http://seekingalpha.com/article/3963242-nvidia-technological-shift-drive-earnings-roof-next-years
>release fiji and 3xx
>amd can't beat Jewidia market share
1/10 Your graph shows the opposite
>>53904861
Op, your graph shows amd gaining more marketshare
>>53904962
>release fiji and 3xx
>amd can't beat Jewidia market share
>1/10 Your graph shows the opposite
Fuck off nvidiot
I just bought a GTX 970 to effectively mine Onecoins with a massive 4GB of VRAM
Polaris/Vega must be a game changer for AMD, or they are running out of money.
>>53905422
Even if AMD releases something truly revolutionary that utterly crashes the competition it still won't mean jack shit. Nvidia are masters of shilling and that's the only thing that matter.
Even if they release shit in a plastic case, people will still buy it because it's Nvidia.
>>53905619
>people will still buy it because it's Nvidia.
No, idiot.
People don't buy AMD because it's AMD.
Have you ever heard a girl talk about how she hates her father? but deep down they are just disappointed.
>>53905888
And they buy GTX 970's instead. Brilliant. Tell me why Nvidia shouldn't fuck you all over when you make a scam like the 970 a best seller.
nVidia as always
> annoucing new cards which will hit the retail market in Q1 2017 or later
> to counter an AMD relase in the next weeks.
>>53904861
Tell your boss that I will buy his products once he writes Linux drivers that aren't a piece of shit.
Until then, Nvidia it is.
>>53906322
Noticed how the only people who complain about the GTX 970 are overly concerned AMD people who worry on behalf of Nvidia users?
>>53906440
though no wood mock ups this time
>>53905888
>People don't buy AMD because it's AMD.
actually yes they do. that's how shills work. amd costs more here and shills still buy them to only get pushed down due to gameworks and shitty drivers, yet they still shill for amd.
i will never understand amd shills.
>>53907542
shills shill things because they are paid to retard. AMDfags are just delusional retarded poorfags.
>>53906471
I guess that's why there's a class action lawsuit about it
>>53906322
>scam
except it really isn't. it didn't perform worse than expectations. it's not like the review cards had the original specs but then they fucked everyone off by releasing a cut down version to the consumers. THAT would be a scam.
it performed exactly how people expected and the slight change in hardware specs didn't phase all the casuals from buying it because it is literally a non issue for them. hell, even the 3.5gb freezing/crashing meme that people were skeptical about before purchasing got disproved very soon after.
>>53905422
Naw nigga, they gopnna b gettin mad cash fo dem licenses from Intel, yo.
>>53907542
Where do you live that a 960 costs more than a 960. A 390 costs more than a 970 or a 390X costs more than a 980?
You're such a hypocrite.
>>53907947
380 more than a 960*
>>53906322
>Tell me why Nvidia shouldn't fuck you all over when you make a scam like the 970 a best seller.
Do you make purchases based off the numbers it says on the box, or do you mean purchases based off independent benchmarks and reviews?
I do the latter. I bought the GTX 970 based on its actual performance, not whatever number was written on the box. I felt the price was justified for what it actually provides.
If you do the former and therefore feel scammed by this, I feel sorry for you.
>>53907947
not sure about the 380 vs 960, but the 970 is nearly $100 cheaper than the 390 AND the 970 outperforms it. AMD fucked up massively with the pricing of their rebrands. They moved the 290 and 290x up in price for no reason other than greed, and nvidia is cucking them because of it.
>>53905888
If you think so explain me why the gtx960 is one of the most sold cards when the r9 380 crushes the shit out of it in EVERY benchmark, unlike gtx970/r9 390 comparisons that are debatable. And in this case "muh housefire" is not an excuse because it draws 180w, nothing that isn't unmanageable or even remotely relevant on the bill compared to the gtx960.
So we have the final proof that nvidiots are really a bunch of idiots, and that most part of nvidia sales are driven by extreme shilling
>but "muh hevc decoder"
That shit is unknown to the 99% of dumb gaymers who bought that card, now fuck off mindless drones
>>53908836
because benchmarks are not the same as real world performance, nvidia cards actually destroy amd much harder when paired with a low end cpu.
>>53909091
>And in this case "muh housefire" is not an excuse because it draws 180w, nothing that isn't unmanageable or even remotely relevant on the bill compared to the gtx960.
>>53904861
>AMD makes shitty cards
>even worse drivers
>wonders why people pick nvidia
>>53909112
meant to reply to this in the other post but:
>And in this case "muh housefire" is not an excuse because it draws 180w, nothing that isn't unmanageable or even remotely relevant on the bill compared to the gtx960.
the extra money on the bill isn't the problem, the problem is that they're shitting out a lot more heat for mediocre performance. the larger dies they ship (tonga itself is 359mm2, whilst gm206 is 220mm2). this means massively higher costs to produce these chips and lower yields for AMD, all because they refuse to optimize their dx11 and opengl implementations in order to push dx12 and vulkan. very poor on their part.
>>53907618
lol. theres a class action lawsuit because some lawyers want a payday
>>53906471
the 970 was a shit card.
>actually making me dig out a speccy from 2014
>>53908836
No Linux drivers
>>53907947
I can buy a high end gtx 970 model like a gigabyte g1 edition or evga ftw for less than the most popular standard msi r9 390. These big factory overclocks by default make the 970 a better performer that the 390.
>970 cheaper
>970 faster
Simple as, mate. Don't get all emotional because your beloved amd literally has no worth here she sells like ass compared to nvidia.
The 970 was a goat
>>53908836
Lol nvidia marketed the 960 as a fucking 4k MOBA card for gooks. I very much doubt it was created to compete with amd in the latest titles since both the 960 and 380 still struggle with the latest titles at even high/medium settings. The 970 is the minimum spec card in these latest titles for a reason.
>>53905619
>Even if AMD releases something truly revolutionary that utterly crashes the competition it still won't mean jack shit. Nvidia are masters of shilling and that's the only thing that matter.
This. It's actually happening for a while now
>>53910188
What is thus truly revolutionary thing amd have released?
>>53910244
rebranded space heaters
>>53910188
>only quantum break
>the rest is shit
>>53910188
>shilling this hard
>>53910244
Architectures that can stay relevant through time so you don't need to buy a new card every 6 months
>>53910313
whatever you say
>>53910188
>>53910313
>>53910374
I'm waiting for the next gen cards with the newer Displayport versions; if I have to wait a few years, then I will wait a few years. There's a distinct possibility I will buy newer (and hopefully cheaper) iterations of VR helmets in the future, and they'll probably require the faster interfaces due to higher refresh rates/resolutions. It's kind of pathetic that we had Displayport 1.3 approved in September 2014, and the GTX 970 is still using 1.2. We just had 1.4 published in March, so I better damn well see 1.3 on the next generation of cards or they'll not get a single cent of my money.
>>53910383
>>53910374
>stock 970 vs a factory overcloked 290/390 housefire
>>53908018
>it took months for "independent benchmarks and reviews" to find out about the 3.5 scam
nice logical fallacy Nvidiot
>>53910396
>>53910374
>>53910188
>buyers remorse: the post
go home and don't come back until you have something that isn't a rebrand, pajeet.
>>53910401
>stock 970
It's a Gainward 970, as you can see here
http://www.techspot.com/review/1019-radeon-r9-390x-390-380/page3.html
1304 MHz clock, which is 250MHz more than reference model. Considering you can OC 390 to 390X level the gap is basically the same, and it's getting bigger each month (these benchmarks are from June 2015)
>>53904861
>release 3XX series
>It's just 2XX series
I wonder why they're losing market share.
>>53910258
>>53910313
>>53910329
>>53910383
Why do amd cards always have shitty frametimes? Just cause 3 had the same problem as well.
>>53910396
>resorting to posting that one ac unity test which causes shitty frametimes
>fails to mention it causes frame time problems for a bunch of amd cards aswell and isn't strictly due to vram access
https://youtu.be/COEwgUI6D3w
Sorry to break it to you but this vram problem is not existent.
>>53910188
please, only the SLI and the 3.5 meme suck with Nvidia, the rest is superior
>>53910401
>pulling implications out of my ass
>>53910458
go home and don't come back until you have actual benchmarks that prove me wrong and not just memes
>>53910469
>Why do amd cards always have shitty frametimes?
>NVidia's bad frametimes are just a meme, pls buy Nvidia it's the way its meant to be played guys!
>>53910488
be careful pajeet, i'll cancel your h1b if you don't make me something that isn't a rebrand soon
>>53910483
Because the 390 and 390x aren't rebranded housefire with only more vram?
>>53910466
>gainward 970
>1152 core clock
>102 mhz more than stock
Why are you lying? Do you not know the difference between a core clock and a boost clock?
>>53910532
1304 MHz boost clock
I've had a Sapphire 7950 for six years now and still feel no need to upgrade.
spec of gpu: http://www.sapphiretech.com/productdetial.asp?Pid=9213647F-B1DB-4945-BE33-8E1353666B91&lang=eng
>>53910466
>Considering you can OC 390 to 390X level the gap is basically the same,
only 390x levels? the 970 has always been outperforming the 390x.
>>53910573
I know right?
>>53910573
>>53910582
>fallout 4
>more than 60 fps
you realize that game is locked to 60 fps max right? that benchmark is very clearly faked, not sure why you would ever use a russian source in the first place either. you are aware that russia is a terrorist state?
>>53910582
Wow that's complete bullshit. I've got a 7950 (as I mentioned here >>53910570) and I get 80+ average framerate on max settings.
Fuck those graphs.
>>53910616
It can be unlocked though physics will fuck up.
The image shows old patch. The newest patch runs better on amd hardware now.
>>53910616
You can unlock the framerate although it messes with the ingame physics, this was made clearly for benchmark purposes
>>53910547
You tried to imply before the card was clocked (core) at 1304 mhz. You can't judge the performance of a card based on its boost clock since that boost clock fluctuates depending on temps and a few other factors. For all we know they clocked it at stock. Why we they use a random aftermarket card to do their benchmarks anyway? It'll be misleading. People want to know the base stock framerates of each card and not random aftermarket card framerates.
>>53910616
>you realize that game is locked to 60 fps max right
Not on pc. The physics is tied to the framerate though so if you get much more than 60 fps, gravity works differently and the game itself runs faster (move faster, shoot faster, etc)
>>53910639
Well they used HIS R9 390 which is one of the lowest clocked 390 in the marked with a 1020Mhz core clock..
>>53910639
>Why we they use a random aftermarket card to do their benchmarks anyway?
because reviewers are stupid. it'd be nice if every reviewer showed stock clocks and various levels of OC on each benchmark, the maxwell cards have insane OC headroom and scaling that seems to be ignored in nearly every review.
>>53910665
Both cards are most likely at stock. It would be stupid to have any overclock on each card. Digital foundry are the worst when it comes to this. They test a 20 mhz oc 970 vs a 1070 mhz r9 390
>>53910717
They seem kinda biased towards Nvidia, yes, but their videos are really informative and well made so I can forgive them.
>>53910582
>>53910717
>They test a 20 mhz oc 970 vs a 1070 mhz r9 390
i assume you mean 200? that isn't even max OC for a 970, it's not uncommon at all for a 970 to get nearly 1500mhz boost clock, meanwhile 1070mhz is the max OC of a 390.
>>53910766
>Launch
nice
>>53910773
AH SHIT WRONG IMAGE
>>53910771
390 can get to 1150 and even 1200 if you're lucky. 1100 is basically guaranteed, which is nice if you consider clock scaling is actually better on AMD cards
>>53910786
The 390 is a overcloked 290
The 390x is a overcloked 290x
>>53910771
No its 20 mhz. They tell you the cards they use for testing in this video:
https://youtu.be/4JHuyLoYU4c
The 390 can go higher but many reviewers don't recommend it since power consumption and heat output go up massively. You have to increase the voltage of the card to get any higher which is what causes that massive increase in heat and power. The 970 on the other hand can get pretty decent oc's on a stock power level.
pascal is a disasterton
amd will drop the hammer in 2017 but nvidiot shill site jonpeddie will still post fake lie market share numbers
convenient how there is only one site for the info
>>53910815
so?
>>53910832
Forgot to add that the only time they bothered to oc that practically stock 970, pic related happened.
>>53910858
The 390 is at the maximum, look the crazy power draw. The 970 is more overclokable.
>>53910453
I bought it months after it came out and the 3.5 thing was well-known.
Besides, if it took them that long to figure out what you seem to exclaim is a major design flaw, then it can't really have been much of a design flaw.
If it actually made it impossible to use the card, don't you think people would have noticed sooner?
>>53910902
Yeah, I see it. Still don't know, my man, how old are these benchmakrs? Because I play GTAV with a 390X every week and I get much better performance than that with a simple i5 4460. The other one I played recently is Far Cry 4, again with flawless performance far above 42fps.
You're still right on overclocking, though, I'd say that's Nvidia's biggest advantage right now
>>53910916
Games started using a fuck ton of VRAM just recently. the 7970 having 3GB was TOO MUCH at the time of its release, people used to say 2GB was enough and the 680 would be fine for years
>>53910979
Texture compression techniques have come a long way since then. There's a reason a 980 ti gets away with only having 6gb of vram at even high resolutions like 4k.
Some people were discussing this in a similar thread a few days ago. They mentioned how the 390 and 390x use no form of texture compression hence why there's 8gb vram. Intel with their igpu and nvidia have been using texture compression for nearly 5 years now.
>>53911054
https://en.m.wikipedia.org/wiki/Adaptive_Scalable_Texture_Compression
Lol the guy who made it was an amd employee as well
>2010
>after AMD released clearly superior RV700 and Evergreen cards
>still selling lower than Nvidia
Just goes to show you that a good product doesn't go anywhere without good marketing.
And goes to show you how insignificant system builders like us are.
>>53911101
those numbers are sales m8, AMD actually did surpass nvidia briefly in market share around the time of the 4000-5000 series, it was close to 50/50 until maxwell came and now it's more like 75/25.
>>53907704
>except it really isn't. it didn't perform worse than expectations. it's not like the review cards had the original specs but then they fucked everyone off by releasing a cut down version to the consumers
This is what I find most hilarious about this whole thing and how it exposes the fact that most people talking shit about a product have never even touched it in the first place, just like >>53906471 pointed out.
A day before that guy ran the tests and found the bottleneck, people were praising the ever living fuck out of the 970. It was crushing benchmarks left and right, people were buying them in droves and posting positive feedback, but once that guy published his findings, it's almost as if all those 970s had suddenly broken from night to day.
>>53910979
>what is caching
>>53904861
the graph shows amd gaining 4-5%
alot of people got burnt by nvidia, but most people, sans hardware failure, are not going to go out and buy a new gpu, much less a side grade at best, also the market share for the highest end, that is in total less then 5% of the graph.
depending on how this graph is made, if its sales for that quarter, or if its an all units in a computer deal, it will change when the new mid range for both come out, as it will be a fairly substantial jump. im honestly betting on seeing amd jump to 35-40% in 2 years time, and if the cpus knock it out of the park, you will see it possibly over 50%.
if people associate amd with a good brand, and lets be real, drivers have not been an issue for amd for years now, however nvidia keeps pushing game ready drivers and many gamers are getting fucked over by them more and more.
>>53905619
nvidia have burnt a fuck load of bridges, gameworks, even by the nvidia buyers, is being seen as a negative now, and with vr coming around and amd being ready, nvidia is milking a hardware architecture 1 sku to long.
>>53911308
what happened was simple, most games dont push 4gb of ram, hell most dont push 3gb even at 1440p
if your game is pushing hardware, and you card buckles, do you just reduce settings and go on or go in depth to find out the bottleneck. most people praised the card, but when they found out the reason their game played like shit was due to being lied to, thats where it went from praise to fuck this card, im buying a 980 because nvidia lied about the 970...
>>53911101
They should get Trump to promote them; get some agency he could form to scout talent.
"Make AMD great again"
Sidebar: they're hiring a few hours away from me, but I don't have experience in ASIC verification or device programming. I'd love to work for them though.
>>53911675
>thats where it went from praise to fuck this card, im buying a 980 because nvidia lied about the 970
You don't just decide to get a $150 more expensive 980. The 970 still sold like hotcakes even after the updated spec sheet. It was by far the best price/performance card of the time and amd had no alternative. It was beating out the 290x in every game whilst being cheaper. It was still a very good card in the eyes of most people except a bunch of false flagging fanboys on nvidia subreddit who were quickly put down because they had no proof they owned the gpu. Ofcourse the Internet being the shithole it is caused those fake reports to go viral. There were never any problems with the card or its vram configuration and anyone who has done even a little research will know this, yet fanboys keep posting false benchmarks like >>53910396 which were actually totally unrelated to the the cards hardware in any way.
>>53909147
This.
>>53905619
nah, i don't buy amd anymore because of past experiences. don't want to waste my money on a flaky product that has lots of little problems.
>>53904861
http://seekingalpha.com is a service for influencing the stock market or at least try, not a source of news.
>>53913121
http://wccftech.com/steam-survey-nvidia-gpu-intel-cpu-popular-steam-users/
>>53904861
Boooo
>>53911927
most people who went for the refund then went and gave nvidia more money for the 980, it was retarded as fuck but it happened.
![frametimes_1080p_gtx_tsuq0[1].png](https://i.imgur.com/bdAuBae.png "frametimes_1080p_gtx_tsuq0[1].png")
>>53911927
>There were never any problems with the card or its vram configuration
The damage control by Nvidia shills over this shit card is hilarious.
>>53915518
>>53911927
>You don't just decide to get a $150 more expensive 980.
Are you this much of a newfag? If you were here during the 3.5 GB disaster you'd have seen half of /g/ were threads about cucks who bought the 970 talking about how they are going to return it (or use the evga's upgrade thing) and get the 980 instead.
>It was by far the best price/performance card of the time
Only literally during launch week. AMD then adjusted the prices so that the 290X was also $330 and the 290 was slightly cheaper, I don't remember the exact price though. This happen literally every time a GPU is released and is faster than the competition, the price advantage never lasts more than a couple days.
Also, people who bought a 290X than (by the reduced price) today enjoy a GPU that is faster than the 970, will become even faster with time as Nvidia will gimp Maxwell after Pascal is released, like they did with Kepler, and don't have to deal with a retarded memory configuration that causes a stutterfest when VRAM usage is high.
>false flagging fanboys on nvidia subreddit who were quickly put down because they had no proof they owned the gpu
It was all a lie, guise! Those people were lying!
It's not like when AMD had the driver update that locked some fan setups at 20% and those people were saying this was killing card (even though that's impossible because GPUs throttle and then shut down when temperature rises, they don't just continue working until they burn). Those people weren't lying shill at all, they were telling the truth! No proof needed, I just know it!
But anyone who mentions the 970 3.5 Stutterfest Edition is a liar who has no proof!
I've ordered a Vive and it should come in soon and I'm upgrading my PC, I was planning to wait for HBM2 but it doesnt look like it will come untill 2017.
What do /g/?
Buy a 970 or 980 untill next year when they release the HBM2 cards? Or wait till June for the cards with GDDR5?
>>53915541
>complaining that in 2015 AMD stopped driver updates for cards released in 2011 and before
>saying Nvidia will support DX12 on "all DX11" cards after Nvidia backpedalled and offered no DX12 for Fermi, which is not surprising since not even Maxwell has basic DX12 support, lacking fundamental features like async compute and complete (tier 3) DX12_0 support, which all AMD GPUs since 2012 do support
>power consumption is important, guise! it's not like the days when Fermi, the original housefire which literally caught on fire, was being BTFO'd, back then power consumption didn't matter at all! but today when Nvidia has an advantage on it, power consumption totally matters, guise!
>64x tesselation, ever
>showing just two cherry-picked benchmarks without driver uptimizations, instead of using techpowerup's summary of several games using the latest drivers
>paying for WHQL certification makes drivers better, guise! WHQL drivers never have issues, like killing GPUs several times on the last few years, failing to support high-end GPUs on Windows 10, causing constant crashes on Chrome and so on
>only drivers the have the very important WHQL certification count, drivers without the certification do not count
>blatantly lying about the fan bug killing cards
3.5/10
because made me reply
>>53915676
>/g/ is somehow representitive of a whole user base
No.
Since you're so sure of this 3.5 problem, link me any legitimate footage that proves it was an issue.
>>53910902
i own a sapphire r9 390 from launch (stock 1010 core) and i can overclock to 1140core/1650mem stable with +100mv on MSI afterburner and it never goes above 78C while gaming at 100% load
whered you get that "at its maximum" information from? do you own a 390?
So what would /g/ say that currently is the best pick. Without shitposting about brands?
>>53918093
Digital Foundry:
>we don't recommend overclocking the R9 390 beyond the 1050-1070MHz territory seen in factory-overclocked cards - power consumption and heat generation rise enormously.
At stock speeds the 390 uses an absurd amount of power. Overclocking increases this by a massive amount. It's just not worth it. You'll need at least a 500w psu for the gpu alone if you want to oc it. If you want to also oc your cpu you'll need like a 700w psu.
I have a gtx 980 ftw oc to 1450 mhz and a 4690k oc to 4.5ghz and it barely even stresses my 500 psu. Also my 980 doesn't go above 65-70°c even with fans at 15%.
>>53909147
>AMD drivers have been giving steady performance gains
>Nvidia drivers have been breaking and gimping cards
Fuck off shill
I really wish bitcoin mining never happened for AMD,they didn't capitalize on it as much as I see it,though I could be wrong.
Can someone with more in depth knowledge explain bitcoin's mining effect on AMD since?
>>53904861
>Consistently release the better product for the $100-$300 range
>lose because of marketing
AMD needs to get it's shit together.
>>53920365
I think problem is simple, they release they shit late.
I was waiting on R9 2xx or GTX7xx to come out and 780 came out first, so I got that one even tho R9 290x would of been just as good.
>>53920365
>lose because of marketing
It's not the marketing anon...
>>53920438
But the 7970 already existed
>>53920365
the problem is that their products aren't better. AMD rebranding their GPUs and firing their engineers has done much more harm to their products and brand than nvidia marketing to gaymurs ever did.
>>53921533
But AMD has offered better performance for the money over Nvidia's mid range for a while now, the 380 & 390 beat out the 960/970 with ease, though the 390 needs 1440p to do it
>>53919317
>amd drivers
>implying it's not the benefit of consoles having amd hardware helping with optimisations
The drivers are still shit
>>53922109
Consoles still aren't the same on the software side as PC.
>>53922150
Optimising for an architecture benefits anything that uses the same architecture. As devs start squeezing every ounce of performance from the consoles they'll get better and better at learning and optimizing for the amd hardware within the consoles. Most recent games are console ports now so
>game engine optimized to run with maximum efficiency in console amd hardware
>game ported to pc and amd desktop hardware benefit from sharing similar architecture as console
>game engine naturally works better on amd since its suited to it
