>die size is around 520-530 mm'2 (compared to 314 on

>>58389263
maybe the 1080 is expensive and with a exaggerated profit margin

>>58389287
the profit margin would have to be insane for AMD to still be able to undercut it and make a nice profit themselves

>>58389297
AMD is known for not actually having a good profit despite selling more and better because of too low prices...

>>58389442
>selling more and better
except they don't retard

Flagship Vega 10 is competing against the Titan X Pascal.

Inb4 a torrent of shitposting, smug anime girls, and pepes.

>Much bigger than GP104
>Slower than GP104

AYYMD IS FINISHED & BANKRUPT

>TITAN X gets over 100FPS in Doom
>Vega HOUSEFIRES gets 60FPS with dips all the way to 30

AYYMDPOORFAGS CONFIRMED ON SUICIDE WATCH

>>58389263
They will still have decent profit margin. Since AMD's contract with glofo is priced solely by wafer and was priced literally years ago they can make stupid undercuts since TSMC charges nvidia almost ~30% above production costs, where AMD is still riding a contract from their divestiture of GloFo.

Nvidia's margins are still absolutely insane, and AMD is still making a small profit on selling the similarly sized fury series for sub 300 prices. Unless Nvidia goes into some insane price war (and consequentially nosedives their own profits) AMD should end up being okay. This would be a good thing however, and we would likely see 400 usd 1080's from it, so everyone should hope this happens.

File: AMD-Vega-GPU_2.jpg (46KB, 630x354px) Image search: [Google]

46KB, 630x354px

>>58389263
>520-530 mm'2
including the HBM2 stacks who take up around 2x92 mm^2 and leave around 350 mm^2 for the GPU
it has the HBM2 on the same chip

>>58390084
No, you retard.
HBM is on package, it is no on die. They are separate dies. That estimate is for the GPU die itself. Those HBM packages are a known size.

>>58389480
Vega 10 can't even compete with Maxwell's Titan. What makes you think it's going to stand a chance against the Titan XP?

>>58390121
Nobody knows final performance of Vega 10.
The only benches we have are from early dev boards that still have an extra 3 inches of power monitoring hardware.

>vega ad
>literal 12yo gaymur faggot
>37fps dips in doom
>"poor volta" while barely competing with 1080
This is why amd is a joke.

>>58389263
ANOTHER

is 687F:C1 confirmed Vega's id?

http://www.eteknix.com/first-amd-vega-10-benchmarks-revealed/

>>58390206
MASSIVE

>less fps than a gtx 1070

ayymd BTFO again
HAHAHAHAHA

>>58390223
DISAPPOINTMENT

File: lol 8pins.png (2MB, 1920x1080px) Image search: [Google]

2MB, 1920x1080px

Are those 2 peripheral ports on that semi-modular PSU connected to the GPU?

What PSU is that? Think we Anons can pool together and figure out if it's 6 or 8 pin?

>>58389503
>>58389517
>>58390206
>>58390223
>>58390236
I made this thread to have a serious discussion, can you kiddies run and go play with your Lego?

>>58389461
>because I say so

>>58390260
Do you know what market share is?

File: IMG_1090.jpg (19KB, 250x241px) Image search: [Google]

19KB, 250x241px

>>58390253
>According to the report from TechReport who got to thoroughly test the Vega sample with 8 GB VRAM in DOOM, this specific sample is somewhere in between the GTX 1070 and GTX 1080 levels of performance. When set to Nightmare settings, the game reported max frame times of 24.8ms. We also recorded some gameplay of DOOM running on the Vega sample while we were at AMD’s CES event.

>not even released
>already lost to 1 year old Nvidia mid range card

HAHAHAHAHA

AYYMD IS FINISHED & BANKRUPT

Ayymad lost again, what a surprise

File: IMG_8521.jpg (67KB, 258x360px) Image search: [Google]

67KB, 258x360px

>>58390252
House fire inductive pins

>>58390287
How is 1080 midrange?

>>58390342
Titan XP and 1080ti is high end. Your shit spilled yet again, Pajeet. HAHAHAHAHA

File: lelele.png (1MB, 1920x1080px) Image search: [Google]

1MB, 1920x1080px

>>58390252
It's a Thermaltake, boys!

>>58390467
somebody call 911

>>58389297
what's insane are the nvidiots who dish out 800 bux for a card that costs 19,99 and 2 dead chinks to manufacture / per batch of 1000.

>>58390731
poojeet seems constipated

>>58390731
yea but silicon is expensive man

So is ayymd releasing any rebrands so I cant get a cheaper rx 480?

>>58389263
I thought the die size was projected to be 470mm^2. Anyway, production costs are nothing compared to licensing and R&D. le die size meme only really rears it ugly head on immature processes.

>>58390817
rebranding is an industry standard so i'd say your chances are good

>>58390112
>>58390084
can't you count? measure mm2 of your fingernails multiply by 4 or 5 = die size

File: 32432.png (2MB, 1920x1080px) Image search: [Google]

2MB, 1920x1080px

>>58390252
>>58390467

>>58391083
just kill yourself

>>58391433
it's an extra pcb for analysis.

>>58391433
probably a troubleshooting/diagnostics addon

>>58391433
A port for monitoring every incoming and outgoing signal in the gpu

>>58389442
Pajeets lose their shit when they make even only a couple hundred rupees.

>>58392115
nice paint skills

>>58391810
>>58391857
>>58391987

Thanks m8s!

File: pork-sausage_link.jpg (43KB, 680x537px) Image search: [Google]

43KB, 680x537px

>>58390287
>Cutting off the rest of the quote which actually states it essentially in a sealed box so it was throttling and other shit which lead them to believe the performance would definetly improve

Fucking memesters need to leave /g/

>>58390361
Pretty sure it goes something like this, anon:

Low end: 1050, 1050 Ti
Mid range: 1060
High End: 1070, 1080
Enthusiast: TITAN X

File: tmp_12208-IMG_09831836212906.png (327KB, 486x683px) Image search: [Google]

327KB, 486x683px

>>58389671
Why doesn't AMD reacquire GloFo?

>>58393292
Foundries are an enormous money sink. It'd be an instant death sentence for AMD if they reabsorbed Global Foundries.
Even with their prospective game changing process IP, its no guarantee that they'll have enough customers of the process to turn a large profit.

>>58389263
I hope for AMDs sake the GPU running DOOM was in no way indicative of the final performance. If Vega 10 somehow has worse performance than the 1080 it'll be the biggest fuck up in GPU history since the FX "just leaf blow my shit up" 5800. The RX 480 is only @ little bit bigger than the GTX 1060, while clocking lower, and only being slightly behind in performance. Surely Vega can't somehow be a step backwards. Bulldozer and Prescott where both a thing that happened, so who am I to say.

>>58391433
Debugging port

>>58390287
>>58390335
Dumb frogposters

>>58393680
Engineering sample running Fiji drivers averaging 70fps in a segment where the Titan X Pascal averages about 80.
the Vega ES dropped to 35fps~ momentarily, and the Titan X Pascal will drop to 50.

Performance really isn't looking bad.

>>58390905
size is always a concern because silicon is a finite thing, you only have what is it, 300cm or 450cm wafers, each wafer is 25-50k and this sets the bare minimum a chip can cost.

granted larger chips have issues with the process not being perfect, and fucking some dies, so there is that too, this happens on every process through, no real way around it.

>>58393680
doom is only really iffy on showing performance because amd has such a massive uptick over their shitty opengl performance

Personally just assume nvidia and amd are equal with a 7-12% boost to amd due to async.

but at the same time, i'm assuming this is a pre good drivers look at the device, along with it being an es, and with the current thought of the big vega being 16gb hbm2, possibly the smaller card. looking at it that way, seams great.

price is everything though, and I don't think amd will fuck the price up.

>>58394206
A 14nm LPP wafer is between $6k and $8k depending on what client is buying them, and in what volume.

I still cannot see the benefits of HBM(2) as of now.

Nvidia doesn't have any problems with bandwith.

>>58395398
>Nvidia doesn't have any problems with bandwith.
Right. Thats why GP 100 uses HM2 for shits and giggles.
Protip: its not just about bandwidth, its about power.

What is somewhat more futureproof ?
Rx480
Gtx 1060

>>58395515
RX 480 no question. Already pulling ahead in DX12, and equaling the 1060 in most DX11 titles.

>>58395398
>I still cannot see the benefits of HBM(2) as of now.

They have double to triple the bandwidth, while using almost no power.

Remember Hawaii, the biggest housefire AMD card? They could almost double it in size and keep the same power usage, because removing GDDR5 gave them that much headroom.

>>58390163
>Nobody knows final performance of Vega 10.

There are four stages to the AYYMD shitposting cycle.

>Can't wait for (product) to come out, (competitor) is finished!
>It's pre-release, you don't know the final performance!
>It just came out, drivers will fix it!
>Just wait for (next generation product), then (competitor) is finished!

You are currently at stage 2.

>>58395618
The Fury series was a total failure.

Even drew more power than a R290X.

>>58389263
it will be good or cheap, but not both
> Zen might be though

>>58395423
You're fucking retarded. Nvidia uses HBM2 because it allows them to pack P100 on a small pcb.

>>58396734
fury x was shitty binning
fury was fucked chips that cant make it as a fury x
fury nano was so efficient they could put 2 of them on the same board and draw less power than a fury x.

>>58399165
>muh small PCB
No, tech illiterate retard, area savings are an insignificant and completely tertiary benefit to memory on interposer.
The radically higher bandwidth per watt, lower leakage from the memory PHY, and decreased die area are why Nvidia's GP100 uses HBM.

>>58396544
we are on stage 2 because faggots saw one dip on 5 month out hardware and are claiming thats it.

realistically, its between a 1080 and titan in performance, and with 8gb of hmb2, we are thinking its the smaller of the two gpus, as in the one that is 400 or less.

>>58399056
the 4000, better then nvidia by a mile, half the price of nvidia
the 5000, better then nvidia, cheaper by it by 150$
the 7000, better then nvidia, 50$ more,
the 7000 series again, the FAR better value then nvidia, and when oc'ed better
the 7000 series yet the fuck again, 500$ 200$ cheaper and only slightly behind,
years later, the 7000 series, on par to better then that same gpu, but had at 300$ or less
the 290x a solid card and beat the nvidia's of its time
the 290x again, traded blows with the nvidia's of the next gen
the 290x again over time, is beating out the nvidia's that came out after it up to the ti line
new api, well, 290x surpassing the ti of the next gen in some games...

yea kind of interesting to watch amd over time. the only time they cost more is when they are significantly better.

>>58390221
Yes, see those DOOM benches: https://www.youtube.com/watch?v=Iq-e1usnCiU

>>58389263
Can't wait for this poojeet company to finally go bankrupt.

>>58395515
GTX 1060 is already spoiled milk. It was a good buy at release too.

>>58399868

I want them to stay competitive. No better way to keep prices down and retain actual advancements.
Otherwise you're gonna get Intel's lineup when AMD stopped trying - 5% improvements every other year.

i can't be the only guy disappointed that there is still no fucking 4k60 ultra card on the horizon? we already know what the 1080ti can do based on the titan xp, and we know amd isn't going to be able to do much better.

call me when the 1180 Ti and the 590 are out.

>>58390286
>AMD has never had more market share than Jewidya
sure thing, Berg

>>58400149
Vega crushes the titan XP for only $350 dude