>Wait for Vega
>>61978126
>wait for drivers
>wait for navi
>wait for drivers
>wait fo x
>>61978126
>.4% worse
>"HOLY SHIT GUYS DOA"
I'm not saying its actually good (Performance per watt is terrible and its competing with an older card) but please make your shilling a little less obvious
>>61978126
> No price:performance ratios
>fake msrp's
>>61978126
nice try, jensen
still gonna buy vega 56
I'm going to try and get in on a Vega56 + Ryzen bundle. I want to upgrade both parts and buying Ryzen + a separate 1070 is too expensive to be worthwhile.
>>61978126
>400+W
>>61978939
it costs more than a 1080 you mong
>>61980375
No it doesn't you fucking nvidiot its MSRP is literally the same as 1080
>overpriced memecoin mining cards irrelevant for gaming
Jesus fuck, if you pay over $100 for a fucking video card, you're an idiot.
>>61981252
vega 64 is $600 while 1080's can be found for $515
>>61978899
>SLOWER THAN THE GTX1080 ONE YEAR LATER WHILE CONSUMING ALMOST AS DOUBLE.
Literal DOA
>>61980000
>$500 MSRP
Yeah no.
>>61981271
>gaming at 1080p/60
What is this 2007?
>>61981324
>Vega 64 is $600
It's actually going for $700.
>>61981324
The MSRP for the standalone card is $500 and AMD just confirmed that yesterday what's so difficult to understand?
>>61981345
More like $400 for the standalone card.
>>61981387
>The MSRP for the standalone card is $500 and AMD just confirmed that yesterday
good luck buying one while you can get a 1080 for as much anywhere
and i'm saying this as someone who bought a 480 and R5 1600
>>61981365
Not AMD's fault retailers are a bunch of greedy anti-consumer pieces of shit and that miners are driving the demand up. AMD is literally doing their best to restock Vega, just be patient and wait until prices stabilize.
>>61981387
they better gonna release the vega 56 aib on 26th august
>>61981387
>He doesn't know
It's actually $499 for Vega 56 and $599 for Vega 64 MSRP, btw Vega 64 is going for more than $700 right now.
>>61981422
>Low stock is the retailer's fault.
For (You).
>>61981422
>just be patient and wait until prices stabilize
fucking Volta will be out by then
>>61978126
>Literal engineering sample on alpha BIOS with no factory overclock, just increased power target, virtually ties factory overclocked AIB 1080 with mature drivers.
Looks pretty promissing t.b.h f.a.m.
>>61981358
>caring about graphics in videogames
You are now required to play only games released before 1995 for two years.
I hope that you will come out of this a better person and have a better appreciation for actual gaming, vs. today's style-over-substance barely interactive movie bullshit.
You want pretty graphics and a fucking neverending story? GO WATCH A FUCKING MOVIE, CHILD.
>>61981597
what the fuck is wrong with you
there is nothing wrong with wanting better graphics, that's literally what everyone has been doing since the beginning of this industry
my favourite game of all time is still heroes of might and fucking magic 3 but that doesn't stop me from admiring how pretty games are nowadays
>Vega is dead
>>61981597
Autism
>>61981686
For everything other than gaming it's good.
Been amd admit it's more of a workstation card which is why t y launched it as one months prior.
>>61981329
Actually it is marginally faster than 1080 but consume quite a bit more power (higher than 1080Ti at peak)
It has potential to go faster but it is entirely depended on drivers and software taking advantage of its new architecture. It is AMD's GF100.
>>61981994
>It is AMD's GF100.
No, since it has no yield problems or anything.
Just consumer drivers are both late and currently unfinished.
It's also no HPC chip, no 1/2 FP64.
It's a mining card that has mediocre but acceptable gaming performance. It will still sell out because of the crpytocurrency crowd.
AMD have a fraction of the budget of Nvidia or Intel. It's a miracle Ryzen arrived in relatively good shape.
AMD graphics driver devs are woefully behind schedule. Performance can only go up from here.
Right now the only sure bet is the 56 though, it's the one most in line with what Vega can do on reasonable wattage, and it does very well against the 1070 already. If current Vega cards never get any faster because of drivers, I would still not regret buying a 56, but I would with a 64
>>61982138
Current gaymen drivers are alpha-tier in quality.
NGG path is not even implemented yet.
http://www.pcgameshardware.de/Radeon-RX-Vega-64-Grafikkarte-266623/Tests/Benchmark-Preis-Release-1235445/3/
DSBR is supposedly enabled.
>>61982150
Sigh I wonder when amd will actually get off their asses with gou drivers...
Navi better not be this bad but considering it's mcm it I'll be even worse.
>>61982326
More money = more code monkeys for drivers.
Anyway the Pro stuff works pretty nice already.
>>61982337
What's the point when all I want is a Vega 56 oc aftermarket for $500aud
Sure beats paying $1000+ for a 64
>>61981994
>faster than a 1080FE in some games, while behind in others
>if it's anything like rx480/580, it'll have virtually zero room to overclock, meanwhile 1080 can OC a good amount
>all of this while consuming much more power
im an amd fanboy, but this is embarrassing
>>61982363
There's no aftermarket 56's yet, besides ASUS one.
Looks like AMD is staggering the launch.
Isnt GTX1080 like 1.5 years old?
>>61982026
It is GF100. The architecture is focused primarily on general compute and geometry throughput a.k.a workstation graphics. Gaming performance is secondary just like GF100 back in its heyday. Vega is hot and eats more power then compention just like GF100 back in the day. There's supply issues as well but Vega is from HBM2 memory not from the silicon itself.
>>61983070
And gaymen graphics needs no geometry throughout?
Anon this is getting silly.
Vega20 will be GPGPU thing. Vega10 is simply a nice all-arounder, just like GP102 is (that one can't into FP16, but is very stronk in int8).
>>61982026
RX Vega obviously has 1/2 FP64 disabled, it's a consumer version. Radeon instinct has 1/2 FP64 and 2x FP16.
>>61983125
Instinct is still 1/16. Read the spec sheets first before shitposting.
Hypothetical Vega20 will be FP64 chip.
>>61982382
1080 can barely OC over its boost speed because the voltage control is locked and power boost already pushes the chip near its overclocking ceiling. It isn't really a good OC'er.
1080Ti has more overclocking ceiling because it hasn't been pushed to its limits.
Nvidia already used up all of their clockspeed juice with 16nm process on Pascal. Don't expect any miracles with Volta. I do suppose that Nvidia could unlock voltage control as a marketing gimmick. If anything Volta is going to be more of a compute monster.
>>61982506
Isn't your entire life like worthless?
>>61983154
Are you saying the MI25 doesn't exist? It says clearly 25 tflops fp16 / 12.5 fp32.
>>61983162
Only GV100 is a compute monster, and it's really-really big (with really-really low yields). The rest of the Volta lineup will be bigger dies.
Think of GV104 being as big as GP102.
>>61983195
ITS STILL 1/16 FP64 YOU RETARD.
I wonder if they'll go taller (moar ALUs per SMM) or wider (moar GPCs) for GV102.
>>61981387
The MSRP sure helps when in the real world the cheapest ones cost €649 (= $764).
https://geizhals.de/?cat=gra16_512&xf=9809_14+10215+-+Radeon+RX+Vega+64
>>61983197
Nah, Volta is going to be a minor evolution of Pascal with more focus on general compute because that's where the real money is going to be in the near-future.
>>61983299
>GPGPU meme, again
About to be killed even in meme learning by dedicated meme processors.
>>61983121
On paper and professional-stuff Vega is able to match the Quadro version of GP102. Gayming has a stronger bias towards shaders depending on the title or you are using AA/AF or not. That's part of the reason why Vega falls behind in gaming. There isn't enough bandwidth to keep all of those 4096 shaders happy.
>>61983336
There's MORE than enough bandwith to feed 4k ALUs, given that DSBR is enabled.
Stop memeing that BW is the problem.
>>61983312
It isn't a meme you idiot. It is the future of high-end GPGPUs.
Gayming tards don't want to face the music that they are becoming more and more irreverent.
There isn't enough capital in the high-end PC gaming market to justify massive R&D costs for large GPU designs.
The masses are more then content with 1060 and iGPU on modern CPUs for their gaming needs. The low-end discrete market is already dead and mid-range is about to face the same threat once Intel designs a decent iGPU that can do 2Megapixel gaming and AMD will have one with Ryzen Mobile (Vega => Navi).
Nvidia already has seen the writing on the wall and have been slowly moving away from high-end PC gaming as their bread and butter.
>>61983405
GPGPU is L I T E R A L L Y a meme. There's barely any workloads that actually benefit from GPGPU in any meaningful way.
>>61983430
fuck off retard
I'm just sad. I really wanted to grab Vega 56, but at $500, that is a no go. That is, unless someone figures out how to enable SR-IOV, that's a whole different story. I guess I'll stick with the cuck 970 for another year and wait for two die Navi.
>>61983441
Jensen please. You and your company are. IBM's cumdumps at best.
Your better invest in visualisation.
>>61982326
>Navi better not be this bad
Vega is far from being bad, and Navi will be a juggernaut of a architecture.
http://www.tweaktown.com/news/55875/amd-launch-monster-navi-10-2019-next-gen-ram/index.html
>>61983537
Both Navi, and whatever NVIDIA will name their MCM GPU uarch will be monsters.
>>61983553
>mfw cute small efficient die scaled up to behemoth destroyer of compute
>>61983430
Some HPC and enterprise markets would completely disagree with you. It is the sole reason why Intel has been trying to cram AVX down its entire CPU market when only their HPC and enterprise customers benefit from it.
Kiddo there's much more to high-end computing then silly gayming.
Nvidia isn't stupid and are having bank. They aren't going to fall for SGI's mistakes. (Protip: most of Nvidia founders are ex-SGI guys)
>>61983610
>some
The key word is SOME. Only and only SOME workloads benefit from GPGPU acceleration.
The entire datacenter market is much much bigger than the part that actually considers GPGPU for anything.
>>61983631
It is still larger, more profitable and [bold]stable[/bold] market then "high-end" gaming GPUs.
Nvidia wants a slice of that pie, also they are get on the whole crypto-currency craze with Volta and beat AMD at its own game.
>>61983405
>Gayming tards don't want to face the music that they are becoming more and more irreverent.
http://files.shareholder.com/downloads/AMDA-1XAJD4/5023073131x0x953561/F850D328-4AEF-479F-B0B6-88703C81A28A/Rev_by_Mkt_Qtrly_Trend_Q218.pdf
Q3 FY16
>58%
Q4 FY16
>58%
Q1 FY17
>53%
Q2 FY17
>55%
Q3 FY17
>62%
Q4 FY17
>62%
Q1 FY18
>53%
Q2 FY18
>53%
Q1 FY17 53% > Q1 FY18 53%
>>61978899
For the low low price of $700 usd.
>>61983659
Nvidia's datacenter revenue is only 2/5 of their gaymen revenue, and that's considering they are riding the ML boom.
It'll drop to silly amounts once dedicated meme processors arrive.
>>61978126
The wait is not over. Drivers will enable FineWine.
Gotta wait. For Vega.
>>61983674
inb4
>mining
>>61983674
The lion's share of that is from mid-range GPUs a.k.a 1070, 1060 and 970 not high-end solutions.
That market is about to get besieged by next generation of iGPU + CPU combos. They just need to be "good enough" for the normies and nu-gaymers.
Miniaturization is a bitch. High-end gaming GPUs are going to become as niche as high-end discrete audio solutions.
>>61983162
Air cooled 1080ti reaches its heat ceiling very quickly around 2k, mine can't even maintain it stable.
>>61983740
Duh, it's 3.5k ALUs.
>>61983722
Nah, since the resolutions are getting higher.
>>61983674
>they are becoming more and more irreverent.
Holy shit. If this keeps up they're going to be a shitty adult swim comedy by 2020!
>>61983752
4K is meme and it is at limits of human visual acuity with 30-40" screens.
Besides, framerate is what majority of high-end PC gamers want not higher resolutions if given a choice between the two.
>>61983784
>4k is a meme
It's not.
But since nothing can achieve 4k@144fps, there's still a LOT of room for growth.
>>61983752
So it overclocks slightly worse than 1080, why say its better.
>>61978939
>poorfags
>>61983829
1080 boosts very high compared to 1080ti.
You are *actually* overclocking 1080ti. To a degree.
>>61983784
I use a 42" screen on my desktop for 4k, I don't get why thats so uncommon.
>>61983797
The normies and nu-gaymers don't care about them.
Only a tiny, but vocal minority does. That market is not large enough to justify dedicated high-end GPU designs.
>>61983852
Yes, midrange GPU market is still the king.
But no, iGPUs won't kill it.
oh yeah i can totally see how the card flops
http://www.tomshardware.com/reviews/amd-radeon-vega-frontier-edition-16gb,5128-6.html
oh wait
>>61983369
fucking abbreviating bandwidth to BW
technology was a mistake
>>61983896
The other Germans even did B3D suite benching. Primshader culling is disabled for everything but selected few workstation tasks it seems.
>>61983847
Its running dow3 at 1860 just off boost atm, manual oc adds very little to the frame rate at 4k really.
>>61983692
>implying it's in stock
vega 56 isnt much of an upgrade over a r9 Fury it seems though i cant find any direct comparisons. And I'd only buy it under $400. Should I just do the Fiji core activation on my fury? It looks like i have 4 cores that could be activated
>>61983916
>technology was a mistake
Fight me fgt.
>>61983961
It is.
>>61983965
>4 cores
You mean 4 CUs?
And no, 56 won't be an upgrade, for now that is.
AMD push Vega to the limit, this card is a turd
>>61983168
He probably bought an Nvidia card that's faster, and cheaper. So he has thst over you wait fags.
>>61984042
well its not good. but for a guy like me who has a good freesync montir, vega liquid running on powersave mode works fine.
its like 5% slower than a 1080
>>61983405
Jesus christ, this level of delusion...
The high end space is a failure for AMD since they can't compete in a timely fashion. Despite them desperately wanting to enter the high end card space for those sweet premium profits. There's a reason the mid range stuff is barely profitable, because whether it's a big die or a little one the cost to manufacture the two is largely the same. The ONLY difference when it comes to manufacturing is how much silicon is used and how many wafers you can maximize. The bigger the die, the more expensive it is.
And catering to miners like AMD does means they are no longer worried abiut enthusiast cards. Just pushing product out the door to salvage this shit architecture.
PC gaming is now a 30 billion dollar a year marketplace. And AMD can't compete.
It was obvious the 1080 Ti would be the better card all along.
11GB > 8GB
I don't give a shit that these cards don't compete with high end nvidia cards.
What I find disgusting about them is the price and power consumption.
>>61984612
AMD wanted to build a hotter FERMI.
>>61983537
Just wait ®
>>61983722
Far from reality. Until the resolution keep increasing gpus will have plenty of market to increase. Today top solutions are breaking even 60 fps on 4k and it's not even considering hdr which will need additional power.
GPUs will stay relevant for at least next 4 years.
>>61983771
Hi>>61983852
There is a very high margin on top CPUs, Nvidia main revenue come from gaming GPUs. Today they even lost fractions of dies since they bin it to lower products.
>>61984562
Mid-range is where the bulk of fucking profit and revenue comes from via sheer volume.
Titans and 1080Ti are just halo products that barely make anything in themselves despite their profit margins per units. They only sell tens to a few hundred thousands at most while mid-range stuff sells by the millions. It is simple scale of economics.
Toyota isn't #1 Car Manufacturer in the world because they sell the best high-end performance cars. GM in its heyday wasn't a juggernaught because of muscle cars either.
The problem is that mid-range portion is aboutl to undermined by next-generation iGPU + CPU combo that offer "good enough" performs for the majority a.k.a normies and nu-gamers that play LOL, COD, Overfail and DOTA2.
History is just repeating itself. Discrete audio cards used to be huge before integrated audio solutions became cheap and good enough. Discrete GPUs are going to face the same fate.
>>61984763
Again, the normies and nu-gaymers doesn't fucking care.
It is just a vocal minority.
>ITT: /v/tards don't understand that times are changing. High-end PC gayming is going the way of old muscle cars. A small minority fueling their own niche by pure nostalgia.
>>61984818
Top-end CPUs are bought by professionals and enterprise users a.k.a Xeons and Epycs.
X series Core chips and Threadrippers are just Xeon/Epyc rejects that only get bought by prosumers/power user a tiny minority.
This isn't the 1990s and early 2000s anymore.
>>61985078
that scenario is a long way off
>>61985078
Lol!
Holy shit... <benderyoureserious.png>
Bwaaaaahahahahaha!
>And then he said onboard audio solutions became good enough!! He doesn't realize Microsoft killed the soundcard industry over night!
Wew lahd!
>But wait, get this, he thinks GM in the muscle car era wasn't big because of it's muscle cars! He also doesn't know the big three have been living on record profits of their full size truck and SUV sales for over two decades now!
Bwaaaahahahahaha!
>Oh this idiot gets even better, he compares making a graphics card to a fucking Toyota Corolla!
Bwaaaaahahahahaha!
>And the Piece de Resistances to this mental midgets argument, is that the Halo products themselves weren't profitable! *snicker* He doesn't know how much money Nvidia makes in the professional and HPC sector, he doesn't know that the cost of manufacturing a 700 dollar 1080ti and a 1200 dollar Titan or a 1070 and a 1080 is the same price!!!
LMMFAOSHIFADMT!
>I just don't have the heart to tell him that companies are actually happy and completely satisfied with Nvidia cards even at 5 times the cost, because they have the customer support and the software ecosystem that makes their hardware actually pleasurable to use!
The poor guy wasted all that time trying to make a coherent argument...
>>61985135
Lol. Maybe because people don't want to pay 500 dollars for a 200 dollar GPU? So they have to settle for the bullshit that is low end gaming.
But lets release a mining driver update before a gaming driver update, we know who our real customers are!
>>61984042
This is not good
>>61985078
>The problem is that mid-range portion is aboutl to undermined by next-generation iGPU + CPU combo that offer "good enough" performs for the majority a.k.a normies and nu-gamers that play LOL, COD, Overfail and DOTA2.
Okay well I can guarantee you an integrated will never be able to run DOTA2 well
Bought a 1070 shortly after launch of aftermarket cards, could sell it today and literally earn 10% profit over what I paid.
>>61985242
Nope, it is unfolding as we speak. It has been happening in the last few years. These transitions take decade or so to really sink in.
Laptops and other portables didn't take over normal desktop PCs overnight. It happned slowly and it took years but laptops/portables have pretty much take over desktop PCs. Desktop PCs are just niche like Big Iron boxes of the old days.
PC /v/tards are just living in denial.Thinking their fragile bubble is perfectly safe from the advances and minimization of computing technology.
People back in late 1990s and early 2000s would have laugh you if said that portables and smartphones would take over PC space.
>>61986232
It is already there if you don't care about the eye candy and "MAH 4K!" a.k.a the majority of those playing DOTA2 and LOL on their laptops and OEM rigs.
Just imagine what would happen if you don't have to compromise eye candy with an iGPU solution.
>>61985387
Kiddo, I got a Vega 64 en route to power a 4Megapixel Freesync monitor at 144Hz. (RX 480/580 and Fury don't cut it)
I'm not denying reality. High-end PC gaming is becoming a niche whatever "we" like it or not.
The world of mainstream PC gaming soon will be dominated by portables, SFFs PC powered by iGPU solutions.
Like I said, minimization is a bitch. Those big GPUs and tower PCs chassis will end-up looking and sounding like old muscle cars of the 1960s-1970s.
Zen+Vega APUs with HBM2 are expected later this year.
>>61987823
This is how much of an in denial faggot you are.
https://www.extremetech.com/gaming/243272-pc-gaming-surges-rest-pc-market-declines
The PC gaming market is growing despite shrinking pc sales. And it's not just shitty low end systems that scrape by on iGPU's and dual core systems. It's a majority high end and mid range PC systems.
Even AMD's marketing is about changing "Personal computer to performance computer".
https://www.youtube.com/watch?v=FwcUMZLvjYw
So take your AMD apologist marketing shill ass back to your supervisor for how to address your bullshit lies, faggot. No matter how desperately you want to call people /v/tards, Vega is a pile of shit.
>>61988016
No HBM. That's next year Q4, MAYBE. Raven Ridge will use DDR4 as VRAM.
>>61988049
Oh, if so that sucks. It's still going to be quite decent, but not the same.
Source?
>>61988060
They've never talked about HBM on Raven Ridge for one thing.
>>61988094
Not specifically, but they've mentioned "hbm in future APUs".
Of course, the upcoming generation of APUs will be a huge jump one way or another, just because of zen vs OldShit.
>>61988118
50% performance in CPU performance, and 40% performance increase in GPU performance is what they have said.
If the rumors are true about Vega, since I can't find a single test on it, but if Vega undervolts and under clocks as well as I think it might, it is going to be an embedded GPU champion. And the Vega cores will redeem themselves on the APU. Now imagine four full fat Vega cores at 7nm, running ultra low power to give better than 14nm Vega 10 performance.
>>61988022
That growth is from Eastern EU/China/India buying what would we considered to be mid-range and low-end OEM systems.
The PC gayming market in USA, Canada and Western EU has pretty much stagnant for a decade and is experiencing a recent but slow decline. The recent VR fad has failed to re-surge the market. The normies and nu-gaymers are getting mid-range to low-end desktop or portable solutions to satiated their gayming needs.
There's no killer app in sight to stop this trend. Smaller PC etailers have been closing shop and companies that specialized in "PC gaming" have been downsizing or closing their doors.
It is worse off on the software side where AAA developers are just re-hashing decade old franchises and aren't taking any risks. Entire genres are either dead or in ICU.
It is telling that most popular gaming franchises right now don't need high-end hardware to have an "enjoyable" gaming experience. They are able to run on OEM rigs and laptops without an issue. Hell some of them just settle for a gayming console.
Most of you /v/tards just don't see the big picture. Shareholders Nvidia and AMD RTG aren't stupid. They have seen the writing on the wall and have been trying to move onto greener fields.
>>61988227
https://www.youtube.com/watch?v=JOcz-H5u3Rk
This video is related
Replace "Buggy Whips" with High-end PC Gaming PCs.
>>61988227
Read the fucking article next time you lying piece of shit.
>Demand for high-end systems that cost multiple thousands of dollars is strongest in the West and European markets, while the Asian markets prefer lower-end to midrange hardware. This likely explains why AMD has launched several Radeon SKUs specifically for the Chinese market — it makes good sense to do so, given spending patterns.
43% of growth was in the high end.
35% of PC gaming growth was in the mid range.
22% of growth was the low end.
A 30 Billion dollar a year market that AMD can't compete in with their GPUs.
>>61988202
Very promising indeed.
I just love when the "baseline" is raised. With APUs that powerful, it'll mean everybody with a laptop or low-profile workstation past a certain date will have at least that much performance.
>>61988476
Let's hope they aren't stuck on 1366x768 resolution screens. I want my next chromebook powered by a raven ridge APU.
>>61988423
Again, that growth all from markets outside the Western EU and North America. That "43% growth figure isn't that impressive when high-end market in itself would constitutes to a few hundred systems sales versus millions of systems sold in mid-range and low-end market.
It is marketing faggots trying to mince numbers around to paint a rosy picture. It is all short-term growth to boot. Give it another 3-5 years and watch that growth figure for high-end market disappear and become a steady decline.
The masses don't care for high-end PC gaming anymore and people who are already in are going to go elsewhere.
>>61989522
The only marketing faggot here is you.
https://www.reddit.com/r/Amd/comments/6urosh/rx_vega_64_direct_from_manufacturer_699/
>$699
TOP KEK
>>61989552
>mfw when /v//tards are screeching in their tiny echo chambers "MAH 4K GAMING!", "PC GAMER 4EVER!"
>>61987945
>Like I said, minimization is a bitch.
You're fucking retarded. Miniaturization won't magically fix the thermal/power delta between integrated and dedicated GPUs.
>>61989742
You aren't getting the gist of the memo.
The majority of people don't like running high-end GPUs because they eat too much power and are big as fuck. They also don't really like massive PC chassis needed for them.
This isn't the fucking 1990s and 2000s where it was acceptable to have your PC sound like it belongs in a server room. Water-cooling isn't some magical panacea either (Pump failures, leaks and possible condensation issues if done poorly)
The trend for PC in general has always been going smaller and being more energy efficient.
Creative used to be a juggernaught back when discrete audio solutions were in demand, but once integrated solutions became "good enough" they took a massive nose-drive and now are facing bankruptcy. Discrete audio solutions only live on in the audiophile/audio professional market. The same fate is going to befall discrete GPU solutions.
>>61989908
Past the emu10k1-based cards, creative's only made shit. It's no surprise they're deep in the red.
>>61989908
High end (>1070) GPUs are already a niche product. Just look at Steam hardware profiles.
1080 is far from the most popular card, and 1080 ti holds a whopping 0.53% marketshare.
http://store.steampowered.com/hwsurvey/videocard/
>>61989956
Exactly, this what most v/tards fail to grasp
Once iGPU solition yield performance similar to $199-$249 discrete GPU solutions. The mid-range discrete market will begin to die. iGPU solutions already killed off low-end discrete market.
The mid-range is where the bulk of "gaming" revenue comes from. Once people stop buying cards en mass the well starts to dry up. Nvidia already knows that its gaming division has been living on burrowed time. AMD's Vega and Navi have always been about the long-game. AMD RTG knows that it cannot realistically overtake Nvidia in the discrete GPU market. How to defeat them? By focusing on making iGPUs being "good enough". Intel is going also up their game in response. Nvidia has been trying to move away from gaming as their bread and butter ever since Fermi. Why do you think had such a massive boner for portable market via Tegra family and now have a boner for GPGPUs for niches that desire them?
>>61990076
>iGPUs will replace the mid range in no time at all!
>been saying that since 2007
Yeah, okay. No one is going to stick massive GPU dies on mainstream APU/SOCs. A 1060 has 4.4 billion transistors, it's almost as many as the full 8-core Ryzen. The relative size of GPU to CPU won't change in the future. Not to mention to even make use of that you need something like HBM on an interposer, and all of the sudden you have a massive chip. It's big, it's expensive, and it'll have higher heat density than a separate CPU and GPU. It's just not happening outside of semi-custom, like consoles.
>>61990215
>been saying that since 2007
But it WILL happen, mark my words.
>>61989660
>Substituting reality to inject my fantasy
Enjoy it while it lasts. People have been saying what you've been saying since AMD launched fucking Llanos. As the market shifts to 4K, and those iGPU's are just now barely getting acceptable 720 performance, how much longer will we have to wait for SoC's to become powerful enough for 4K? 10, 20, 30 years?
Meanwhile in the real world, high end PC sales have skyrocketed. The market is growing while the rest of the PC market contracts.
>MUH MINICHURIZATIONS!
Yeah, ok...
AMD can't make a high end chip to compete so you have to apologize for it by coming up with pure fantasy.
>>61990215
This is already happening right before you eyes. Give it another 3-5 years. Ryzen Mobile (Zen + lower-end Vega) is just the beginning of this. Laptops and other portable didn't take over mainstream PC overnight. Discrete audio cards didn't disappear from shelves overnight either.
Nvidia's founders were ex-SGI guys. They know first-hand what miniaturization does to your market. SGI died because single chip solutions start to work of their massive workstation/mainsframe systems. GPUs that can do mainstream workloads and causal gaming are getting small enough that you only need to allocate a portion of the package for it.
This isn't about trying to achieve 4K/144FPS gaming on an iGPU. It is about achieving 1080p gaming and content viewing a tiny GPU packaging a.k.a what normies and nu-gamyers go at. We are practically at near that milestone.
I'm not saying that high-end PC gaming or high-end GPU solutions are going to disappear but rather they end-up being tiny niches that are fueled by nostalgia. The masses get by with their iGPU solutions driving their portable PCs and SFF PCs.
>>61990367
They have had a high-end GPU that can compete against Nvidia in the workstation and professional market which have far greater profit margins and isn't going away unlike the high-end PC gaming.
High-end PC gaming is becoming more and more irreverent even to Nvidia who dominates that field. To them it is a market living on burrowed time. They know that once Intel and AMD iGPU start becoming "good enough" that well is going to start drying up and there isn't enough capital to afford the massive R&D budget that go towards design a GPU solution which take years to develop. Why do you thnk that have going GPGPU first and then distill that design into "GPUs" by axing GPGPU logic and use cheaper memory solutions (GDDR5/GDDR5X versus HBM2). They don't have any x86 license so they cannot easily develop a x86 SoC solution to compete against Intel and AMD. They must move elsewhere in order to survive unlike their spiritual predecessors SGI and 3Dfx.
>amdfags confirm that they're subhuman trash who would spin anything to make their company look good
>>61990675
They are just playing the long game. Nvidia would do the same thing (and probably do a better job too) if they had a x86 license on their hands.
Discrete graphic cards are going to become an niche just like discrete audio cards are now.
>>61990875
>Discrete graphic cards are going to become a niche
How so? you'd just get the same discrete card with multiple smaller dies stacked instead of one monolithic die.
They still need a fair amount of cooling which takes up most of the space.
Nooooooooo!
It was supposed to be /ouryear/!
>>61990491
Vega FE professional cards undercut those profit margins, because they can't provide the same level of experience as Nvidia and CUDA. Good job, I guess?
>>61993198
Again, that will only be cater towards niche that want performance no matter the cost but that market has always been small fries.
The normies and nu-gaymers who make-up the bulk of PC gaming crowd today don't care for high-end GPU gaming solutions. They opt for mid-range shit and that segiment is about to feel the pressure from iGPU catching up because GPU performance per generation has been on the decline ever since Kepler/Tahiti came out. The fucking 1060 is only somewhat faster than 960 that it replaced at a similar price point. iGPU have been getting faster and have already destroyed the low-end discrete market. Just look it is a PITA to find a low-end discrete card that isn't massively overpriced for its tier.
>>61978126
get out nvidia shill
wait for Navi, btw
>140 more watts than 1080
>10% less fps than 1080
>$200 more than 1080
I thought AMD was supposed to be the good budget choice??
>>61993262
CUDA is dying and AMD has been making considerable strides in the professional market.
There's a reason why Nvidia suddenly release drivers that made their Pascal Titan "good" at professional shit when they are just "extreme" gaming GPUs.
>>61989572
It's this way to deter miners
If you buy it in a bundle it's cheaper than the 1080
I don't get it how can it have way more prosessing power and be worse for games will drives make it faster
>>61993294
Actually is closer to 5-10% more FPS and frame-timer is much better than 1080 outside of "Gameworks" BS.
The prices difference is closer to $100-150 because the damn market is inflated by the whole mining craze.
The power consumption is really the only thing that really kills Vega and came out so late.
>>61993337
>frame-timer is much better than 1080
Post a link to a review showing this across a wide range of games.
Is this the Bulldozer of their GPUs?
>>61993336
Architecture choices make all of the difference.
Just how the GF100 and Fury on paper should fast but they were underwhelming at their debuts.
>>61993337
I've seen 1080s sell go as low as $400.
$700 is a fucking joke. It shouldn't cost more than the 1080 whatsoever if they want it to not be a complete fucking waste.
>>61993361
Nope, it is a GF100. Superior at general compute then competition but worse at gaming performance with much higher power consumption. GF100 destroyed Evergreens at general compute stuff.
>>61993376
1080s are going around $500-$600 new. You will not see $400 USD unless you are going for a used 1080 on fleabay/craigslist on somebody looking to get rid of one in order get a 1080Ti upgrade.
Blame the whole crypto-currency effect inflating prices with etailers even if GPUs in question aren't even good at mining.
Just bought a 1080 GTX for 622$ CAD (~495 USD) with taxes and shipping included.
I wanted vega but they are not selling anywhere near msrp in canada.
I predict vega 56 will be a similar disappointment so I bought the 1080 right now before all the people get disappointed and drive up the demand for nvidia cards.
>>61990076
>>61989908
>>61987945
>>61985135
The amount of memory bandwidth needed by modern GPUs alone demands that they use GDDR5/5X or HBM to actually achieve the performance they do. AMD and Nvidia are even putting GDDR5 on their lowest end cards, and that's not something that's going to end up on CPU dies or on mainboards any time soon.
>$699
>yfw you didnt wait for vega
>>61981597
I love Morrowind too anon but better graphics isn't a bad thing.
>>61981597
>not taking the best you can get
Protip: people cared about graphics in 1995 too.
>>61993517
HBM2 will easily solve this issue which is why AMD has been throwing so much weight behind it. It has never been for high-end GPUs in the first place.
AMD RTG never went after high-end discrete solutions in the short-term. They gave up once Maxwell came out. They knew they best bet is going after Nvidia's real bread and butter a.k.a mid-range shit by making iGPU good enough.
Hell, ATI made the bulk of its back in its heyday through the low-end discrete market but that market was already murdered by iGPUs from SoC solutions.
>>61993765
Honestly, RX 470/480 performance in an iGPU would pretty much kill off all but the highest end enthusiast GPU market, so if AMD were to do something like 1/2 of a full Vega die with 2-4GB HBM2 on die with a quad core Ryzen die that shit would sell like hotcakes and suck up most of the dGPU market.
>>61983162
>1080 can barely OC over its boost speed
True-ish but somewhat misleading, the FE with default settings is heavily restricted by its low fan speed, 82C thermal target and its power limit. Just by increasing fan speed and pushing power target to the max (only +20%) and not even touching clocks should get any card to 1800-1850MHz stable core, 1900-2000+ is definitely possible on a decent one. The completely stock 1080 stabilizes around ~1700MHz, that's what the FE does.
Custom cards with a bit of luck can run 2100-2150MHz and if not should at least bounce around in the 1950-2050 area. You could say it's not "really" OC since Boost does most of the work, but in the end there's a pretty large difference between a FE 1080 running default settings at ~1700MHz and a top-tier custom card around ~2100.
>>61990076
>Once iGPU solition yield performance similar to $199-$249 discrete GPU solutions
Not gonna happen any time soon, you think midrange dedicated cards are going to stagnate while iGPUs become some sort of monster? Do you think game requirements are going to stagnate, especially once a new console generation pops up in a few years? Just the extra TDP you can afford to throw at a GPU on a dedicated card is going to make a large difference. Resolution is also on the rise, 1440p is more and more common, so is 4K. Even consoles are getting on the 4K train.
APUs are going to get better for sure, but they'll remain at the bottom of the market for a good while still.
>listening to /g/
You asked for it.
>Actually make a good card by accident
>Fuck up the bios with secure boot bullshit
>fuck up the software
>now clock bug due to drivers
https://youtu.be/SBx73n-fgdE
FFFFFFFFFFFFFFFFFF
>>61981717
Work.
Ryzen 1600x and Nvidia GPU is the only smart combination.
>>61994505
Why not drop $50 more on a 1700 and OC it to ~3.8GHz?
>>61994234
They are already been stagnating for the most part. Shit the 660 was the optimal mid-tier choice for years until 1060 came but even then it is not that massive bump from the 660.
>>61994118
Founder Edition means shit. It is just pre-order-tier cards from Nvidia who want to charge a $100 premium for that. The card themselves are shit-tier when compare to later AIB 1080 and 1070s.
The first batch of FE 1080s had thermal and power issues because Nvidia cut corners on those cards. It was overshadowed by first batch 480 power issues though.
>>61993299
More fantasy from an AMD shill.
Shocking.
>>61978126
It's possible that later driver revisions will improve it. But in any case, just because it's not as fast as its rival's flagship, it still can be a viable budget alternative.
>>61994550
Because you can hit 4.1 automatically on the 1600X and performs the same or better depending on the game.
>>61994943
>improve
They need to make it work first.
>>61994924
Nope, it is more like you haven't been following closely with the professional scene.
Nvidia didn't decide to make their Pascal Titan good at professional-tier stuff with a driver update out of the goodness of their hearts.
Vega Frontier forced their hand into making Pascal Titan more than just "extreme" PC gayming solutions.
>>61981686
I think Apple just got the best of the best binned cards (that probably run at way lower voltage, temps) for mainly one reason: To drive their 5k displays and still have a lot of headroom left over for applications
>>61994943
>budget alternative
Tell that to AMD, retailers and those fucking miners.
>>61978126
I will, thanks for the suggestion.
>>61994943
>budget alternative
>for more than a 1080
>>61978126
>vega finally comes out
>bro just keep waiting
>>61994399
>>61994399
The card is on alpha drivers. It's basically just an overclocked Fury in gaming until the RTG driver Dev team actually fucking implements primitive shaders in drivers to overcome GCN's 4 triangles/clock limitation. Until then, who fucking knows what Vega is actually capable of once it isn't massively front end bottlenecked?
>>61997469
It's not overclocked Fury.
It's bigger Polaris since PDA is working.
>>61997479
Fair point, though the 4 triangles/clock bottleneck is in front of where PDA kicks in, so it's not like Vega's even getting the full benefits of PDA at the moment.
>>61997491
4tris/clock is not the problem.
Discarding is the problem.
Vega is as good as dead in gayms without primshader culling.
>AMD markets FE as a card for gamers who create and creators who game
>mediocre gaming performance
>"I-it's not for gaming!"
>AMD markets RX Vega explicitly as a gaming card
>mediocre gaming performance
>"I-it's not for gaming!"
Why do people here relentlessly defend AMD? Locking the BIOS on a gaming card was the final straw, AMD is as bad as aIntel and Nvidia. Fury X was their last good GPU, Vega is a "fuck you" to their fanbase.
>>61997523
they do this with anything amd creates
remember the ryzen threads right after ryzen was released?
>b-but it's not for gaming it was never even advertised for gaming (not true)
>>61997523
You should read the thread first before shitposting loudly.
>>61997500
From what I understand 4 triangles a clock is the problem because it places the bottleneck in front of the fixed function culling of PDA. Primitive shaders is not merely a matter of culling but rather of reducing buffer swamping with primitives that would be subsequently culled by the fixed function culling and by deffering the generation fo attribute data until after that culling has occurred. As far as I understand the white paper, this is how they arrived at their maximum 17 triangles/clock throughout with primitive shaders active.
>>61997606
Yes. It culls the primitives before vertex assembly.
In all honestly that's an extremely elegant solution to inherent problems of GCN.
Kudos to Pajeet & his team for that.
>>61997523
The gaming drivers are alpha tier at the moment, and in pro applications Vega FE trades blows with a Quadro P6000 in geometry bound workloads. You should probably ask yourself what is different about the pro divers than the gaming drivers at the moment if you want to learn anything.
>>61997541
people are saying in this thread that it's a workstation card and that it's good in mining (for this price and power consumption it is NOT)
hell some posts imply that vega wasn't meant for gaming, then why the fuck did they release them as gaming cards?
>>61997630
It's a general-purpose graphics chip, akin to GP102. It is means for anything 3D, just drivers that enable major uarch features in gaymen are not ready yet.
Only DSBR works currently.
I still can't understand how Vega10 can have whopping 45MB of SRAM on-die.
>>61997644
>It's meant*
Fix.
>>61997630
Doesn't matter, they already fucked up the launch. Just like the Fury X which also had a disastrous launch people will make up their mind and not buy it, meaning any driver improvements are wasted. AMD's gaming marketshare will simply continue to decrease as the only people buying their GPU's are miners and people interested in compute. Soon games will become even worse optimized for AMD, their APU's aren't successful, their mobile marketshare is trash. The 1060 stomped the 480 in gaming marketshare, and Nvidia will continue to hold a monopoly in the high end. Workstation is now the only market AMD an hope to compete in, they've essentially entered a Matrox or 3DFX tier death spiral in gaming.
>>61997698
>will become even worse optimized for AMD
Right after fucking Sweeney appears on AMD's keynote?
Anon don't be silly.
>>61997698
>their APU's aren't successful
but they aren't even out yet m8
the laptop cpus too
>>61997713
Every time a game comes out the Steam forums are flooded with AMD owners complaining about bugs. Games devs don't give a shit about AMD. Also Sweeney shills for anyone who will pay him, he shilled for Microsoft then as soon as they bought the Gears IP he burns bridges with them and starts screaming about how Microsoft is killing PC gaming, after he and Cliffy B tried killing PC gaming the whole of 7th gen. He shilled the PS4 nonstop leading up to its launch, outright lying about its capabilities then promptly dropped it as soon as it launched. No one should give a shit about Sweeney or Epic anymore, they turn on people at a moment's notice.
>>61997800
>Games devs don't give a shit about AMD
and that's because all new AMD gpus are being bought by cryptofags
the real marketshare right now is something like 10% amd and 90% nvidia (not counting old gpus like hd 7000 series and below)
>>61997800
No, Nier: Automata is not the only game in Steam, anon.
Stop shitposting.
>>61997523
>Why do people here relentlessly defend AMD?
fanboys and shills
>>61978126
Daily reminder AMD wants to be seen as the good gamer guy but they do not have your interests at heart. The want the compute market/miners first priority. Gamers are at the bottom with whatever scraps they may give for an absurd cost, Sure nVidia has done some shitty stuff but they are focusing on gaming and progressing tech in games. AMD have used their underdog image to manipulate and market to their fans. Looks like I'll be going back to nvidia from a long break since geforce 4 days.
>>61997825
That's my point. It's bizarre /g/ is actually happy gamers are getting screwed. People here keep saying workstation is more profitable for AMD to focus on. That's probably true, but having 0% marketshare on Steam probably isn't a good thing for AMD, regardless of muh compute.
>>61997722
I'm talking about stuff llano, Trinity, Richland, Kaveri, Godavari and Carrizo. no one gave a shit about these things. Only consoles used AMD's mobile tech and Puma/Jaguar was based off an older arch, not Bulldozer. AMD's put a fuckton of R&D into APU's and it hasn't yielded much profit.
>>61997983
because it was shit
ryzen isn't anymore, it's like the best thing for mobile because it can get really low power usage
the only problem will be with the GPUs inside the APUs because vega is trash
>>61997983
Considering Zen empoys power management straight from Carrizo - the payoff is MORE than nice.
>>61978126
QUICK MOMMY I POSTED IT AGAIN
>>61994777
>Founder Edition means shit
I know, but when you see a review where Vega is going up against FE, that's what it is, a ~1700MHz 1080, which is a fair bit slower than than what anyone gets in the real world with custom cards.
>>61994745
A 1060 is over twice as fast as a 660, so yes, the bump is massive indeed. It's pretty much twice as fast as a 960, which is its direct predecessor. What you are saying is complete nonsense.
I got myself a GTX1070 for 360€, you mad?
>>61998130
Enjoy your buyers remorse :^)
>>61998098
moot point considering that the 480/580 is a better midrange gpu in every single way if you can get one at a reasonable price before the miners do
>>61998130
vega 56 is a all around better gpu, you fucked up
>>61998146
Ayyyylmao
>>61998157
Kek'd
>>61998194
>I like to eat SHIT, do you have a problem with that?
>>61998215
>trying this hard
Triggered much
>>61998233
well he's not wrong but i doubt you could have picked up a vega 56 for that price
i got myself a gtx1080 for 500 bucks and it was well worth it
>>61998243
I could have gotten a GTX1080 for 477€ myself, but then I'd need a new case.
Also I don't want to have a house fire.
>>61998291
my 1080 doesn't go over 70°C even at full load from the duration of full F1 GP
the card itself is really big indeed i have a full ATX and it's a tight fit
>>61998243
>you could have picked up a vega 56 for that price
where?
>>61998291
Where? Even the most generic GTX1070 are hard to find that cheap...
>>61998335
Well, you need to watch the prices 7 days a week to get yourself such an offer.
They do pop up sometimes.
Will probably end up between 1080 and 1080ti once drivers are good
>>61998382
and that will be when the gtx2080 releases with a better price than vega 64 and it will be useless again
>>61998382
No.
Not really.
It'll have like close to 4 times more worth of triangle throughput when the drivers will work.
>>61998394
AMD cuts the prices for Vega64, launches Vega20.
>>61998398
>vega20 with the performance of a gtx2080 but double the power draw and 30% more expensive
>vega20 gets driver optimised and fights with the gtx2080ti but only after nvidia releases gtx3080
>???
>amd is still behind
>>61998419
You dont understand anything, don't you?
>>61998439
i do understand everything and it will end like this, amd should have fixed the drivers before launch
inb4 the drivers give only a 5-10% boost
>>61998439
I understand that amdfags are always hyping shit and when it fails they do some damage control and the cycle repeats.
>>61998498
The only thing you understand is how to shitpost.
>>61998538
amdfags are top shitposters I agree
>>61983784
Put on glasses, you're blind. 4k upgrade was one of the most beneficial upgrades since SSDs you can get.
>>61983537
>http://www.tweaktown.com/news/55875/amd-launch-monster-navi-10-2019-next-gen-ram/index.html
>Jan 15, 2017 12:22 am
>according to my industry sources - Vega will be great
Oh boy
>>61998333
dude ,read the whole sentence
please
>>61999088
holy shit, i did read it but for some unknown reason i didn't see the word "doubt"
sorry m8, i'm currently on a high dose of caffeine because i didn't sleep right
>>61999142
no problem
>>61998774
I got 20/20 vision buddy and seen 4K displays in person. That jump from 1080p and 4K on 30-40" is noticeable but hardly earth-shattering unlike the jump from 480 to 1080p. It is a different story for massive displays. Displays that will unlikely find a home around a gaming PC.
The normies see that way as well and price tag and requirements for 4K shit is simply not worth it in their eyes. It is current fad among the A/V crowd trying to convince people to upgrade their 1080p shit when 3D failed at it.
4K gaming is a bloody meme on smaller screens. 4K and beyond for computing is only good for professional work (graphics, medical imaging, monitoring stations, coding etc.) where having more screen resolution is always welcomed.
/v/tards still thinking the market even gives a flying fuck about them.
>>61978126
Wait for Navi
>>61999191
These wojak edits are getting sillier and siller with every fucking day.
>>61999191
>mfw waiting for Navi ever since Maxwell/Hawaii/Fiji
Pascal is alright, but it's two whole node improvements instead of one. Polaris was ok too and has aged pretty well already. Navi/NvidiaMCM will eclipse everything these past 10 years. AMD's 460 could get extremely efficient, imagine a small 7nm chip of that size scaling up by putting 2-4 of those together on a MCM. AMD already has Infinity Fabric, they have a good headstart already.
>>61999219
i swear i just studied this picture longer than i should
gaymin???
this is /g
>open source drivers
>monstrous compute performance
what is there not to love about this card
>>61999376
Well, RX Vega IS a gayming card.
>>61999392
The problem is that /v/idiots aren't interested in learning about or discussing the aspect of GPU performance that would actually make it a /g/ related discussion: WHY.
They don't care how or why GPUs perform how they do, so the only thing they can do in GPU threads is shitpost because they don't actually understand how GPUs work at all, nor are they interested in learning.
>>61999663
Well at least ONE anon in these threads got actually interested in GPUs (GPU as in massively parallel computation device, and not a toy).
That's quite an achievement for a shithole that is /g/.
>>61999376
Locked BIOS
>>61999663
I think people here are still missing the point that RX Vega is explicitly a gaming card. Compute doesnt matter, if RX Vega cant game its a bad card.
>>62000053
You don't even understand that even the workstation version of Vega isn't just a compute card. It trades blows with a Quadro P6000 in 3D rendering applications.
The problem is that the gaming drivers just aren't even close to finished because they focused on getting the pro drivers feature complete first. The drivers are literally alpha tier, I mean, for fucks sake they don't even have primitive shaders enabled yet and without that implemented Vega just behaves like overclocked Polaris because it front end bottlenecks on 4 triangles/clock limits of GCN.
>>61999928
I'll concede that point, even though i've read there was an option to replace it on the fly under linux.
>>62000053
to me its a card that trades blows with the titan Xp in compute while being able to play games at 1080 levels of performance.
HBCC is the most exiting feature for me, being able to map RAM as virtual VRAM without having to change anything about the application software is huge!
>>62000353
>to me its a card that trades blows with the titan Xp in compute while being able to play games at 1080 levels of performance.
no one gives a single shit about anything compute related when it's a gaming card, and with current prices it's performance to price ratio is really low
jesus you're like those people who get in a pc os thread to spam it with "b-but muh android has a huge marketshare" or "b-but debian is really popular in servers" when it's clear people are talking about normal use on a normal pc
>>62000492
Learn English and proper punctuation, Rajeesh.
>>62000526
could you quote what was wrong in my post? thanks
>>62000653
I can't even read it, it makes my eyes bleed.
Seriosly, learn proper fucking punctuation.
>>62000130
No one gives a shit about the p6000 you fucking shill. RX Vega is a GAMING GPU. Vega 64 currently has a horrendous price to performance and it wont get better anytime soon. Its just like the Fury X, shitty drivers at launch made the card an overpriced disappointment. RX Vega is a failure regardless of compute.
>>62003175
>GAMING GPU
Why, are Quadro P6000s slow in games? Idiot.
>compute
Rendering 3d graphics is compute? LMAO
>>62003253
>replying to another shitposting retard
But why.
>>62003253
Stop shilling. Again, no one gives a shit about the p6000. If AMD can't make gaming drivers for a gaming GPU then fuck them, they deserve to fail. Im tired of early access hardware.
>>62003253
So, according to this, Vega 64 should be on par with a GTX 1080 TI or a Titan, but in reality it can't beat a GTX 1080.
Is that all simply down to their primitive shaders not working yet? I have my doubts about that.
On paper, the Fury X should have beaten the 980 TI, but it never managed to do so. I think it'll be the same thing with Vega, except this time AMD have been relegated one tier and still can't keep up with Nvidia's offering on that tier.
And that doesn't even mention how Vega 64 draws at least 100W more power than even OCed AIB GTX 1080s.
So we have
>Bad pricing
>Bad performance
>Bad power consumption
It's just a huge disappointment all around. After all of Raja's big talk and after the (almost) miracle that was Ryzen, I was almost certain that Vega would be a hit and that AMD would do to Nvidia what they did to Intel.
Turns out, Raja's talk was just talk and he never did manage to use a loo.
>1070 on newegg.ca is still hundreds more than it should be
WHEN WILL IT END
>>62003314
>impatient manchild gets angry over a faceless corporation prioritising less volatile markets over retardedly volatile ones
Whoa!
>>62003331
On paper, Fiji had too much ALUs given the triangle output.
There's more to GPUs than simply shader math.
>>62003280
If I can't have the faint hope of baiting someone into actually learning something, there is no reason to post in GPU threads on /g/ at all. Ever.
>>62003331
>Is that all simply down to their primitive shaders not working yet? I have my doubts about that.
>On paper, the Fury X should have beaten the 980 TI, but it never managed to do so.
Primitive shaders is literally designed specifically to address the reason the Fury X never beat the 980 Ti when it should have. Plus Vega has Polaris' primitive discard accelerator, whose benefits cannot really be seen right now because they come later in the rendering pipeline than the front end bottleneck issue. Plus Vega introduces DSBR, whose benefits can also not be seen yet because they happen even later in the pipeline than PDA and there is still a massive bottleneck in front of it.
If RTG really does actually get Vega to have an effective throughput of 17 triangles per clock with primitive shaders, then yes, Vega 64 should trade blows with full GP102 in games.
>>62003450
>If RTG really does actually get Vega to have an effective throughput of 17 triangles per clock with primitive shaders, then yes, Vega 64 should trade blows with full GP102 in games.
This is what I'm thinking and why I'm considering buying a Vega 64, but my trust in AMD has been shaken so much I'm not sure I'll go through with it.
Guess I'll wait for board partners to release their cards. If Sapphire or XFX create another marvel like the Fury Nitro I'll probably end up buying one. Or I'll just get the liquid cooled one.
>>62003500
Isn't CLC one a bundle-only offering?
Anyway Sapphire would definitely pull another Fury Nitro tier card.
>>62003500
>Guess I'll wait for board partners to release their cards
I would do that anyway, they'll have vastly better cooling. Who knows, maybe RTG will actually manage to get the drivers feature complete by September! Christ. Fucking madmen hardlaunching the card and paper launching the drivers.
>>62003547
They are (probably) staggering the launch for a reason. As you see, there's no partner cards announcements besides ANUS one.
>>62003500
>If RTG really does actually get Vega to have an effective throughput of 17 triangles per clock with primitive shaders, then yes, Vega 64 should trade blows with full GP102 in games.
I was originally aiming for a Vega64. Still that power consumption and the price of the ref card, which is already 60USD more than a decent AIB GTX1080, scare me off from buying any AIB Vega64 because AIBs will be definitely more expensive.
I may just end up with an AIB Vega56. Just don't know whether they are out at 28th August. And if any AIB Vega56 already costs more than any AIB GTX1080 I just don't have any legitimate reason not to buy a 1080 instead even when I'm using Linux.
>>62003644
>ever considering ref cards
Even though the PCB is absolutely stellar compared to trash that is nvidia refs, the cooling is shit.
>>62003672
By "aiming for a Vega64" I mean something like ASUS Strix Vega64. Unfortunately in Hong Kong, ASUS things, not just VGA cards, are always >100USD more expensive than similar products from other brands like Gigabyte. That doesn't matter though because Sapphire ref Vega64 here is selling 685USD even when there are limited stocks. Still I speculate that AIB Vega64 will cost even to 1080Ti level, which is around 709-767USD.
I really hope AIB Vega56 will come out as soon as 28th August and they will cost between 459 and 509 USD, but it seems both of them are just forlorn hopes.
>>62003913
We don't reeeeeealy know how AiB cards will be priced.
Right now there's severe lack of supply.
I don't really know why the fuck RX Vega was even selling, given its lackluster day0 performance.
where the fuck is vega48 for 299?
>>62003947
Maybe those buyers were just fanbois and miners. Just hope those AIB cards will be cheaper than what we have today.
>>62003994
Vega11 is probably Q1 2018. And it's going to be 32-36 CUs.
https://twitter.com/JayzTwoCents/status/899321072960512000