>Radeon VEGA chipped RX 580 benches at around that of a GTX

>>59076466
only if the price is good. the 1080Ti will be alone at the top, but it will be likely not worth the money they're asking

>>59076466
look at the settings, they don't seem to be maxed

if you can find a gtx 1080 benchmark with the same settings - that'd be a fair comparison

>estimated MSRP
Don't trust the pajeets, ever
That said I think it's totally reasonable to expect 1080-level performance

>>59076611
This. Full Vega should have nVida Titan X performance, It wouldn't be unreasonable to expect 1080-esque from a cut-down chip.

>>59076466
Ya a d the rx480 was supposed to be $200 as well
For months it was 250+

I seriously doubt AMD would sell a GPU that matches a 1080 for $250

A more realistic price would be $375+

>>59076635
>should have
Don't ever trust the pajeets, ever

>>59076653
not like you'd want the 1060 3gb or the 480 4gb anyway

>>59076635
Yeah I bet it should totally match maxwell
And it should be $108
And it should come with a free blowjob from Lisa
Shit I would pay twice that MSRP for a GPU like that

>>59076669
the 480 4gb for 175 is mighty fine my dude

>>59076466
1. Where'd this come from?
2. Does the 580 have HBM?
If so it ain't gonna be $250.

I can't wait for Vega to release so I can do a Zen/Vega build, but even I can't pretend HBM and its pricey interposer don't add significantly to cost.

>>59076730
Probably The 590/Fury will have HBM.
580 will be GDDR5(X?)

>>59076466
>want to buy RX480
>Ryzen turns to be actually pretty good
>now worried Vega will do the same not only on the high-end
Should I wait?

>>59076774
I dunno man I imagine Vega being their tip-top chip to replace the fury and MAYBE a cut down version filling the role of the 590, but not the midrange.

AMD could easily give Polaris the same treatment as the rest of the GCN line. It's designed from the get go as a high yield, low cost, low power chip. They could just scale it up and throw in some tweaks and easily make a 1070 competitor, maybe even a 1080 competitor.

We'll see come launch time, but full Vega is a BIG FUCKIN CHIP with some really expensive goodies in tow. It's probably gonna be a real force in the GPGPU market, but it ain't gonna be cheap.

>>59076895
if your getting by, yeah

>>59076653
>I seriously doubt AMD would sell a CPU that matches a 6900K for $500

>>59076466
this is most probably the polaris refresh the TX2 vega cant only be 15% faster than a 480

Wait for the benchmarks. /g/ went retarded before the RX480 released saying it was going to be a 1070/1080 competitor.

>>59077017
The official AMD releases for the 480 were actually spot on about its performance.
No, really.
Just they were misinterpreted (actually they were deliberately misleading).
I did the math on their 'crossfire AOTS benchmark' and found the actual performance of the 480 by comparing other AMD cards in crossfire DX12 AOTS.
If anyone else but me had a brain there wouldn't have been that level of hype.

>>59077038
That god for (((You)))

>>59076466

>glare quality low
>half resulution on terrain: on
>shadow quality medium
>1080p

the 970 can do that too

Honestly if vega is a 1080 +20% I'd take that over a $900 1080ti that blows its doors off

Unless we're talking 120fps on ultra at 4k it's just a waste of money, I'd take 144p 144hz any day over 4k/60

>>59076466
I'll do crossfire with Vega and Ryzen R7 1800x. 4k monitor setup fuck Nvidia and Intel fucking faggots.

>>59076466
What game is this?

>>59076466
>ashes of singularity
>that one game that no one cares about but AMD gets good results in because it is highly optimized towards them
Oh, right.

File: 1486486069897.jpg (66KB, 800x629px) Image search: [Google]

66KB, 800x629px

>>59076466
Not gonna get my hopes up. All I want is GTX 1070-like performance for around £300 that I can use with my FreeSync monitor. Anything better is a bonus.

>>59076895
I keep looking at RX 480 deals and I'm tempted. I have an R9 270 and a 1440p monitor, which basically means I need to use lowest settings to get 60 fps. Waiting for Vega is tough but I just keep telling myself it'll be worth it over the RX 480, particularly if I want to target 100+ Hz.

>>59076936
yeah I remember when people actually said and thought this lol

File: 1486966302356.jpg (226KB, 600x583px) Image search: [Google]

226KB, 600x583px

>>59076466
>RX 580
>5
They're going to fucking run out of numbers if they keep bumping the gen so fast

>AoTS
>lowered settings
Not even remotely tricked

>>59077554
It's a brand new architecture, like Zen is. I think they're justified in bumping the number. RX 4xx = Polaris, RX 5xx = Vega, although I wouldn't be surprised if we see lower-end RX 5xx cards using Polaris and being rebrands.

>>59077583
But then the 400 series didn't have any high end cards :^(

>>59077640
Yeah it sucks but Polaris doesn't scale that high in terms of performance. If there was an RX 490 it'd just be a rebranded 390X or Fury.

>>59077640

Was sad they didn't do anything for the high end stuff, but atleast the fury and fury x dropped in prices and that the 480 was actually good past all of it's controversy.

But if vega is getting the same treatment like ryzen, then i'll happily wait.

>>59077017
^^ This. Didnt you learn the 1st time round?

File: 1467295551467.png (49KB, 808x805px) Image search: [Google]

49KB, 808x805px

>>59076466
OH GOD

THEY'RE COMING FOR US TOO

Vega 11 is RX 580 with GTX 1070/1080 performance.

Vega 10 is RX 590 or Fury 2 or Rage Pro or whatever the AMD naming committee (currently comprised of single monkey who lives in a cage in the lobby) draws out of the hat. Performance around Titan X

Polaris 10 becomes RX570/560

>big vega xt =~titan/1080ti
>big vega pro =~ 1080
>little vega xt ~>1070
>little vega pro ~<1070

>polaris rebrands (10/12?) for below.

That sort of market segmentation makes the most sense in my head. It would be nice if vega does bring a huge performance uplift, but it's unlikely to be that much.

>>59076466
>$250.
Too good to be true

>>59076466
> >Estimated MSRP is $250
Doubt that. Price-wise, they can sell $400 RX580 (>1080), $300 RX570 (> 1070) and it's still going to be a bargain deal.
However, nobody except AMD knows how much it's gonna cost to make them, maybe they won't make a profit even at $300.

>>59076466
Even if it's a 500 dollarydoos it's still not as much as a 1080.

(If it delivers similar performances obviously)

>>59078742
I agree it's gonna be a 500 bucks card

But what's the point without CUDA ?

Only for gaymes ?

>>59076730
Again, I ask this allot. How much does hbm2 cost?
Because amd could eat the inflated cost somewhere in the we pay 10$ per gb of gddr5 price.

I say amd could eat it, but here is just my thought on why they would.

1) they have dick in the gpu market, at least consumer wise. Putting out a hbm2 gpu with insane power for end of product life prices... that could change the landscape.

2) amd is selling them mostly to businesses first and consumers will never touch high/good binned gpus. Though amd sold many gpus throughout the 480 and below life, the impact that only having low binned shit had hurt them for a long time, at least hurt them at the cost of after markets getting close to 300$ can hurt.

Both of these factors could easily combine into this
- the consumers are getting shafted on binning again
- they want to get market share, so why not subsidize the gpu a bit
- If they get a foothold with a gpu that is 1080+ power, how many people will upgrade for the next 3-4 years? they will have entrenched market share.

Do I think that they will go and sell it for 250$? fuck no, but 350-450 for the cut and uncut version, hell yes. if amd is making a 480 price level sku, they may have a 4 or 8gb version that is more heavily cut for 250~

Amd is doing something that seems close to their 'we got a new architecture' kind of performance, they have the shit that made the 900 series so efficiently implemented, and if they manage to saturate the gpu, they got a nearly untouchable gpu on their hands that conforms to the vulcan and dx12 standard.

Not saying I fully believe this, just thinking of what I would do give my market position, and the year + of building hype with an overwhelming success I have at my back.

>>59076895
If you need a gpu, get a 470, close enough to 480 performance and will get you by, resell it when vega comes along or use it as a backup in case shit happens.

If you can hold out, wait till vega or navi. Was personally going to wait for navi but life treated me good recently and I could be looking at vega.

>>59076466
>Estimated MSRP is $250.

I AM CUM

>>59077640
Polaris was a stop gap pipe cleaner product.

>>59077449
you mean the game that was the first to properly implement dx12 so its going to be a staple benchmark for years to come?

>>59078889
HBM2 apparently costs a lot. Nvidia is having a lot of trouble securing decent supply for their offering, whatever it is who gives a fuck, and in reality the interposer brings a whole new level of yield woes to the table.
On the launch of the fury X, a failed die was included in each box as a sort of keepsake. These dies were otherwise functional until they reached the interposer mounting process. The interposers themselves are so sensitive attempting to remove the cooler on a reference fury card can easily cause a micro fissure rendering the GPU completely worthless. On many custom fury variants such as the Sapphire Nitro, a sturctural support was added to improve stability for the massive coolers entirely because of this issue.

I'm absolutely sure the process has matured, but it doesn't change the fact there's an added layer of complexity involved with producing a GPU with HBM on top of the cost of te HBM itself.

Rebrand

>>59078963
Fuck AotS. That game was only relevant because the demo provided insanely versatile multi-gpu scaling and upon release didn't even offer support for multi gpu at all.
It was a neat tech demo at first, but should be recognized for its failures to deliver on promises rather than commended for the lies it spun in an attempt to bring attention to a mediocre RTS no one really cares about.

I want to see more examples like DOOM and much, much less like AotS.

>>59078963
>the first to properly implement dx12
"Properly implementing dx12" means optimizing for all DX12 capable platforms because you need to do that under DX12 seeing as a developer you have a greater responsibility at a lower level.

Which means the cases of games favoring one platform will get worse, not better.

Also, no one plays Ashes so no one cares about the benchmarks for it because they don't relate to any other game.

Is vega only newer high end cards or are we getting poolaris rebrands?

File: Boyz-II-Men.jpg (69KB, 596x479px) Image search: [Google]

69KB, 596x479px

>mfw 250$ in the US will turn into 300+€ in Europe

>>59079249
dead dies were given to the first batch of the radeon pro duos. It might not have been that many and the fury had already been out for months by that point.

So yes that happened, but it wasn't a very widespread thing. What did happen though was aggressive price cuts on the fury chips after both nano was underwhelming and then pascal launch later. 28nm bulk seems to have been pretty good by that point considering how the 980ti did as well.

File: pisscal rekt.jpg (79KB, 466x635px) Image search: [Google]

79KB, 466x635px

>>59076510
>>59076611
>>59076635
>>59076653
GET REKT GREENS <------ lmao that was a dank meme i r89 r9001 390/3.5gb of v for vendetta ram, novidiots be like do a 1080 and walk awayy lmao, 144 wayy sli novideo grillforce gtx 480 clocked at 3.5hz 1080p 3.5 fps gimpworks edition, human eye can see only 3/5 ram anyways, nivea water boiler titanal xp at 3.5p, envidia fryforce shitan rascal 1080shiTi at 87.5% GTX(GetTheXtinguisher) (ngreedia pisscal xp "malware experience" thermi burning sillayy bridges edition with GreedSync support) + krappy fake $hilltel gore i$emen at 7700K (Kelvin) maybelake i0 killowatt lake "cooler not included in a box" coffin lake "end of life" edition at 0.90 GHz. 8GiB? What do you need 4GiB for? I'm giving you 3.5GiB! mumble mumble 3.5!?!?!? mumble mumble mumble 390?!?!?!

File: RX480-Position.png (1MB, 1392x768px) Image search: [Google]

1MB, 1392x768px

Not Nvidia but AMD wrecked themselves.

>>59079438
what

>>59079438
thank you anon, I laughed a lot

>>59079438
>SweClockers
hehe mm

That image is fake by the way, the person just manipulated it to show a OCed 1080 as a Vega GPU.

If we're talking for real, I don't think Vega will be epic. AMD do the right thing now, they sell new "generations" even though they're barely improving. Why do they that - well, they don't have enough money to supply both CPU and GPU R&D departments. The last permormance improvement was 4 years ago, with 290X. Then they slapped more RAM on a 390X, then they did Fury and HBM - which was good, but even top of the line Nvidias offered better performance per dollar. Then they introduced Polaris, a GCN 14nm shrink with new video ports, but no improvements over the same 290X and now they're going to announce a Vega, which is Polaris with HBM. A final meme card.
If anything, it will be super power effective, but again: now when they have money from those Ryzen sales, they will deliver a next big card a couple of years after Vega.

>>59079249
Nvidia kind of fucked themselves not getting their hands dirty with the interposer tech early on. AMD has a straight up design information advantage over them for this.

Granted, it might not count for much considering Nvidia has $texas to throw at whatever problem they run into.

>>59079249
nvidia wanted hbm2 a year before it was even close to ready, and were willing to pay several thousand dollars for a few chips.

Hbm2 is now at or getting close to decent yields.

>>59079438
Quality post

>>59079286
aots is still the best case scenario for dx12, to implement most if not all of what dx12 has to offer that gpus can deal with.

because it was also the first to do this, it will be a staple benchmark for years to come.

kind of like crysis, and just cause 2, and crysis 2 for dx11

And relatively soon it will be a battleground for whats better, vulcan or dx12

>>59079350
properly implemented as in not giving negative performance gains across the board.

so a game that both stresses the cpu, and the gpu, and takes advantage of a new api, is not a valid benchmark tool...

>>59080079
> a game that both stresses the cpu, and the gpu
> is not a valid benchmark tool
Why don't you benchmark in CS 1.6 then?