>literally 100s of thread about "got new card today"
>spent $1000 for their new card
>Pascal shipping first months of 2016
>10x the current gpu performance
kek
why the fuck are you not waiting? you are really desperate for a new card urgently (sure..) or just money to throw away?
cause people are dumb.
this is too complicated anyways just buy 980ti its the best
>a fucking convolution kernel ASIC
>>51640313
I'm guessing the successor to GM204 launches early 2016 with 8gb 12ghz GDDR5X vram, 35-40% faster than Titan X at 4K for $599-649. I'm also guessing there will be two cut additional SKU's based off the same chip for $499-529 and $399-449.
Buying a new card today is just throwing away money.
>>51640313
>10x the current gpu performance
>actually falling for this
you're not wrong about it being a dumb time to buy but even nvidia admitted that was based on "very rough estimates"
also, they said this 10x faster estimate was relating to "deep learning apps," whatever the fuck that means
>>51640313
>10x
I assume nobody actually believes that
>>51640313
>first months of Pascal
We still don't know don't we? Could easily be Q2+
>10x the current gpu performance
>10x the price
>mfw
>>51640927
>3.5k for the GTX 1070
>>51640313
>10x
Not on gayming mi familia
How do I overclock my 290x if it won't let me change the voltage
Also I have 3570k but have no clue had to OC processors
>got my first job recently
>actually excited for new tech and not the dread of my current hardware being outdated
Feels good
It's going to be just like the people that bought 980s and didn't wait for Big Maxwell. I'm going to laugh just as hard.
Are 4k / 1440p monitors worth it
I think that's the only reason for buying Pascal series
Anything seems overkill at this point
>>51640313
Pascal is not actually going to be 10x better the Maxwell. You really think I am going to go from 120fps in BF4 to 1200 fps? I wish.
>>51641710
yeah fps is only gpu depended right?
>>51641753
depends on the game, but mostly, yes.
>>51641983
AHHAHAHHAHAHAH
>>51640313
>>10x the current gpu performance
Are you one of those people who thought the PS2 could render 10,000,000 polygons per second?
maxwell was 10x the scientist that pascal was
>>51642004
you showed him
>>51642004
You realize that in most games, the GPU is the bottleneck, right...?
The only exceptions are MMOs and HL2 engine games. Pretty much everything else is only bottlenecked by the GPU and not the CPU.
Educate yourself, retard
Things will move along as they always have. GTX1080 for $599, 30% faster than 980ti. GTX1070 for $399, identical performance to 980ti. Eventual Titan Y for $1k, 30% faster than GTX1080. Eventual GTX1080Ti for $699, 10% slower than Titan Y.
>>51640882
>>51642017
>>51642128
you guys fucking hell
The 10x includes 2x from using 2 GPUs enabled by better interconnects and dropping to lower fp16 precision for calculations.
This 10x is for compute. Read the fucking reports you cunts.
but you guys only playing fucking games and worry about FPS only.
>>51641710
>You really think I am going to go from 120fps in BF4 to 1200 fps?
you really don't know shit about technology. you really are so ignorant.
>>51642211
>>51642211
this
10 times the performance is for compute, FP16, or "half-precision floating point" calculations.
>>51641710
>10x the performance = 10x FPS
wow
go back playing fucking BF4
question: why would nvidia release a 10x more powerful card if they could just release a 20% better card every year and sell more that way
>>51640313
That is fucking HALF PRECISION! It's for specialized machine-learning applications, not general-purpose graphics.
How much of a fuck up will I be if i brought a 970/390 for 500 bucks?
Or should i buy a 380 now then resell/ return? in 2 years for the new tech of gpus
>>51642401
because it is a competitive market?
you understand they don't have a monopoly on GPUs
>>51642435
I never said it is for general-purpose graphics but they will be definitely at least 40% better then the cards today.
just remember
1.7 yields
300 series
pork shoulders
woodscrews
LITERAL SUICIDE DRIVERS
killing panels
>>51642476
They're close, so unless AMD pulls something out of their ass that blows NVidia's offerings at the time completely out of the water, they can basically do whatever the fuck they want.
Also they have a lot more money to spend on advertising than AMD, so pretty much every major tech youtuber is an NVidia shill.
>>51641688
>Anything seems overkill at this point
Depends on what you're doing. I mostly use my GPU for compute. Machine learning, mainly, but also physics simulations and fractal rendering. All of those will take as much hardware as you throw at the problem. Give me a $50,000 cluster and I'll have it running full tilt 24/7.
But for most gamers, there isn't that much point. You rapidly run into diminishing returns.
>>51642476
Just like Intel doesn't have a monopoly on CPUs amirite?
>>51642201
i feel like they'll change up the naming system. it seems weird releasing a card designed for 1440p/4k called the 1080
>>51642515
>they have a lot more money to spend on advertising than AMD
i don't know where you take this informations from.
actually amd revenue > nvidia revenue
>>51642533
AMD is not real competition. You would not put an AMD CPU in a high-end gaming rig.
People would absolutely put an AMD GPU in a high-end gaming rig. Nvidia is more popular, but AMD is a strong competitor.
>>51642508
>Suicide drivers
>Crimson
Sure is selective memory in here
>>51642580
>literally only affected retards that can't into checking settings before stressing their GPU
>literally fixed the next day
>>51642550
kek, did you never even take a business class? revenue isn't everything. Green's business focus is a lot tighter than AMD's, and yet is still at about 80% of the revenue (and $800m more profitable).
So yes, green has a lot more money to spend on ads.
>>51640313
>first months of 2016
I wish
>>51642453
Arent those like $300? Im trying to figure out which of the two to buy
currently building a new rig, with the 6600k as my cpu. still havent decided on a vid card though. so i should just wait?
is there a 'cheap' card i can by around 100 thatll run csgo perfectly fine?
>>51640313
>first months of 2016
[citation needed]
>>51642618
random kid from the internet pretends to know everything about the spending of $7/8 billions company.
yes sure kid. you took business class thats why you cant afford leaving your parents basement
>>51642646
csgo is very cpu depended. if i were you i would buy a second hand card and then upgrade it
>>51642597
>it's okay when AMD does it
>>51642646
Any modern $100 card will be fine.
Even your iGPU might be fine for CSGO. CSGO is quite light.
I'll be buying a fury X in about 3 weeks. Both to support competition to nvidia and because I ain't waiting 6 fucking months to get a new card.
I'll ask here because graphics thread so close enough
Bought a R9 290 off Gumtree a year ago, dirt cheap. Worked fine right until last week when it died. Fans spin, no graphics output on any port.
Don't have the receipt so Gigabyte won't help. What do?
>>51640349
>slower than fury x
>slower than Titan x
no thanks.
Planning a whole new rig, what parts should I hold off on getting until we get a better idea of pascal GPUs specs?
I was thinking:
Mobo (stupid proprietary shit maybe)
CPU?
PSU
>>51642736
What games are you looking to play that's coming out today that can't be maxed out by your stuff right now?
Or are you retarded enough to enjoy sub 40fps 4K gaming?
>>51642693
>>51642718
thanks mates. any suggestions for gpus around 100? i have no idea about gpus that'll run csgo finre
>>51642753
shove it in the oven
google graphic card oven fix
>>51642808
again CSGO is very CPU depended. you can use a shitty card but with a good CPU and run high fps
get a 280x second hand
>>51642825
thanks again mate. you the best
>>51642840
280x is a very good card
>>51642566
>high end gaming rig
you mean sli and 4k screen?
>>51640313
>>10x the current gpu performance
>kek
>why the fuck are you not waiting? you are really desperate for a new card urgently (sure..) or just money to throw away?
got proof of this?
i'd love 10x the double-precision performance of current gen but guess what
that aint fucking happening.
>>51641294
And those people that bought Pascal but didn't wait for Big Pascal, and the people that bought Big Pascal and didn't wait for HUGE PASCAL!
Why fucking upgrade? The next card is just 6 months away!
>>51642811
Some really nice dude on a forum is scanning his receipt for me, if that doesn't work I'm gonna play Auschwitz simulator with it.
>>51642927
read the thread
go watch nvidia announcement
go read reports
use fucking google
>>51642928
>Not waiting for Mega Massive Big Band Pascal, HD Turbo Remix
Scrubs
>>51642766
pascal wont have some crazy high power consumption youll be fine with a 650W or whatever
>>51642973
It's more like 5-10% at best in actual real world performance. HBM is a huge power draw and heat increase for fucking nothing since there's nothing to really take advantage of it yet.
Most of Pascal is workstation shit. Are you getting nvlink for your workstation? For your gayman rig? No? Then enjoy your bottleneck and usual bi-annual performance jump.
>>51643021
wat
>>51640313
>Waiting
Not everyone is an AMD poorfag. If there was a known release date and known price points, I'm sure some people would be willing to wait a month or so tops. But for those of use building new machines now, we'll be fine. 980 ti for gaming is going to stand up just fine to Pascal X80 for a good amount of time. Theres no sense for 900 series owners to get first gen pascal unless they have a GTX 950 or less.
>>51640944
ayyyyy
Is the 28nm meme truly dead?
>>51642597
fucking apologist, tell us all how did you manage to type that with so many cocks from AMD trannys in your mouth and ass?
>half precision computing
>not full precision computing
You niggas love being keked
>>51643058
Go read about nvidia's press release demonstration about Pascal and Volta. They said that they're having trouble keeping power draw under acceptable levels because of HBM, and that PCIe x16 would bottleneck it.
The consumer-grade Pascal GPUs are going to be on PCIe x16 and will have major bottlenecking. They need nvlink to get the most out of the new cards, and those will be available for their workstation cards.
For the regular consumer it's going to be the same shit, different year. If you have a 980 Ti now, it's going to still be viable for a long time, and it's going to be a while before PC games start demanding HBM since there's a huge market share out there consisting of non-HBM cards. In 5 years, it'll be completely different story, but for now you're not going to be missing out on much. Just look at how long it took to make 1 gig VRAM cards obsolete. If you've got 4-6GB, you're fine for a while, unless PC developers decide to stop trying to code efficiently and PS4 ports suddenly need 100x the specs to run properly on PC.
Consider how much AMD put stock into their own HBM offerings and how they're not the messiah cards they were hyped up to be. Even Maxwell was touted as such compared to it's predecessor, and when 780 Ti's are allowed to be compared on benchmarks, they still compete.
>>51643001
>not waiting for Waitdeon WaitD Waithundred X-remewait edition
>nvidia slides
These are reliable as bathroom graffiti
>>51643099
4K on the 980Ti isn't there yet, better to wait for Pascal, the price of 4K monitors will have dropped by then too, it's not about being poor or not, it's about being smart with your investment
>>51643716
980 Ti sucks at 4k if you're on reference. Many games it can do >40 on non-reference cards. Older games it will absolutely smash them on 4k.
4k's also as gimmicky as it gets. It will do absolutely everything other game gaming at that res, and 1440p/2k is going to work very well.
>>51643385
the 1gb hd 5870 still handles new games fine.
75-85 fps battlefield 4 for me(high, not ultra) , and YouTube videos of other games look quite smooth too.
http://www.gputechconf.com/
APRIL 4-7, 2016
SILICON VALLEY
THANK YOU BASED NVIDIA
>>51642804
not him, but my main monitor is 1680x1050, and I can't justify a fury without a 2k monitor
I also can't justify a 2k monitor without a fury
current card is a 660ti and it just werks, even with a missing capacitor, I'm waiting for it to die before I replace it
>>51644123
And there you go. When 1gb VRAM cards can still compete, we're not facing some kind of performance wall. This shit will never happen. It's bad for business.
>>51644140
If you think that Nvidia are going to be launching new products at GTC, you're in for a real disappointment. Even if they were planning for a launch around that time (which they're not, it'll be later), that's not where they'd be doing it, because that's not what GTC is for.
>>51644222
Nvidia launched TITAN X and Quadro M6000 at GTC 2015
http://www.anandtech.com/show/9096/nvidia-announces-quadro-m6000-quadro-vca-2015
Don't post if you don't know SHIT
>>51644239
Yet 980 Ti trounces the shit outta the Titan X.
>>51644310
What has that got to do with anything?
Nvidia will be demostrating Pascal to customers at GTC 2016 and will launch it there
You were proven wrong, now stop embarassing yourself with your FULL RETARD cluelessness
>>51640313
>>51640349
>actually believing "10x" performance
Wow, you guys are retarded.
>>51644345
>>51644310
>>51644239
>>51644222
AWH SHIT YALL NERD FIGHT OVER SOMETHING ARBITRARY
>>51642211
Even that is doubtful.
>>51644345
You save $400 and lose nothing in real world performance.
>>51644185
from personal experience in battlefield 4.
gtx 260: 40-50 fps with occasional drops (medium)
hd 5770: 45-55 fps with occasional drops (medium)
hd 5870: 75-85 fps with occasional drops (high)
hd 6870: crashed on high. on medium 70-80 fps.
gtx 650: 40-50 fps medium. rare drops.
gtx 680 4gb: 75-90 fps, no drops, ultra.
i believe those stutters(frame drops) were because of vram bottlenecks, or a weak gpu for the gtx 650, since they mainly happened during scenes with explosions and fire. the game uses what vram it can, but with older cards sometimes it's not enough causing it to bottleneck. that's my opinion at least.
tested at 1600x1200 with an i5 2500 and 8gb ram.
>>51642880
I'd like to see the number of 980 ti sli machines that exist of there using 8350s.
>>51644239
Titan X was announced at GDC you goddamn retard, not GTC. There is literally no chance of them unveiling new consumer hardware at GTC.
>>51644432
>i believe those stutters(frame drops) were because of vram bottlenecks
No.
>they mainly happened during scenes with explosions and fire
And this is why. This shit hardly uses a lot of VRAM. You spend far more VRAM on textures and shit. These are a result of some shader compilation taking place too late, which is either a driver problem or simply that your system is overloaded.
>>51644490
Launched != Announced
Keep digging that hole though
>>51641294
980 ti was 10 months later.
https://www.youtube.com/watch?v=xQhb3C2hQoE
5:33, CEO Jen-Hsun launching the TITAN X at GTC 2015
Stop embarassing yourselves if you don't know SHIT
it's gonna be the same gains as every year you morons. pascal isn't magical. it will be the skylake of gpu's.
>>51644494
thanks for correcting me. I'm not an expert, was only an opinion, but now that's good to know. I appreciate it.
>>51644488
fx 8350 might be faster at 4k, but most people still won't buy it because at 1080p Intel is faster and they don't think to look up benchmarks.
so many people still buy the gtx 960 when the 380/x is better.
>>51644432
And this is why I'm not worried about Pascal.
Consider how long they've had the Ti line for nvidia cards, and how they often are omitted from benchmarks. There was a huge dramatic fallout because a reviewer was honest about AMD. These companies pay for a lack of objectivism by offering goodies. 680 Ti,. 780 Ti, high end 7xxx series, 290X so on, and so forth. These cards are still competitive. The highest-end cards from earlier generations get ignored, and they do this because they stay relevant for a long time.
Nobody is going to upgrade every year if they know their hardware is still worthwhile for 3-5 years, especially if they sink lots of money into the high-end hardware.
>>51644364
This.
>>51644488
>machines that exist of there using 8350s.
Can you even decide to even the look more like?
I got a 970 for $240. I don't want to pay $350-400 for the equivalent when Pascal releases. I also don't want to wait. I also seriously doubt the performance will be 10x as good.
I don't use 4k so I don't really see why I would need Pascal.
10X is for deep learning
Pascal is built on 16nm FF+ process, 2x the density of 28nm, much more power efficient because of FINFET transistors and micro architecture improvements will increase performance and power efficiency significantly.
>>51644586
You must be short.
Because this shit goes over your heads.,
>>51642928
I'm holding out for Ginormous Johnson. Gonna overclock my 55" 8K monitor up to 288 Hz FEEL that glorious radiation peeling the skin off my face while playing Fallout VII: the Chernobyl Zombie Hordes.
I'm waiting to get a magazine with bonus POWER9 + Maxwell with NVLink. All joking aside, I'll pick this shit up super cheap at my local Best Buy early next year. I need a new gaming machine.
>believing a GPU manufacturer's timelines
Both AMD and nvidia have an awful track record on timing their releases. You'll be lucky if it's Q2 2016
>>51643385
>The consumer-grade Pascal GPUs are going to be on PCIe x16 and will have major bottlenecking.
PFFFTTTHAHAHAHAHAH
I just bought a 960 SSC for 150 dollars (with shipping) just before black dickday. Am I retarded?
>>51644681
Nah.
I'd just rather bicker about other autistic shit.
>>51642643
390 unless u like physx and hair works, their dead even in the majority of AAA games
>>51644839
No, considering that 150 dollers wont buy you a pascal card at launch anyways.
>>51640313
It's like OP finally realized that the entire point of Black Friday is to sell off old stock to poor morons in preparation for new stock that they'll sell to rich morons.
>>51642453
>How much of a fuck up will I be if i brought a 970/390 for 500 bucks?
>$500
>970
Fucking what? Where do you live?
If you have a large PSU, go for the 390.
If you don't, go for the 970.
>>51644839
>I just bought a 960 SSC for 150 dollars (with shipping) just before black dickday. Am I retarded?
You probably should have waited for the 960ti coming out around Christmas.
>>51640313
Just money to throw away. Jely?
>>51642566
apparently the competition is real enough for an nvidia shillboi like you to defend your expensive purchase so much
>he payed 600$ for a 5 fps gains
>10x
Yeah... compared to the GT 140.
>>51640313
>>Pascal shipping first months of 2016
summer 2016 (because it taped out in June 2015)
>tfw still running a GTX 460
I really need to upgrade, but I just don't play enough games to justify spending hundreds of dollars on a card...
>>51644871
That wouldn't be the point for not waiting. For others, it would be for the maxwell prices being dropped because of the newer arch pascel being released.
Although I got the 960 real cheap, it could have been cheaper, but way to long to wait for however long.
>>51644909
If you can't afford $400/year you are really poor.
This is coming from an actual poorfag.
>>51644963
i just upgraded from a gtx 260 to a 670.
a 460 should still be fine.
see >>51644432
>>51645152
>i just upgraded from a gtx 260 to a 670.
670s are probably more expensive, or only slightly cheaper, than 970s right now.
You fucked up.
>>51645159
got it for aud 160 used.
a 970 is around 500 dollars.
we don't all live in America you cunt.
>>51640313
I'm still on a 560 ti classified.
Waiting to see what they come with.
Cards still holding up so I can wait I guess.
>>51645209
>we don't all live in America you cunt.
Didn't know prisonland gouged you so hard on those prices.
Good infrastructure has its prices, I guess.
>>51640313
waitfags are the worst
>>51645348
cause you dont get to use that magical word 'shill' on them?
>>51645232
prisonland gouges hard on every price.
(higher wages = things cost more)
and the dollar is getting worse.
>>51645348
last time i was a waitfag i waited for amds bulldozer
>>51640313
If you believe the 10x performance crap, you must be a pretty dumb person.
I bet on 60-70% peformance increase TOPS.
>>51645404
Wrong aud is worth 1.25 to the dollar and considering that they probably tax the shit out of you for having something imported, the math is there. Now go back to your boot licking.
>>51644555
how much of an issue is it if you need to play on high instead of ultra?
also most benchmarks perform less than in real life because of driver updates.
>>51640944
>bad memes
>tfw your dad works for nVidia and brought you home a GTX 1080 to "test"
>>51640313
>10 times the performance 10 times the cost
nice try, they're lucky if it's even 50% faster, look at the Fury Cards, a highly overclocked 390X can keep up with the base fury
>>51645410
bulldozer is fine for multi-threaded workloads
>>51645714
C-can we see it?
http://www.3dcenter.org/news/reihenweise-pascal-und-volta-codenamen-aufgetaucht-gp100-gp102-gp104-gp106-gp107-gp10b-gv100
GP100
GP102
GP104
GP106
GP107
GP10B
THANK YOU BASED NVIDIA
>>51645714
why would they give the janitor a gpu?
>>51645714
well my dad works for 4chan and he could ban any one of you so lol xD
>>51644891
Hes defending AMD gpus as competition to nvidia you illiterate.
>>51643021
>HBM is a huge power draw and heat increase
Consider,
Heat generation is equal to power usage.
HBM1 provided a considerably smaller footprint and half the power draw for more bandwidth than GDDR5
Therefore, for the same speeds/bandwidth HBM uses less power, and thus makes less heat
>having trouble keeping power draw at acceptable levels
It was a shot at discrediting AMDs work on HBM.
If they used GDDR5X to achieve 1TB/s of transfer the power draw would have been somewhere around 120w for the memory subsystem.
It's about 60, or half, using HBM2.
The chart they provided clearly shows that the only other options are much worse. You just can't read.
>bottlenecked by x16 slot
You're an idiot, but we already know you are so it's OK.
NVlink is meant for very high power HPTC/cluster compute situations as a faster path for GPU-to-GPU communications.
Even today you can run a 980Ti balls to the wall on an 8x 2.0 (4x 3.0 equivalent) slot and only lose, maybe, 5% performance.
The rest of your post is purely speculative uneducated nonsense.
>>51640313
thats bullshit.
its probably Q2 earliest
A more important question, when will be gonna see games that uses the featureset of DX12?
>>51643021
HBM uses a fraction of the power that GDDR5 does.
>Nothing to take advantage of it
Yes, that's why both manufacturers have designed completely new architectures to TAKE advantage of it. Until now they have been limited by the bandwidth of the memory they use. Not any more.
>5-10%
Considering this is not only a new arch, but a die shrink, and moving to HBM2, it will be more than 5-10%. The supposed "100%" that nVidia claim is in Compute.
>>51646253
>>51640898
>>51644758
>>51640313
Pascal was slated for Q3 2016.
This is before we take into account any delays caused by the retards at TSMC for taking so long to get their yields up.
Not to mention any production issues that could be faced by lack of HBM2 supply. AMD being the co-inventors, have claimed the lion's share of all early HBM2 production.
Pascal will not be Q2.
Pascal will be lucky to be Q3.
Nvidia has no benefit in outdoing their own current cards by even twice the performance until AMD gets their shit together.
Even if they can release cards that blow everything out of the water right now they won't because it's better to up things incrementally year by year.
>>51640313
>Dumb enough to believe some nVidia jew telling '10x performance of current GPUs'
fucking retarded shill
>>51642211
>>51642248
>10x is for the compute where nVidia sucked because they didn't have the hardware for it
>Now nVidia gets to put '10x the performance' because of their trash hardware from before
lmao this is funn
I just want Fury X2...
>>51643385
>They said that they're having trouble keeping power draw under acceptable levels because of HBM
good joax
the real problem isn't HBM, it's their shitty chip design. They still haven't learned how to push double precision compute without having their power draw shoot through the fucking roof.
>>51646388
unlike AMD?
>>51642164
that's bs. modern games are all made for consoles and run shit on pcs. to run 2015 games on max settings and >1080p you will need a i7 processor. the best gpu wont help you if you use a shit processor. both can be bottlenecks, you are just retarded
>>51642646
You can run CSGO on just a 6600k
>>51646398
Well AMD's cards do double precision.
>>51646414
>the best gpu wont help you if you use a shit processor.
This is the wrong way around idiot
The CPU that's in both the xbone and the ps4 are low-wattage AMD netbook jaguar cores.
8 cores, yes, but they're 15W IN TOTAL.
![gta-v-cpu-4k-vh[1].jpg](https://i.imgur.com/u7EN7qO.jpg "gta-v-cpu-4k-vh[1].jpg")
>>51646414
Ok.
So explain to me how an Athlon 860k can keep this near an i7 in 4k gaming.
>>51646281
Wasn't nvidia lining up with Samsung to circumvent the HBM2 problem?
Anyhow, anyone suggesting any date at this point is just speculative. I'm waiting for January atm to see if they announce something concrete by then.
Before that it's just a pointless war amongst consumers
>>51646480
If Samsung was to produce HBM2, it would be on license for SK Hynix.
No doubt that AMD still claim the majority of Samsung's production, but the increased production alone could limit the production problems.
That's if Samsung can start turning out chips in time.
>>51646398
AMD was able to push 1/4th clock FP64 right up until the 290X launched at 1/8th, then with the Fury X it was reduced again to 1/16th clocks.
Meanwhile Nvidia had to push their FP64 clocks to 1/24th for the entirety of kepler, and then was reduced again in Maxwell to 1/32nd clocks in order to get power draw under control.
Fermi on the other hand pushed (roughly) ~1/4-1/6th FP64 clocks. It's harder to get an exact number for them because of how each SM was structured and fluctuations in power draw and clock speeds.
>>51646457
>hyperthreading off
That's not very fair is it
DO I BUY A FRACTAL DESIGN CASE OR NZXT. PLEASE ADVISE.
>>51646536
Fractal Design any year
>>51646530
Because it provides no performance boost in games. In some games at actually reduces performance.
Do you honestly feel having HT enabled would make such a tangible difference to make my point null and void?
>>51646530
>>51646549
Actually, scratch that. Look at the chart again,
4th result from the top. HT enabled 4790k.
>>51641026
Change the time on your computer to 5 hours from now. Congrats you're over clocked.
For a simple consumer, how is HMB2 beneficial for me?
Right now, all I can see it does is to push the prices up into the skies.
Didn't Nvidia suggest they would use GDDR5X?
Is AMD gonna use that aswell for their middle ranged cards?
I can't see how the memory is the bottleneck and the limiting factor, even for today's most recent games.
The GTX960 only has a 128bit width bus, but when overclocked, the performance gain just correspond to the % of more clock and memoryspeed.
>10x
nice fucking meem
Isn't the new nvidia series supposed to have a new connector? (Not PCIe) or was that just a one time meem ?
>>51646580
>Didn't Nvidia suggest they would use GDDR5X?
nVidia would be shooting themselves in the foot with that decision.
Here's how that would play out.
>Pascal rolls out with GDDR5x
>Marketting try making it seem like a good thing
>Cards prices super high because "New Memory technology"
>Only 24GB/s per module
>Increased power draw vs the already power hungry GDDR5
>Cards are bandwidth starved and struggle in new games
>Devs get lazy in their optimisations due to the increase in graphics power on PC
>Pascal suffers further
>nVidia pr image deder
Not even nVidia are stupid enough to use GDDR5x out of spite for AMD having invented HBM.
>How does it benefit me
Smaller cards
More efficient cards
Cooler cards
Able to push higher resolutions with ease
If the industry didn't think that HBM would be beneficial to the customer, and thus their profit margins, then they would never have adopted it with such haste.
Just think of the possibilities
Half of next Generation's GPUs, all being the size of, if not smaller than the Fury Nano. And outperform it.
My dick tingles at the thought.
>>51646638
There is no increased power draw
It's built on smaller process
Please don't post if you don't know SHIT
>>51646635
nvLink?
They're trying to push it for servers and super computers. But it will flop. Companies aren't going to lock them selves into a vendor like that, not when their only advantage over an offering from AMD is CUDA.
No, companies will float by on the hardware they have then swap out to AMD, or wait to see if nvLink is dropped.
That's if it ever becomes a thing.
>>51640313
>10x mixed precision compute performance
>probably won't be even 10x in mixed precision compute
>expecting any supergenerational improvement in gaming from pascal
so are you retarded or just shilling?
>>51646652
>Please don't post if you don't know SHIT
I suggest you let this fine gentleman know that too then
>>51643385
>>51643021
http://www.nvidia.com/object/nvlink.html
NVLink is awesome, these people don't know SHIT
NVLink is more energy efficient than PCIe and superior
http://www.nvidia.com/object/exascale-supercomputing.html
US Govt has already chosen NVLink and it won't be dropped any time soon
>>51646658
Read about one possibility of using an adapter from two PCIe slots into the proprietary connection, I mean sounds pretty good if that would be possible if it would mean two times the bandwidth
Nvidia fags always buy new cards every year. Once games stop optimizing to their one year outdated card everything starts running like dogshit.
If you cared about value you would have just bought an r9 290 two years ago which still maxes out pretty much every game.
>>51642597
Meanwhile at NVidia
>Only affected a tiny percentage of cards
>Still fixed the next day
It's almost like it's the same shit, right?
>>51646658
No one will swap to inferior, inefficient power guzzling AMD GPUs
>>51646499
>clock
Could someone explain this or provide links for further explanation?
>>51646694
I don't recall 320.18 being fixed the next day.
or 196.75
Or the Win10 driver that was wiping EDID on laptop monitors.
>>51646698
Nice argument.
I especially enjoyed your intelligent and coherent sentence composition.
I hope we can enjoy an educated discussion about religion and politics some day.
>>51640313
I spent $150 after rebate on a 960 because I didn't have a computer. I'll just buy a new one and sell this one off for the same price to some gullible nerd, because pre-rebate it's $180.
>>51645404
>(higher wages = things cost more)
I find it hard to believe people still post this retarded shit. Our minimum wages are slightly higher than nations such as America and Canada. The average wage is still lower and living costs overall are much higher. How many years are people going to post this shit out?
By the way, Australia wasn't a prison colony when the states became a nation. It was founded by legal landowners. It would be like saying America is a nation of Puritanical heretics despite the fact that it was a fucking long time between the early European settlers and the time America became a nation. I like to think people are lazy as opposed to be stupid as fuck.
>>51640313
16 NANO METER
6
N
A
N
O
M
E
T
E
R
WOW
O O
WOW
>>51646638
the 970 did enough damage to their image and people still buy it.
>>51647044
People who don't know what's happening buy them, the people that follow news and shit avoid these kinds of things and other people just don't care nor do they understand what the fuck everything means, they just see that it's being advertised on tech sites and you even get FREE GAYMES with it
>>51647078
not true.
lots of people online claim the 3.5gb isn't an issue. and then they completely fail to mention the false advertising.
a friend just bought a gtx 970 because he checked online and believed it to be a good gpu. all he did was read a few forums.
there are people fucking defending the 970 and pushing others into buying it. fuck that.
Do you guys think my CPU's will last me?
i5 3470 and i5-4460.
I want them to last me until 2018/2019.
Going through multiple GPU upgrades, im good right?
>>51647127
Anon the average person doesn't give a shit. We know what happened with the +0.5 meme and for many good reasons, we care.
But the average person that asks for general info simply goes for something like this:
>hey, tell me a good card for gayman
970 is fair enough
>can i play high res most games?
yes, eventually you'll lower some options and you're good to go
>ok thanks bye
Mind also that this type of person will continue to not give a shit for the next 5 years.
In his own way, he'll be smarter than all of us simply by being less anal about every single detail
>>51647503
I could care less about 970 memes. If it weren't poor value compared to 780ti and 290/x I'd get one regardless.
>>51642566
Just too bad there aren't any drivers for them
>>51647503
>>51647622
Same. I honestly got one, knowing the issue, and honestly it's still a great card
>>51647362
People on 2500K's are not upgrading to Skylake, because there is not much point. You are probably good for another couple of years.
>>51642525
just rent time on a local supercomp desu
>>51647679
Great. Thinking of getting Pascal
>>51646457
Please anon, don't tease stupid people with charts like that, they wouldn't understand even if you explained it.
>>51641278
>almost 20k in the bank now after paying my car off
>I'm too cheap to buy anything new because I always think I will regret my purchases
>tfw still on 5y/o MBP
>>51640313
just got this an hour ago op.
>>51648050
mwuah
>>51648050
lol you bought a artistic 3.5 card and youre talking about getting Pascal. Wow you nvidia faggots really dont know shit.
>>51646580
>Didn't Nvidia suggest they would use GDDR5X?
Anything that doesn't use GP100 will use GDDR5X, so that's all card beneath a 1080Ti/Titan. Even 1080/1070s just aren't cost efficient enough to have HBM2 onboard.
>>51644617
>mfw my 8350 OC'd to 4.6Ghz somehow beat the [email protected] I just switched to (Dual 390x's).
I regret everything...I'd go back but I sold the combo.
>>51648050
>EVGA
oh boy coil whine galore
>>51648465
The cheapest Fury is less than $100 under the cheapest Fury X even with the water cooling. It must not cost that much.
>>51648579
I doubt that the 1080/1070 will use HBM mainly because Greenland might be the only AMD GPU that will use HBM2. We might see the Fury line return as 970/970 competitors given that only two GPUs will be released, and GDDR5X is not that far off from HBM1 in terms of bandwidth.
It's all speculation, but Nvidia only confirmed that their flagship GPUs will have HBM2.
>>51640313
Meanwhile my 290x can STILL handle anything out today. Who here /literally/ has one of the best aging cards?
>>51648677
4GB or 8GB?
Because if you only have 4, you've been keked out of at least two years of more usefulness
>>51642580
nVidia did it first anyway, and like 2-3 times.
>>51648677
>rebranded 7000s
>>51648677
isn't that a current gen card?
>>51648677
Feels good man. Especially cosnidering in early 2014 I was debating between a 780 and 290x.
>>51647503
>In his own way, he'll be smarter than all of us simply by being less anal about every single detail
I envy this person.
It used to be so simple.
>Walk up to a store.
>Pick some AMD or nVidia with high number
>Play games
>>51648531
Is it that bad? I have EVGA 980Ti in my mind that I consider buying
>>51640313
also 10x the price most likely
>>51648677
>still
>>51648756
>Hawaii
>7000 Series
:^)
>>51646678
>US Gov.
>Efficient
>>51649062
Yeah. Now that is more or less reserved to console gayming.
>Here, take this 400 gimme my cow ow duudy Xbox HA!
On the other side you have someone freaking out over a proper cable to use for his monitor because someone somewhere in some forum told him he was chosing improperly
>>51646694
See >>51646712
>>51640313
sorry op im still using a 7870 from 3 years ago
>>51649145
7xxx series was GOAT
>>51649077
he is just trolling
all cards have coil whine, some more than others.
it doesnt matter what brand you buy, shit radeons suffer from the same shit.
>>51640944
With 40 gigs of vram but only 35 of it is fast and 5 is slow
>>51640313
but the card I bought last week was only 230 bucks and I wasn't going to afford/need a top of the line pascal thingy anyway
>>51646457
It can keep up in heavily GPU bottlenecked situations where performance is shit anyway (like the benchmark you posted).
It's harder on a CPU to get 60+FPS at 1080p than ~30FPS at 4K. The higher the frame rate you want the more the CPU and CPU overhead will matter.
>>51646678
It's irrelevant for consumers anyway, you won't get NVLink motherboards or 'gamer' cards.
