

File: Screen Shot 2017-04-19 at 6.56.05 PM.png (98KB, 1868x332px) Image search: [Google]

98KB, 1868x332px

>>59976409
>high precision event timer
HPET is virtualized in ESXi

http://www.vmware.com/content/dam/digitalmarketing/vmware/en/pdf/techpaper/Timekeeping-In-VirtualMachines.pdf

File: Screen Shot 2017-04-19 at 7.01.19 PM.png (299KB, 1074x1424px) Image search: [Google]

299KB, 1074x1424px

>>59976409
>>59976550
And it seems that "enlightenment" is just Microsoft's term for calling its paravirtualization drivers which would be analogous to VMware Tools.

https://globaljournals.org/GJCST_Volume13/4-A-Performance-Comparison.pdf

>>59960559
Bullshit. Nvidia calls Pascal the "10 series", why the fuck would they go from 10 to 20?