>290x Hype!... ...Disappointment >FuryX Hype!... ...Disappointment

>>61256428
Hi Jensen! How's V100 yields?

>>61256428
290x was pretty good anon, albeit a housfire

raja and the hbm meme has literally killed radeon

>>61256439
nvidia have been on top and stayed on top since the end of tahiti.
7970=680
290x<780ti
furyx<980ti
vega<1080 & 1080ti

>>61256468
>290x<780ti
HAHAHAHAHAHAHAHAHHA
Don't do that anon.

File: negative IPC vega.png (18KB, 824x394px) Image search: [Google]

18KB, 824x394px

>>61256438
breddy gud
we habe bositive IBC gains

>>61256483
Now show the SPEC.
Also who the fuck uses term IPC for bibeocardz?

>>61256428
like 10 years ago

>>61256468
not sure if bait or just stupid AMD user

File: 59664.png (37KB, 650x450px) Image search: [Google]

37KB, 650x450px

>>61256480
at launch it was. It might not be in 2017 but that's irrelevant. I'm not waiting 2 years for the card i spent $550 on to become good.

I hope AMD drops out of making gaming GPUs, they clearly don't give a fuck

>>61256488
it makes sense when both chips have the same 4096 processors. vega is just fiji with OC and 50-100W higher tdp

>>61256428
I've never owned ati/amd card for the last 15 years.

>>61256533
So moar ALUs makes a new uarch?
Dude you're genius.

>>61256523
people used to say that about the cpus before zen. Where is the radeon version of zen? we need some gpu competition.

File: 1440969217941.jpg (20KB, 945x388px) Image search: [Google]

20KB, 945x388px

>>61256523
Hi jen hsun

>>61256511
780ti was $700 and only marginally ahead of the 290x because Nvidia payed half the developers back then to use gameworks.
290x was a far superior card and would have been the standard for high end gaming PCs if it wasn't for Bitcoin.

>>61256523
Maybe you should wait for them to release a gaming GPU before saying that?

>>61256536
>Where is the radeon version of zen?
Probably in 2016 when they haven't diverted all R&D money to Zen and told RTG to make something from pennies.

Good decision might I say, Zen is far more important than any GPU

File: poo.png (312KB, 595x654px) Image search: [Google]

312KB, 595x654px

>>61256540
yes good sir buy amd and good drivers coming wait for it

>>61256535
no but if the only paper difference between 2 cards is the clockspeed and then when at the same clockspeeds they perform the same it suggests that there have been no architectural improvements. the least they could've done is chuck some more stream processors on it. although then it would've had an even higher tdp.

Does anyone find the situation of gamers foaming that they're not being focused on delicious? They literally cannot comprehend that GPUs are used for something else and it's driving them mad.

>>61256560
You need to go back.
>>61256562
You can mention cryptos and feel the rage.

File: instinctPage_02.jpg (533KB, 2600x1277px) Image search: [Google]

533KB, 2600x1277px

>>61256562
NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

>>61256428
I'm glad AMD finally have a competitor for the 980ti

>>61256569
>You can mention cryptos and feel the rage.
Kek, people are literally triggered they can't buy a GPU because someone else is buying them and making money doing it, there's a hint of both obsession and jealousy in the whole situation.

File: 1405865353819.gif (442KB, 420x594px) Image search: [Google]

442KB, 420x594px

>>61256428
Tbh it's dumb AMD fanboys own fault.

AMD has released 2 gaming benchmarks of Vega so far (Sniper elite and Doom) and both showed it at around 1080 level of performance, which is what you would get if you take the FE edition and add 10% for higher clocks and better drivers.

Ofcourse fanboys wanted to insist it would beat the 1080ti at a lower pricepoint even though there was no evidence at all to believe that.

It still is a massive disapointment though, performance at the level of a non oc'd 1080 with more power consumption than a 1080ti.

The blow is even harder because fanboys hyped up the product like they do each year

>>61256562
>They literally cannot comprehend that GPUs are used for something else
they're being used for calculating ur mum's weight
no end in sight

>>61256428
>>290x
>Hype!...
>...Disappointment

What are you talking about the 290X was great, bought in near launch and it only died a few weeks ago and it performed really well.
I replaced it with a 1070 so I guess the housefire thing is a meme since the 1070 runs hotter.

File: 780ti-bf1-1080p.png (57KB, 741x796px) Image search: [Google]

57KB, 741x796px

>>61256511
The 290X had overtaken the 780 Ti within six months of launch, having been barely behind in the first place. These days they're not even in the same ballpark.

>>61256562
tbqh the last time they went full enterprise we got faildozer.

>>61256603
Your point? Zen is clearly not Bulldozer.

>>61256603
Zen is full enterprise and we go, well, Zen.
EPYC totally shits on Intel.

>>61256468
not sure about fury x but all previous ones are now faster than the nvidia counterparts

>>61256610
zen was "desktop first"

my point is they haven't got a good track record of predicting markets.

>>61256624
Zen is desktop first? The fuck are you smoking?

>>61256624
Zen is clearly and evidently a server design, what the fuck are you smoking?

>>61256627
Zen wasn't dekstop first? Wtf are you smoking?

>>61256428
I still have 290X 8GB, it's still better than most shit on the market.
1070+ is better.
Nothing from AMD is better period.

>>61256632
Zen is clearly a server design, fuck, every Ryzen currently available uses leaky Zeppelins not suitable for EPYC.

>>61256428
where was the disappointment? those where cheap as fuck cards and performed great

>>61256641
That's not what he said you daft cunt.

He said it's dekstop first, meaning it was released for the desktop first

>>61256576
>Smart Home
What did they mean by this?

>>61256654
Because you needed to test and iron out bugs before launching Zen for servers you moron.

>>61256661
No shit you stupid cunt, that wasn't the point though

I'll grow out of AMD as soon as Nvidia supports freesync, stops encrypting their BIOS, and lets people disable GPU boost.
Until then, looking forward to vega.
I've actually owned a 560ti, a 760, a 1060, and a 980ti but since going high refresh last year and realizing how much I like it the value I place on adaptive sync has gone through the roof.
If you don't care about it though the 1080ti is actually good performance for the money.

>>61256747
>I'll grow out of AMD as soon as Nvidia supports freesync
This. The moment nvidia support freesync I'm jumping ship.

>>61256468
The HD7970/R9 280X got faster than the 780 in recent games.

>>61256818
>Pay top dollar for an AMD card that needs 5 years reach maximum performance

>>61256900
It's not reaching maximum performance. It's just aged better.

>>61256912
>It's just aged better
This is the worst thing to ever come out of r/AMD. There is no FineWine™, there is no aging. AMD releases cards that simply can't compete with their equivalent until a number of years later, and by then, it's pointless as newer cards that are cheaper perform better. So, it's not better aging, it's AMD incompetence.

Stop giving them a free pass to be shit.

>>61256944
>denying that kekler aged like fucking milk compared to GCN
Oh no you don't.

>>61256954
Kepler didn't age like milk, it was poorly designed to begin with, and Nvidia acknowledged that and changed it. Meanwhile AMD has been dragging its feet with GCN, praying that developers will do what AMD won't.

>>61256973
Kekler is poorly designed now?
Now that's a new excuse for it aging like shit.

>>61256428
With Vega the hype has been pretty much all negative

>1 year late
>will never beat 1080ti
>event if it beats 1080ti it will be crushed by volta within months
>will only match a 1080
>wont even match a 1080

So judging from history Vega should be a great success. It's also not at all impossible that this is AMDs plan this time around: to avoid showing a strong hand until it's absolutely necessary. Nvidia has beat AMD too many times with a prepared response (980ti to Fury).

>>61256979
>So judging from history Vega should be a great success.
Yeah, in 2021, when drivers are finally optimized.

>>61256428
>When did you grow out of AMD /g/?
Pretty sure most people grow out of Nvidia.

>>61256954
Kepler was great for its time. But by current standards it's not enough. The AMD equivalent was crap by comparison, and has only now caught up to kepler - an architecture that has been superceded by better architecture. This is AMD being way behind the 8-ball.

>>61257011
Eh, not really. It's not even about FPS, Nvidia simply has more features. AMD has made great strides to catch up of course but given Nvidia's marketshare advantage it makes sense that they have broader feature and 3rd party support. There are programs like Moonlight which is probably the best free mobile streaming app out there, and even has VR support now. Gsync is way more expensive and mostly a ripoff, however Nvidia has Lightboost on older 3d monitors and ULMB for more modern monitors, which is useful for those who hate sample and hold. There's Nvidia Inspector, nothing on ATI/AMD's side even comes close, Radeon Pro has been abandoned. And even the official Nvidia drivers have better features, I recently sold a GTX 980 used to some guy. He was switching from a 390 to a 980. You know why? Because AMD's aspect ratio scaler in their drivers wasn't working properly, he was a CS:GO player and needed 4:3 support. Nvidia's aspect ratio scaling is fantastic, that's just something I had always taken for granted.

The point is for niche use cases, Nvidia makes more sense than AMD.

>>61257037
Are you referring to the GCN architecture or the 7900 series in particular? Because the GCN architecture has received very few CU changes until Vega and still successfully combats everything prior to Pascal. GCN is much more scalable and has the potential to utilize more ILP than both Kepler and Maxwell and well before they were around. Pascal had to add more silicon and aggressive software optimizations e.g. pixel level preemption to par AMD's level of shader parallelism. Meanwhile on the geometry and RBE... Different story. Vega is just catching up to Pascal with Conservative Rasterization, ordered views, DSBA, primitive discard (from polaris) and TBDR.

>>61256912
It is reaching max performance not aging. AMD cards "should" be faster than nvidia at launch but they aren't and they take years to perform as they should. The fury x was an 8 teraflop beast while the 980ti was only 5.8teraflops. Even today the drivers still aren't good enough to show the furyx outperforming the 980ti as it should with that level of compute discrepancy between them. The 1080 is also only 8 teraflops so in theory the furyx will eventually perform the same as the 1080 (if vram wasn't a factor) but since amd have essentialy dropped support for the furyx since they sold so few that will never happen. "finewine" is good if you're in the market for a used card but if you're literally buying a worse card based on the speculation that it might perform better in two years time you're an idiot. "Finewine" aging has no guarantee and tends to not even occur if the card is not rebadged. The 7970 and 290x aged well because they were both rebadged to the 280x and 390x respectively. As vega yields are looking to be very low and the silicon is looking to be very expensive the likelihood of a rebadge is extremely low and vega is doomed to age just a averagely/poorly as the fury x did.

>>61257158
>It is reaching max performance not aging. AMD cards "should" be faster than nvidia at launch but they aren't and they take years to perform as they should.
>"finewine" is good if you're in the market for a used card but if you're literally buying a worse card based on the speculation that it might perform better in two years time you're an idiot.
It's refreshing to see common sense on this depraved shithole. Just a shame the fanboys won't see the reason in it.

>>61256468
The 780ti is quite a bit weaker than even the R9 290
The 290x outperforms the 980 in many games

>>61256428
For the record the R9 290 was an amazing value card at launch. Whooped the more expensive GTX 780. Can confirm it is a housefire though since own a reference cooler one.

>>61257158
Not sure where this myth is coming in that Fury support has been dropped any more so than any other AMD chip. Last I checked Fiji was still holding its own.

I agree that FineWine(tm) is overblown and not a real reason to buy an AMD card. It's not intentional on AMD's part. They've just been making chips with strong as fuck compute capabilities that brute force their way through shit for longer than Nvidia's chips can. Nvidia has always been heavily tied to software and driver optimization and to their credit it works extremely well for them.

>>61256428
290X and the Fury (non-x) weren't disappointments

Vega looks fucking awful though

>>61256428
My R9 290 works fine though, not a disappointment although it seems to be the GTX 480 from AMD when it comes to noise, power use and temps

>>61256783
>3.91ghz
Intel is absolutely fucked.

File: t3_6d8s2h.gif (2MB, 400x225px) Image search: [Google]

2MB, 400x225px

>>61256428
Woah there faggot the 290x was a beast and still is. That card can play any game on ultra today in 2017 as long as its 1080p.

>>61257683
Fury X was a pretty big disappointment IMO.

7870 and its 270x refresh was gud
7970 and its 280x refresh was gud
290x and its 390x refresh was an insane beast monster
fury and fury X was a big fat fucking flop though no reason to own that piece of shit card it wasn't a big enough upgrade over the 290/390

>>61256428
>290x
>Hype!...
>...Disappointment
>FuryX
>Hype!...
>...Disappointment
>Vega
>Hype!...
>...Disappointment

Yet the people that did buy those cards.. 95% of them didnt regret their decision because a R9 290X is still competitive.

Life of AMD cards is long...

A owner of a gtx 1080 here, i had a gtx 970 before that. wish i had gotten a Fury X instead of these two....

>>61256944
This, waitfags fucking rekt

>>61257099

>4:3

>>61257872
Non-x fury was decent, though

performance between 980 and 980ti for 980 (sometimes less) prices, with 4GB HBM instead of GDDR5, it was decent value and competitive

god knows what they were thinking with the Fury-X though

>>61257882
still no reason to upgrade from a 290x at 1080p so I don't care what happens with vega

File: perfrel_2560.gif (69KB, 400x1057px) Image search: [Google]

69KB, 400x1057px

>>61256511
290X was a good card, it got a shit rep because the reference models had jet engine, shit coolers which couldn't even stop it from throttling. Aftermarket cards were much, much quieter and performed better too since they were no longer kneecapped by the abysmal cooling. Even so the 290X was slightly faster than a Titan at high res when it released, so it was a fine card.

>>61257920
What does having HBM versus GDDR5 add though if it doesn't translate into higher performance?

In theory lower power consumption maybe but Maxwell was much better in that regard, even with GDDR5

File: LastLightFury.png (15KB, 599x553px) Image search: [Google]

15KB, 599x553px

>>61257965
Non-X Fury was a better performing card than the 980, though...
>Pay GTX 980 money
>get a card that was better than the 980
how was this a failure?

>>61257998
I'm pretty sure it was similarly priced to the 980, and I'm also not so sure if it outcompeted it in every situation

In terms of price/performance the regular fury was definately better than the x though, I agree

>>61256944
t.nvidia shill
- From a GTX 1080 owner

>>61258023
it was ~$450

>>61258029
I feel like nvidia was holding back since the 580 and the 1000 series is the first time they've made cards that actually push the envelope

the 680/780/980 cards were so fucking weak and shit tier

>>61258023
The Fury is better in like 9/10 benchmarks than the 980

http://www.anandtech.com/show/9421/the-amd-radeon-r9-fury-review-feat-sapphire-asus

The reason I know so much is that I was a 980 owner that got burned by the release of the R9 Fury so it is all still fairly fresh in my memory, it was simply a better card, mind you, the 980 was a trash card to begin with and barely justifiable over the GTX 970, so it was low hanging fruit for AMD

>>61256428
>When did you grow out of AMD /g/?

> Bought a Phenom X2
> it was a piece of shit
> But, let's give it another chance
> Buy a 7870
> it's a piece of shit

Fuck me, never again.

>>61258073
Allright, I didn't know too much about the 980.

Only Maxwell card I owned was the 980ti, and have a 1080 right now

>>61258091
why did you upgrade?

>>61256428
>>290x

I have a 290x and don't have any problem with it...

>>61258059
>and the 1000 series is the first time they've made cards that actually push the envelope
you mean rehashed maxwell? sure they managed to add clocks due 14nm move, but pascal is the most boring release from nvidia in recent decade

volta looks the same, maxwell was glorious though

>>61257960
I honestly credit Hawaii for some of the really nice aftermarket coolers that we have today. AIB partners had to step up their game and for the most part they did.

>>61257998
The problem with the Air Fury was that it was never in abundant enough supply to actually compete with the 980 and it came much later. If you could find one at the right price it was a no-brainer but they were hard to come by. By the time those Sapphire Nitro Fury variants were on a firesale at Newegg (for like $289 or so late last year) it was already getting tough to justify spending that much money on a 4 GB card, even if the GPU itself was powerful as hell.

Also, Fury non-X was competing with the unjustifiably (at the time) priced 390X at $400 which really never should have existed at that price point as the 390 was a much better buy at $330 or so.

In the end there probably just weren't enough Fiji chips to go around. It was 600 fucking square mm.

>>61258108
the 1080 and 1080 ti are very strong cards
meanwhile the 980 could barely beat 290x which was a generation behind

>>61258104
Was having lag in some games at 1440p at ultra settings, so I sold it, invested another 120$ on top of it and got a used 1080 with almost 2 years of warranty left on it.

Very happy with the purchase, lag is gone, power consumption is fantastic, and the cards runs very cool

>>61258029
Actually, no, not a shill. I've had GTX470, R9 280, R9 380, GTX970 and now currently RX 480. Had to RMA the 280 3 times, and on the third time they sent back a 380. The 380 also had to be RMA'd 3 times, and on the third time they sent back the GTX970. Still have the GTX470 and use it as a back up for when/if a card needs to be RMA'd, gave the 970 to my brother and bought a 144Hz freesync monitor and RX 480.

>Happy with the GTX470 - still works fine
>Unhappy with 280
>Unhappy with the 380
>Happy with 970 - still works fine
>Happy with the 480 - works fine, but can only use driver 16.11.2 because every newer driver introduces a problem

File: u r.png (139KB, 1371x597px) Image search: [Google]

139KB, 1371x597px

>>61258029
>>61258151
Forgot the screenshot

>>61258126
just because AMD had to redesign whole gpu line nvidia looks strong right now
nvidia will have too do the same soon if they want in MCM game that amd started

remember when they had to scrap whole GPU line? something similar happened to amd now

>>61256593
I hear that once they get 64GB of HBM2 on a GPU they will finally have enough available memory with a high enough bandwidth that they will able to calculate a thousandth part of his mom's weight and then be able to extrapolate the rest.

>>61258166
Except for the last 5 years Nvidia Always managed to release high-end cards that could compete with AMD. Sadly it seems that this time around AMD looks a generation behind on performance and performance/watt.

And Volta will only make the gap bigger

>>61258166
MCM is so far off though, it's even farther off than volta

>>61256468
>vega<1080 & 1080ti

look at the price to perf

AMDfag

/g/ is the wring place to shill, go do that in /sjw/ the anons are retards there.

>>61256428
B-but what about muh CPU? FX-4300 is so gorgeous.

>>61258208
We can't look at price/performance since the price is not know yet, idiot

File: 796098790_193432.gif (372KB, 240x280px) Image search: [Google]

372KB, 240x280px

>>61256428
GEE I WONDER WHO COULD BE BEHIND THIS POST

>>61258193
>last 5 years
yeah, when do you think they started working on new GCN and two consoles?

>>61258203
coming to you q1 2019

>>61258222
You're not funny

File: 1480093420510.png (497KB, 579x640px) Image search: [Google]

497KB, 579x640px

>>61258222
Mshallahal malahl, brother

File: allahu-akbar-intel.jpg (29KB, 500x500px) Image search: [Google]

29KB, 500x500px

>>61258275
Salaam alaikum!

File: 1497603683431.jpg (30KB, 521x278px) Image search: [Google]

30KB, 521x278px

>>61256428
YOU STILL DIDN'T GET IT !!
Its not simply Benchmark test results its all together cost to Yell, Power consumption to Yeld. not only much FPS '''
And specially much depends from drivers code optimization etc.

NVidia bribe game devs with free stuff and much support only to put into code of graphics engine stuff to slows down opponents...

AMD making good stuff but not over expensive thats' why all new consoles using AMD APU.

It is like PC versus Aple Stuff is stuff and it works, some people will go PC .. some will go Trendy over expensive Aple, and that it.

If RX Vega can be bought for $499 it's basically a day one buy for me. It'll BTFO the 1080.

>>61258108
How is Pascal boring? The 1060 is the first x60 card in a while to significantly improve over the last gen (660, 760 and 960 were all basically the same thing), the 1070 was 980Ti performance for significantly cheaper plus more VRAM, the 1080 is about 30% faster than the 980, and the 1080Ti is the first consumer GPU to hit 4k/60 with near maxed out settings on a majority of modern games. Even the budget option, 1050Ti, is interesting. The 1050 and 1060 3GB are the only boring GPU's. I'm really not sure what people expected. How come AMD gets a pass when their refreshes improve performance by like 5%, but Nvidia's "refreshes" offer a performance bump equal to an entirely new arch and people still whine.

>>61258304
>AMD making good stuff but not over expensive thats' why all new consoles using AMD APU.
AMD is making the hardware for consoles because Nvidia and Intel know there's virtually no profit to be made.

>>61258270
This has to be a bot at this point. It replies to every single happy merchant pic.

>>61258383
That's why Nvidia was so happy to get the switch.

>>61256562
feelsgoodman, the constant dick-measuring over 7 frame-differences in their AAA trash """hobby""" needs checked daily

File: b3867bf0_vbattach235410.jpg (205KB, 1600x900px) Image search: [Google]

205KB, 1600x900px

>>61256428

>>61256428
>290x
>Hype!...
>...Disappointment
The rest are true. The R9 290X was/is god tier though so kys

>>61258481
The Switch let them dump all of their unused Tegras on Nintendo. They didn't have to do any R&D, the technology was already there.

>>61258335
>The 1060 is the first x60 card in a while to significantly improve over the last gen
so is 480, see my point? amd had to rework everything so nvidia looks good

technology wise pascal is boring as hell, nothing new just more clocks and ALU

>>61257011
Yeah I had to have my 6800gt back when those things were hype as fuck, I bought a 390 about a year and a half ago and haven't looked back since. Adaptive frame sync is fucking amazing and Freesync is a better deal than G(oy)sync by hundreds of dollars so it's a no-brainer.

File: CHEd7MCXAAEeL5D.jpg (35KB, 553x369px) Image search: [Google]

35KB, 553x369px

>>61258538
Too bad about all those tegras in the Shield tablets and toaster ovens they made.

Is there a technical reason AMD couldn't push Polaris any harder? Why didn't they release a 480X or something that could match the 1070/980Ti/Fury X?

>>61258598
it's essentially normal GCN with less power consumption
also 14nmLP can't clock very high
vega got the clocks because pipeline is redesigned

>>61258538
>Nintendo is Nvidia's bitch
>AMD is Sony and Microsoft's bitch
Really makes you think...

>>61258552
>Freesync is a better deal than G(oy)sync
Except goysync can be used in windowed mode, whereas freesync is restricted to exclusive fullscreen mode only.

>>61258714
anon..it's not for about 6 months.
welcome to the future! yay!

>>61258383
Their margins were expected to pass 20% and they get funding for R&D. That's how most of polaris was developed

When will based Vega be available for purchase?

>>61256428
>shilling corporations
Why is this a thing? What kind of sad loveless life do you lead?

>>61258774
30 july to 3 august

>>61258714
>still trying to troll with outdated info
must suck to be uninformed, don't you feel foolish

>>61258714
GYSNC SUCKS, IT USES FORCED VSYNC
AMD LETS YOU TURN VSYNC ON OR OFF WITH FREESYNC, BEST OF BOTH WORLDS BABY.

>>61258616
>14nm LP can't clock very high

So about that Nvidia GTX 1060 made at GloFo?

>>61258596
wait, Nvidia used Tegra CPUs to make these little ovens? um wtf?

I want an AMD easy bake oven too! It should be kawaii!

>>61258714
lol this old song and dance
next you'll be telling me HDMI2 isn't supported either :^)

File: 348s.jpg (20KB, 348x348px) Image search: [Google]

20KB, 348x348px

>white
>pale as an albino
>no muscle definition whatsoever
THE WAY IT'S MEANT TO BE PLAYED

File: NV-tattoo.jpg (7KB, 288x210px) Image search: [Google]

7KB, 288x210px

another fat guy
another nvidia tattoo
these fuckers should save the ink money and cop a gym membership

File: BwEgjKlIcAAOOa4.jpg (65KB, 600x600px) Image search: [Google]

65KB, 600x600px

at least this poor bastard has the beard genetics to cover his nonexistent jaw and chin.

>>61258909
this guy's brain and heart ventricles are overclocked too high. he shaves his dome for extra cooling.

>>61258511
>>61258596
>>61258878
>>61258909
>>61258929
does ANYONE but fat ugly guys like Nvidia products?

I totally don't understand why people make these sort of threads and posts.

Why would you proclaim to someone else they shouldn't buy something. What's the point in the first place. Everybody should be glad AMD still makes cards even solely for the competition it brings to the table, keeping nVidia on their toes and maybe forcing them into lowering prices or investing more in better graphic cards themselves.

There's absolutely no point in hostility towards competition when you're the fucking consumer.

https://youtu.be/Vr-P9xcifT4?list=PLx15eYqzJifcv2LU02X_UkeGjBoXN-9Lt

>>61258966
hostility towards Nvidia is fine because they go above and beyond to screw their cuckstomers out of every penny they can, IE charging hundreds of extra dollars for Gsync for absolutely no reason whatsoever.

>>61256979
Vega would of been a success if it launched 2 months ago rather than 2 months from now as it appears to be doing. AMD is late to the party since NVIDIA s half way throught it's refresh cycle and it will be a matter of just waiting 2-3 months for volta, something that will beat the current line (and vega) up by 20% guaranteed. Why risk buying amd?

Went from 290x to 1080ti after I saw how bad Vega was going to be. Don't regret it one bit.

>>61259011
>ah the old, "people only buy the most expensive flagship card" line.
nah, people buy the lower/middle end card mostly, which is where AMD is dominating, even without coin miners.

>>61259023
lol sure you did, I bet that 1080P 60 hz monitor of yours is perfectly fine too.

File: 1236690855685.jpg (90KB, 402x348px) Image search: [Google]

90KB, 402x348px

There is nothing wrong if RTG died.
Nvidia can be a GPU maker while the third party make different model with varying prices

I don't know why you guys are so autistic defending AMD, most companies or division died and rise all the time and the world still moving own.
We use to get more than just two GPU company before now, and no one freak out like this autistic child board back then when they gone.

Get a fucking life.

>>61257924
This, Navi or bust.

Also by late 2018 when navi comes out we will have 4k monitors with freesync 2.

File: What now faggot.png (2MB, 1919x541px) Image search: [Google]

2MB, 1919x541px

>>61259040
What now faggot? I have you know I sold my 290x for 250 USD as well so an upgrade to 1080ti wasn't that expensive.

>>61259006
And people who buy Nvidia are fully aware of that. If they chose to desire to still purchase their shit they should regardless of what AMD does. Maybe they're totally not interested in adaptive monitor refresh rates. Whether it's Freesync or Gsync.

Nvidia being able to charge bucketloads for their cards is exactly whats currently bad in the consumer market. AMD hasn't released anything really in high-end ballpark segment for years now. If the tables where turned AMD would've done exactly the same, don't kid yourself into thinking that might not be true.

And that's why monopoly is bad and competition is good.

>>61258166
Vega should of been an in house prototype used to help advance to the neXT level. Not a full fledged step and consumer product. AMD is fall ing for the Intel trap of tiny improvements each gen which means green will btfo of them every time.

>>61259031
>which is where AMD is dominating
And yet nvidia will ship more units

File: 1402904467267.png (289KB, 655x606px) Image search: [Google]

289KB, 655x606px

>>61259053
>I don't know why

because you're fucking retarded. why do you think there are only 2 big GPU vendors left in the first place?

>wanting a monopoly in the first place

>>61259031
Except the 1080 and 1070 are more popular for gaming than the 480 according to Steam. Polaris cards are mainly selling due to mining. High end cards are way more popular than people here think. The PC market is slowly segmenting into people who buy low end GPU's and laptops with integrated graphics, and people who want the best of the best. AMD's target market, budget gamers who still have some money, is shrinking, while Nvidia's target market, enthusiasts with lots of money, is growing.

>>61259159
There is a niche group of games who want their GPU to act like a heater who still buy AMD. 300W pull = comfy winters.

>>61258309
> $499
Seems a bit high no? I would expect $350 if they intend to compete at all. Especially with those bench marks.

>>61259159

>Except the 1080 and 1070 are more popular for gaming than the 480 according to Steam.
because there's nothing above a 480/580, so if you want more performance there's no other option than going nvidia.
popularity means fuck all if there's only one choice.

>>61259200
Don't blame Nvidia for AMD being incompetent.

>>61259159
The 1060 is by far the most popular GPU in the Steam survey. More people are rocking Nvidia's "budget gamer" card than 1070, 1080 AND ti owners all combined.

You're assertion of Nvidia being targeted towards people who spend 500 bucks + on video cards is pretty baseless by your own metrics. Which are pretty dodgy metrics to begin with seeing as AMD seems completely misrepresented, for instance, the RX 580 isn't even on the list.

>>61259200
That's my point, people say the 480 is successful because it appeals towards budget gamers, but it's still not a huge marketshare despite having good price/performance. AMD lucked out that there was another mining boom. Midrange GPU's aren't making any money anymore, the 1070 and 1080 sold like hotcakes and Nvidia's high margins on enthusiast-tier GPU's means higher cash flow.

>>61258724
>>61258808
>>61258822
>>61258854
Goysync in window mode works for me

>>61259252
That's probably because the 580 just released. The mining craze that has completely decimated the number in stock probably doesn't help either.

>>61259358
Well there will be a flood of cheap 580s in the market post boom. I am definitely picking one up for a 2nd build.

>>61259358
Or Geforce experience makes the Steam survey prompt more often.

>>61259258
GPUs are like toys for company that has something like EPYC.

>>61256428
When will you grow out of overhype?

As a Fury X owner, I'm still happy with my card, though the fact that some games are STILL produced poorly optimised for AMD hardware does shadow its true power somewhat.
Take Fallout 4 compared to the Witcher 3 for example. In TW3, you can turn off the settings which aren't AMD friendly, which means that the card performs where it should.
Fallout 4 however, sure, you can turn off the godrays - and fight with it every time you start up the game, and if you go to Boston? Ha, goodbye FPS counter, you're going down faster than a young woman near Bill Clinton, and not by your own choice.

If it wasn't for instances like that, I wouldn't even consider leaving my Fury X. As it stands, I know Vega won't provide, and I doubt Navi will do much better at this point, so I don't know where to stand. The other issue is I bought a Fury X specifically because I've got a Freesync monitor, and thus while still an option, Nvidia isn't a great option for me either.

>>61256483
It's pretty impressive that they got 50% or higher clocks in the same TDP

>>61259157
There's a lot of monopoly happening in this world,
Only /g/ retarded enough to take this shit seriously.

It's sad because AMD makes really good budget cards, butt people still buy Nvidia because they can't be bothered to do research. Still using my 270x, but I'm hoping Vega turns out well. The watts are my biggest worry right now.

Btw did they ever get crossfire working in window mode?

I'll just leave this here.

https://www.youtube.com/watch?v=PUR8l2tOZ3g

>>61256944

I'd rather buy a GPU that gets better the longer I own it (AMD) than buy a GPU that gets worse the longer I own it (Nvidia).

The 7970 initially competed with, and was inferior to, the 680. Now it outperforms the 780. In the short term the 680 was the better buy. In the long term the 7970 was the better buy.

Vega RX will be better than expected (not that difficult given the current circumstances with FE). Probably cheaper than expected. And more available than expected.

This is going to be Ryzen 2.0 for AMD with reviewers being overly skeptical at launch all the while RX Vega still going to be popular by demand.

Low expectations while having a good product is exactly what any manufacturer wants. Why would you want to cause a massive disappointment yourself by overly hyping your product. Let the consumers themselves decide when they can actually buy your product.

290x was legit i still use one and get 100+ fps in battlefront 2 alpha

>>61258335
>How come AMD gets a pass when their refresh
Because amdrones are desperate for a new card to waste money on.

>>61258714
>freesync is restricted to exclusive fullscreen mode only.
Try again shill

>>61256428
>When did you grow out of AMD /g/?
Never because I'm against corporate jew cancer (Intel, Nvidia).

>>61256428
>290x
The 290x uses the 780ti as a cock sock and is now comparable to the GTX 980.

>>61259184
No one cares about max power draw. O no! My computer pulls the power of 4 light bulbs instead of 3. Truly an am bankrupt!

>>61259200
Which is why amd needs a titan killer, but seems to be incapable of doing so, even when given an extra 8 months to do so.

>>61259184
We don't even know what the TDP of RX Vega is, fuck off shill.

For the price, before etherium mining, my rx480 has done me better than any nvidia card would

GPUs are now just another profit venue, not their main product.
They should and will divert R&D to CPUs.
GPUs are still toys to most of the market compared to CPUs (80B vs 15B TAM)

>>61260130
AMD's the company that has an army of corporate drones willing to defend what's shaping up to be the most disappointing GPU launch in recent memory.
>I-it's not for gaming! Drivers will fix it!

File: 1499418278869.png (224KB, 1273x667px) Image search: [Google]

224KB, 1273x667px

>>61260511

>>61260511
A disappointment is still better than a scam in my book (GTX 970), so please keep defending that cancerous company.

>>61260505
Retarded gaymurs kept buying NVidya even when AMD had the better product. The entire market is a waste of time, every consumer in it is a mouthbreather that falls for the dumbest shit. AMD should just pull out of the AIB market and focus on consoles, APUs, workstation, prosumer, etc.

>>61260556
>2xx rebranded to 3xx oc
>4xx rebranded to 4xx oc
>amd is better hurr durr

>>61260165
>Power draw doesn't matter!
>Heat doesn't matter!
>The only thing that matters is the GIGAHURTS!
AMDrones appear to be taking cues from Intel these days.

File: 1458848012974.gif (1MB, 300x169px) Image search: [Google]

1MB, 300x169px

>>61260675

>>61256428
Can someone give me a quick rundown on why Vega is embarrassingly lacking in performance? I mean what is the point of Vega and where does it stand in the market?

>>61260713
>Can someone give me a quick rundown on why Vega is embarrassingly lacking in performance?
Bulk of it's uarch features is disabled in drivers.
>I mean what is the point of Vega and where does it stand in the market?
It's worth for iGPU IP alone.

>>61260675
>Release update to your GPU lineup at the exact same price point
>Still get shit on by retards who prefer getting scammed out of performance

This is why consumer AIB market is a colossal waste of time. Brainwashed consumers have all been given full frontal lobotomies.

>>61260703
>muh 3.5 gib

>>61260734
>It's worth for iGPU IP alone.
I meant the FE anon.

>>61260680
It would be funnier, if Vega could beat a 1080.
It's a less efficient arch and can't push the numbers even Pascal can.
It's already over. Nvidia could delay Volta by a year and still be ahead.

File: 1491612867984.gif (1004KB, 300x169px) Image search: [Google]

1004KB, 300x169px

>>61260703
>>61260748
>>61260675
>not posting the correct version

>>61260759
They said Vega is Q2 2017 and FE is Q2 2017.

>>61260748
>Defending the shit company even after they lost in a court of law

You NVidya shills should be fucking exterminated like the cockroaches you are.

https://www.gtx970settlement.com/Home.aspx

>>61260713
The focus is on mobile and APU's. AMD has essentially abandoned the high end market to Nvidia. What's weird is I thought they wanted to combat Nvidia in the professional and GPGPU markets, but if the new Radeon Pro's are just Vega FE's with certified drivers I don't see why anyone would switch from Quadros. Radeon Pros would have to be like half the price of Quadros to compensate for the fact that performance isn't as good and it can only do CUDA through emulation.

>>61260799
>the new Radeon Pro's are just Vega FE's with certified drivers
Source?

>>61260817
he said "if" so he isn't claiming anything.

>>61260675
Nvidia never leaves when they entered the rectum of a normalfag.

>>61260799
>but if the new Radeon Pro's are just Vega FE's with certified drivers
What else would it be? Vega10 is Vega10. WX Vega will have certified drivers and software profiles.
>CUDA through emulation
>HIP
>emulation
Hands off dat crack boyo.

>>61260817
SR-IOV, certification, ECC, and whatever workstation goodies the $1000 extra gives you.

>>61256483
>AMD its going to be around 1070 > 1080
It barely beat a 1070, this shit better be around $350 MSRP.

>>61260817
It's just a guess. Radeon Pro Duo 2016 is 2 Fury X's in Crossfire so I can see the new Radeon Pro being based off of Vega FE. I'm just wondering what incentive there would be to switch to Radeon Pro over a Quadro if it's weaker than the Titan Xp in a lot of professional benchmarks.

>>61260903
>$350 MSRP
That'd be neat.
I should probably sell my R9 Fury while people are still paying €400+ for it. God bless the EU.

>>61256533
>le vega is fiji die shrink maymay
Sure, and Pascal is Maxwell rebrand too, retard.

File: vega-v-furyx-specviewperf.png (32KB, 808x594px) Image search: [Google]

32KB, 808x594px

>>61256483
Seems the only part of the drivers that are actually finished are the workstation parts.

>>61261079
Pascal is Maxwell: moar ALUs edition.

File: WAeuoHz.png (353KB, 2880x1620px) Image search: [Google]

353KB, 2880x1620px

This is official btw, and only a few months old, far latter after AMD knew if there was a problem with Vega.

Just wait.
Vega will either be shit or it will be fucking amazing, 3 weeks to go.

>>61261285
Can lower RX 480 30 watts with no performance loss.

>>61261285
It took nvidia 4 years to reach GCN2 levels of compute density. I guess they're even.

>>61261302
And that's hurting AMD's image, they they get more units from the wafers but they bin less aggressively, meanwhile nobody really know how much units Nvidia throws away because they couldn't reach a certain voltage target.

>>61256428
>290x
>disappointment
What?

the 290x is literally the best AMD gpu in the last 10 years, still competitive in games and mining even 4 years after its release. Can you say the same thing about its contemporaries, the 780/780 ti?

>>61261754
>the 290x is literally the best AMD gpu in the last 10 years
That's not 5850.

>>61256428
a very long time ago
i don't care about consumer technology basically
and i've yet to see a high end server with any amd hardware the past few years

>>61261821
https://www.supermicro.com/products/nfo/AMD_SP3.cfm
?

>>61261754
It's also literally the last time AMD actually made something new in that GPU range, meanwhile Nvidia has made two more generations of -80 tier cards while AMD has been the GPU equivalent of Intel just rebranding the same shit over and over again.

>>61261884
Why would AMD waste R&D money when GPUs are barely fucking profitable for them?
No one buys Radeon since forever.
Making ebin halo products when no one buys them makes no fucking sense.

File: 1430725832538.gif (253KB, 447x415px) Image search: [Google]

253KB, 447x415px

>>61259890
>than buy a GPU that gets worse the longer I own it (Nvidia).
Can this meme fucking end already

https://www.youtube.com/watch?v=DvCsT_dk-1E

https://www.youtube.com/watch?v=dh9Akfu90Gs

Nvidia cards do not get worse over time you fucking troglodytes.

The only reason AMD cards from 4 years ago got better is because AMD is still using the same fucking architecture for the last 4 years, which is why they are miles behind Nvidia right now and can't even compete in the high end segment anymore.

>>61261909
Their GPUs don't sell because they're not good. Ryzen is selling because it is good, they just need to make a GPU that is actually good as well (and no, releasing rebrands and faulty products that need 6+ months of waiting for driver updates isn't good).

>>61261940
Frogposter being retarded as usual, all in order.

>>61261947
>Their GPUs don't sell because they're not good.
What? Fucking R700 into Evergreen failed to help Radeon break past 50% marketshare.
It's pointless.
Also CPUs and GPUs are vastly different. GPUs are toys aimed at (man)children while CPUs have more general audience.

>>61261884
something like 80% of GPU's sold are in the $150-$300 bracket

flagship products are a huge money sink, not worth it

>>61261970
Stick your head in the sand and don't take a look at the actual evidence like the simpleton you are

>>61256533
it makes sense if its the same uarch
wich we already know its not

>>61261821
it may exist, but i havent seen it in the real world yet

>>61262177
meant for >>61261843

>>61256979
>to avoid showing a strong hand until it's absolutely necessary
How desperate do you have to be to honestly think this? When have AMD GPUs ever exceeded expectations on release?

>>61262177
Of course ytou haven't seen them since the platform is like 2weeks old.
>>61262213
>When have AMD GPUs ever exceeded expectations on release?
R700? Evergreen? Tahiti?
Also fucking R300.

>>61262228
>R700
They took a different approach and had cards that came close to the performance of the competition for lower cost, they never exceeded expectations in raw performance.

>Evergreen
Same as above

>Tahiti
Yeah definitely exceeded expectations with power consumption.

>R300
Back when ATI were still on their feet. G80 is still killing AMD, it's NVIDIA's Core architecture and they are still refinining it and pushing it to the limit before they move on to something brand new.

>>61262295
>they never exceeded expectations in raw performance
Oh they did, given the die size.
>Same as above
Evergreen literally skullfucked nVidia so fucking hard it resulted in leather jacket man rushing out thermi.
>Yeah definitely exceeded expectations with power consumption.
Well memed.
>G80 is still killing AMD, it's NVIDIA's Core architecture and they are still refinining it and pushing it to the limit before they move on to something brand new.
And you know what happened to Intel that polished fucking P6 for eternity?

>>61262350
>And you know what happened to Intel that polished fucking P6 for eternity?
RTG's equivalent of Ryzen is a long way off, Vega is a flop.

>>61262356
>Vega is a flop
Come on. I know you can shitpost better.

>>61262375
Whatever helps you sleep at night, 300-375W TDP, 500mm2 die, HBM2 and GTX 1070 level performance is a flop in anyone's book.

>b-but FE is not a gaming GPU!!!!
When in the history of ever has a non-consumer or 'gaming' GPU performed terribly in gaming?

>>61256428
290x was great fuck off. Only the reference cards were a problem and Nvidiots just ran with it because they got bullied so hard with Fermi even though it wasn't the same case at all.

>>61262406
Well that's a better shitpost!
But i know you can do better. Much better.

File: 1441337594807.png (71KB, 220x220px) Image search: [Google]

71KB, 220x220px

>>61262422
You just don't know when you stop embarrassing yourself do you?

>>61261754
Tahiti(those fucking 1300MHz+ monsters) > Hawaii > Evergreen

In terms of longevity

File: 1471073089004.png (115KB, 273x302px) Image search: [Google]

115KB, 273x302px

>>61262437
That's worse than the previous post.

>>61262406
>>61262437
>vega FE crushes the fury X in anything that isn't gaming
>"BUT I SWEAR THAT DRIVERS WON'T FIX IT FOR GAMING! IT'S DOA!!"

File: 1476302163700.jpg (46KB, 625x625px) Image search: [Google]

46KB, 625x625px

>>61262461
It crushes it only in SPEC.
Probably the only thing that works perfectly on it right now. Also
>giving him replies

>>61262461
>unironically thinking drivers will 'fix' anything
Game optimizations are for new games, if the card performs terribly in games that are a few years old or more, it's not going to magically improve no matter how delusional and new to tech you are.

>>61262481
>i dunno what TBR is
(You).

>>61262454
>>61262473
>posting idoltrash
>being this much of a filthy casual
Get off my board you fucking redditor.

>>61262489
You mean that 'feature' they conveniently removed from the Vega Architecture marketing information on their website?

Just because NVIDIA can pull it off and get improvements from it doesn't mean RTG can.

>>61262490
>le ebin lebbit maymay
(You)

File: 1446797336943.png (610KB, 817x720px) Image search: [Google]

610KB, 817x720px

>>61262524
>proving my point
You can't actually be this retarded can you?

>>61262513
>Just because NVIDIA can pull it off and get improvements from it doesn't mean RTG can.
>Just because powerVR can pull it off and get improvements from it doesn't mean NVIDIA can.
That's you.

>>61262553
NVIDIA can and did, RTG clearly didn't because VEGA is terrible at just about everything you throw at it. Its only saving grace is the half rate FP16.

>>61262513
ATi has used TBR in Adreno and Xenos you stupid child.
PowerVR had it in 1995.

Holy shit

>>61262567
double rate*

They even gimped the FP64 on their 'prosumer' card, they could have one upped NVIDIA with decent FP64 compute like NVIDIA once did with the Titan.

>>61262567
Man this is getting boring.
You can go back to /v/ and shitpost there. People will even give you replies. More replies than i'd ever give you.

>>61256617
and miles behind the cards made to replace them. if you have to wait years for your card to reach another manufacturers card's performance is it really a good thing. especially when two generations have passed in between

>>61262585
It's still 100 times better than 1/64 FP16 on anything but P100.

>>61262597
I guarantee you don't even use anything that requires FP16 and RTG already have the Instinct series for AI/Machine Learning, there's no reason for the FE to exist other than to show everyone how terrible an architecture it is.

>>61262607
FE only exists to fuel your shiposting and meet Q2 2017 target for Vega.
Stop shitposting.

>>61258203
um, not sure if you know this but amd is mcming the thread ripper and epyc, and vega has all the hardware necessary to do it on the gpu side too, it seems to me to be a real world test bed for all the tech in a gpu so they can get as much data as possible.

The moment amd mcm's their gpu, they can have entire new lineups every 6-9 months

>>61262625
>just WAIT
lol

>>61259131
we will know that in a few months if the drivers are not in a fall back mode anymore, or if the hardware does have a real flaw.

>>61262631
>The moment amd mcm's their gpu, they can have entire new lineups every 6-9 months
NVIDIA are already working on this and have published papers regarding it. At this rate they will probably beat AMD to it in the GPU realm.

File: 1.jpg (57KB, 455x400px) Image search: [Google]

57KB, 455x400px

>>61262633
>*shitposts loudly*

>>61262643
>a fall back mode
Even AMD reps call this a meme, stop parroting this garbage as an excuse for the performance.

>>61262655
Stop shitposting.

>>61260931
Honestly, after looking at prices for the Fury X, I'm really fucking tempted to sell mine.
The only issue I have is mine is on a custom watercooled loop and the block I've got on it cost me some money, and selling it would probably confuse retards who don't read descriptions properly and try to put it in dry.

>>61262651
It was simulated, not even a real fucking silicon. nVidia is yet to prove their ability to engineer any kind of complex NUMA system.

>>61262651
ah, not likely, at least not in non compute loads. also, considering volta has a 800mm die chip, its not happening in voltas life time, while it could happen in navis.

the issues with mcm are large, and amd has experience doing it and doing it well, where as nvidia has next to nothing in that regard, so expect theirs to have more issues than amds.

>>61262674
They're definitely more capable than RTG.

>>61261125
>workstation card has drivers optimized for workstation parts
holy shit

>>61262687
Wrong, FE is a gaming card

AKa "Gaming mode"

>>61262686
Nah, they were late to everything, even to unified shaders.

File: 7e7.jpg (62KB, 600x516px) Image search: [Google]

62KB, 600x516px

>>61258283
>Jewtel
>congratulating Ramadan

>>61262700
FE is a "prosumer" card (thanks Huang for new meme words).

>>61262687
>I don't know how drivers work: the post
>>>/v/

>>61262700
It's a "hybrid" card, like the Pro Duo was. Either way this 'gaming card' and 'workstation card' meme is no excuse, Quadros do just as well in games as Geforce cards, Firepros historically kept up with their Radeon counterparts. Retards who know nothing about GPUs have just made this fact up.

>>61262705
Not to 12_1 feature level apparently

>>61262700
wasn't the gaming mode doing nothing at all?

>>61262718
No such market exists, it's a gaming card with gaming drivers throgh and through

>>61262726
>Not to 12_1 feature level apparently
That's nvidia's stuff. Aka no one cares.
>>61262731
Tell that to Huang who created and killed the only prosumer card evar aka the OG Titan.

>>61262663
amd reps claim the drivers 'are not gimped' which was a fucking retarded question.
it's like asking 'did you hold back the drivers on purpose' and amd said no
obviously amd did not, the right question would have been 'are all the features for vega currently implemented and working properly in the drivers'

Seeing as we know its useing fijis render method and not the tile based rasterization through tests, and we know for a fucking fact there is a tile based rasterizer in it, something isn't working and Its likely the drivers for visuals aren't ready yet, or at least aren't stable yet. when nvidia went to tile based, they had to scrap a fuck load of their drivers completely, and we all know amd doesn't have the best driver team ever.

>>61262730
Yes. In fact FE shipped with drivers that disable most Vega features.

>>61262730
The gaming mode didn't do anything with the Radeon Pro Duo either, it's a gimmick.

>>61262756
>That's nvidia's stuff. Aka no one cares.
Apparently RTG do because Vega finally has 12_1 FL.

>>61262761
>amd reps claim the drivers 'are not gimped' which was a fucking retarded question.
https://forum.beyond3d.com/threads/amd-vega-10-vega-11-and-vega-20-rumors-and-discussion.59649/page-138#post-1990723

>>61262718
um... prosumer has been a word for decades, it means not professional but not consumer garbage either, you saw it more in tools or cooking appliances

>>61262771
>Yes. In fact FE shipped with drivers that disable most Vega features.
[citation needed]

>>61262772
>https://forum.beyond3d.com/threads/amd-vega-10-vega-11-and-vega-20-rumors-and-discussion.59649/page-138#post-1990723
And Vega FE still works like Fiji!
What a wonder.
>>61262780
Yes and nVidia was the first to release prosumer GPU and then kill it since it was eating Quadro sales.

Are Vega preorders up yet?

>>61262789
gamersnexus and pcper benchmarks.

>>61262792
It's a new driver, not a 'Fiji' driver or a 'fall back' driver, fuck off already.

>>61262812
Yes it works like Fiji. What a wonder, aint it?

>>61262772
he is being a cunt about how we are describing whats happening, I fail to see how that's a refutation, and i'm also assuming they are part of pr, would like to know if their words hold more weight than that.

>>61262823
He's under the NDA to say anything of value.

>>61262792
amd could still legitimately call their high end cards prosumer as they don't gimp the 16 or 64 flops.

>>61262811
They just show the card performs poorly, nothing about drivers purposely disabling features.

Let me spell it out for you because you're tech illiterate and fucking stupid. You can't 'disable' tile-based rasterization with driers, it's a hardware-level feature and is how the hardware handles the task, not how the drivers tell it to do it. I'd like to see someone modify Maxwell or Pascal drivers to make them do full screen rasterizers.

>>61262821
Welcome to GPU drivers, now stop perpetuating this 'it performs badly because Fiji drivers' meme.

>>61262849
>Welcome to GPU drivers, now stop perpetuating this 'it performs badly because Fiji drivers' meme.
It performs badly because it behaves like fucking Fiji.
>>61262839
I don't think RX Vega will ship with 8-Hi stacks.

>>61262829
If you work for the company you aren't really under an nda, but he would likely hold back everything they say, which is why you have to word questions in ways it allows them to say something without saying something at the same time giving you solid information.

>>61262839
FP64 is already 1/16 rate, same as Polaris and Fiji 'gaming' cards.

Someone brought up the idea that the geometry throughput 2x per clock and TBR are tied to the primitive shaders.
Which is weird since Raja did say those features will be developed driver side and not need implementing by devs.

In any case, seems like the biggest changes are not working in FE at the moment, be it TBR, more geometry throughput and HBCC

The first one is the nastiest though, it leads to higher power usage and a bandwidth bottleneck Vega is currently experiencing over 1450MHz

>>61262849
> I'd like to see someone modify Maxwell or Pascal drivers to make them do full screen rasterizers.
They do it when TBR is inefficient.
Yes, you need to configure and enable it in drivers.
>>61262870
And B3D dwellers are too autistic to ask him a proper question.

>>61262862
it likely wont, amd showed off the vram working even with 2gb in deus ex at max, they are either trying to use less vram or make cards with less vram viable once we go to 4k. there may be a 16gb version but it would be the pinnacle version.

>>61262862
It performs badly because it's a bad architecture.

They also claimed 11 polys per clock and it's still 4 just like Fiji.

Basically they lied.

>>61262839
Even fp64 in GCN needs more registers than what consumer cards have, that's why there's no fp64 chip for a while after Hawaii.

fp16 doesn't, it's just a 32bit pipe split into two.

>>61262878
HBCC seems working. But we don't know HOW it works.
>>61262893
Hello Jensen! How's V100 yields?

>>61262904
Hey Raja, how's that 0.2% marketshare in the datacenter?

>>61262917
When is Nvidia making a CPU?

>>61262917
>GPUs in datacenter
The finest memes brought to you by nVidia™.

>>61262876
cant find vegas double rate, and I know that amd has been skimping double for a while, but the titan x has

single - 10157
double - 317
half - 159

at the very least polaris amd cards have equal half precision to their single.

>>61262921
Never since they can't even make a non-housefire SoC even with REFERENCE ARM cores.

>>61262897
nvidia gimps the fuck out of their half precision on consumer cards.

>>61262921
GPUs are growing far faster than CPUs, CPUs are a dying market

>>61262955
Even on fucking Quadros. Only P100 has 2:1 FP16.
>>61262963
> CPUs are a dying market
HAHAHAHAHAHAHAHHA.
FUCK.
Shitposting so loud i can barely hear myself laughing.

>>61262955
There's actually no reason for AMD to not gimp fp16 on RX, in fact I think they will put it at 1/4 or something, usable, but nowhere near the FE

>>61262955
And you don't even need that so why do you care?

AMD would do it too if they were in the market position to get away with it, it's 100% useless to consumers.

>>61262963
Yeah, you can claim that when Intel isn't posting revenue 7 times larger than you.
From those dying CPUs.

GPUs in datacenter are useless outside of a few niche applications.

>>61262963
Call me up when your Goy Processing Unit can serve TBs of media and documents

>>61263016
>your Goy Processing Unit
the irony of you probably posting this on a machine with an intel processor

>>61263024
Intel makes decent SSDs and good NICs. nVidia makes toys and memes.

>>61263024
Intel is in the business of making more than gaymer gear.

>>61262977
>>61262963
the gpu market can only replace at best 30% of the server market, the rest of it the gpus just are not suited for it.

>>61263039
Wrong, Intel make the best SSDs. Samshit is housefire trash that only performs at the speeds it advertises for 3 seconds. Samshit is more reliable in having the least reliability and performance consistency.

>>61262981
If they did anything like that it would likely be polaris like where its just equal half flops.

>>61263049
>30% of the server market
Only in wettest dreams of JHH. Like, ARM will be more relevant than GPUs in datacenters soon enough.
>>61263059
t. intlel shill

>>61263048
Yeah they're in the business of ripping you off, failing in the mobile market and making terrible acquisitions because their management is a mess.

>>61263071
>t. intlel shill
Posted from my Ryzen system.

>>61262983
they were in positions where they were better or near equal in the past to nvidia in market share and they still never did anything truly shitty.

File: SIA_World_semiconductor_revenues_May.jpg?w=1600.jpg (318KB, 1600x1348px) Image search: [Google]

318KB, 1600x1348px

>>61262963
Audible kek

>>61263082
Intel still makes quality NICs and good SSDs. It's still 146% better than being worlds first meme company.

>>61263089
That isn't truly shitty it's called market segmentation.

Removing features that are useless for consumers from consumer products makes perfect sense. If you're one of those Linux autists who sits around thinking he needs workstatiion hardware then you should pay for workstation hardware and not pretending your gaming hardware is something it isn't.

GPUs are good at self-crashing memes and some workloads that can run a billion cores and don't have serial code, for rest it's a joke.

>>61263106
Again, the best SSDS. Their enterprise grade SSDS (700 series) are unrivaled in reliability and performance consistency.

>>61263136
Enterprise I agree, their consumer ones are junk, and use unholy combinations of siliconmotion controllers and sometimes not even their own NAND

>>61256428

I'm still running my 290x. Great card.

>>61263136
I was talking about consumer stuff tho.
Their NIC are the shining example of what consumer NICs should be: simple as fucking stick and jsut werks.

>>61263071
that 30% isn't necessarily far fetched, much of what servers are used for are data processing and in those cases where it's not about ram storage limits its about a cluster of slow but many cpus, an area that gpus excel.

>>61263122
If you are purposely gimping hardware that could otherwise do a task to not do it because you didn't pay enough, that is fucking shitty, you can call it whatever you want. I don't care when double precision gets taken out because that requires actual die space, but half precision do don't believe does.

>>61263140
You shouldn't even buy any other SSD, if you really need something cheap then MLC drives from Toshiba are your best bet.

>>61263146
>I was talking about consumer stuff tho.
Intel 730/750 series just use enterprise grade controllers, they're still consumer drivers and have proper enterprise equivalents.

>>61263156
GPUs excel at simplistic number crunching. Anything more difficult and with branching makes them fucking useless.
GPUs are fucking toys.

>>61263191
and much of the demand for large clusters of slow but many cpus only need simple number crunching, 30% may be an overestimate either by amd or intel, it was one of them that said that figure, but It is likely an accurate estimate, especially looking at movie cg and the farms they use.

>>61263218
>especially looking at movie cg and the farms they use
That's HPC-tier niche.
As i said, GPUs are toys. nVidia should be happy to be IBM's cumdump.

Also what happened to Tegra?

>>61263269
Being used in the Switch

>>61263294
That's it?

>>61263307
It's literally billions and billions of dollars in revenue.

>>61256593
Clever girl

>>61263315
It's fucking nothing since nVidia dumps them for peanuts. Because no one else uses them.

File: Screenshot from 07-07-2017 22:38:09.png (438KB, 1625x898px) Image search: [Google]

438KB, 1625x898px

You mean to tell me this all is suddenly GONE in the span of 3-4 months?
You mean to tell me that half a year after tapeout and AMD announcing features, they'll all suddenly GONE a few months latter?

Yeah, no, that makes as much sense as Zen ending up with 20% more IPC over BD

>>61263356
They are not gone. The features (besides ROPs being clients of L2 and maybe HBCC) lack driver implementation.
RTG software team is lagging soooooooooooooooooooooooooooooooooooooooo far behind hardware dudes it's embarassing.

>>61263356
Actually working but shit implementation(worst case, very unlikely looking at current tests)
Dark silicon, pretty bad, but can be fixed with a respin.
Simply not enabled, likely.

Combination of 3? I have no idea, probably just as likely

>>61263418
It's definetly not a silicon bug since pajeet&friends had enough time to do even base layer respin and smoke some more cubans.

Give me something between 1080 and 1080ti for $400 and you got yourself a buyer AMD.

>>61263460
That's maybe Vega11 and it's much-much later. Just buy nVidia if you want card here and now.

>>61263476
Vega 11's release date is currently only RUMORED
If they can't get a Vega 11 out soon(2 months) that would imply that Vega 10 yields are very good.