Where were you when VEGA won? Nvidia finished and bankrupt.

Not this shit AGAIN

Fake news

Alternative facts

>>60193347
>>60193375
>>60193381
>3ip's
>4posts
:thinking:

It's fake like all the other "leaks"

FAKE!!!
Nvidia won't let this happen

>>60193320

VEGA YASS!
JEWVIDIA BTFO!
Based red team.

>>60193320
AMD shills so desperate they literally start making up benchmarks

File: ???.png (299KB, 512x374px) Image search: [Google]

299KB, 512x374px

>>60193320
post link or fake news

NO NO NONONONO FAKE DELETETHIS

$500.

Nail. Coffin. Volta. Kek.

>>60193347
>>60193375
>>60193381
>>60193736
>>60194011
>>60194155
>>60195814
OYY VEY

If they were targeting Volta from the start, this makes perfect sense.

It'll probably be lower but not by much, Vega is a new arch on top of a dieshrink, Nvidia's last arch was Maxwell and Fiji was only like 7% perf/watt worse than Maxwell there.

>>60193320
Oy-vey, delet this goy.

File: 4maAVRQTt9Q_fpvyhL9Li_HiHGnLVdi2U31C7hAD-Jo.jpg?w=1024&s=420c4eef6de74fcc3d8fb180a6fa88c0.jpg (143KB, 1024x702px) Image search: [Google]

143KB, 1024x702px

>>60193320
That silence must feel like a chokehold to Nvidia

>>60196102

Nvidia has been spooked for a long time.

>>60196131
Nah, they just went full retard. Wanna ANOTHER Titan?

>>60196139

Ever since Mantle got adopted. They've been counting their days.

>>60196139
I don't understand the Titan issue, Titan XP1 was released early because Nvidia didn't have yields for a fully enabled GP102 die, they only got a full die in production recently hence the P6000 and Titan XP2 release was so late after the first Pascal Titan.

It's not strange, it's a huge amount of work to get such a huge die running on 16nm, AMD's Vega is even bigger by 15% so it's gonna take them slightly more time.

>>60196165
Well yeah, when MS went with Mantle-like apporach and said Mantle itself was donated to Khronos group, it went really-really bad for NVIDIA because their hardware is not suited for newer APIs.

>>60196183
It serves no fucking purpose because no one will buy it anyway. At least P6000 has a market audience.

>>60196279
Flagships rarely serve a purpose, they're made in small quantities to increase brand recognition.

>>60193320
>no link

FUCK OFF AMDRONE

>>60196039
Not sure about that, took me 5 minutes editing the DOM.

>>60197018
Putting that aside, the last time AMD brought us a massive arch revamp we got Zen and Tahiti, that some managed to even push to 1350MHz on air.

AMD seems to have a knack for new architectures (bulldozer not withstanding, that was their only fuckup)

>>60193320
>leaked info about vega losing
Ayymd shit and bankrupt
>leaked info about vega winning
FAKE NEWS

File: makebelievesign.jpg (58KB, 525x463px) Image search: [Google]

58KB, 525x463px

>>60196039
>Fiji was only like 7% perf/watt worse than Maxwell there.

>>60197121
They aren't exactly calling this a new arch, though. They're calling it "the biggest architecture update since GCN".

So how the fuck does it compare to anything?

File: Untitled.jpg (192KB, 800x601px) Image search: [Google]

192KB, 800x601px

>>60197191
>it's the board pierro

>>60197257
The Maxwell GTX 750 ti perf/watt can't be matched by anything AMD has and it came out like 3 years ago. 7% lol

>>60197299
>GTX 750 ti
It's almost like a breathing joke.

File: amdramadan.png (507KB, 578x644px) Image search: [Google]

507KB, 578x644px

>>60197339
>breathing joke
Whatever the fuck that means lol.

File: 374.jpg (76KB, 709x520px) Image search: [Google]

76KB, 709x520px

>>60197379
The answer lies in the mirror.

File: perfwatt_2560_1440.png (29KB, 500x650px) Image search: [Google]

29KB, 500x650px

>>60197299
Clearly at the high end, they were 9% behind.
So no, Nvidia never had a huge perf/watt advantage there, a huge perf/watt advantage would be evergreen vs fermi.

You seem to like making an ass outta yourself.

>>60197445
TDP AMD 300
TPD Nvidia 250

That's over 20% , not 9%

>>60197189
no link

>>60197464
If you think TDP means what you think it means I can show you a 15W CPU drawing over 30 watts.

Also everyone measured TDP differently, TDP also isn't TBP

Man, i really hope Nvidia's plants inside AMD can destroy all the engineering samples and wipe their data on vega.

Having something like this exist is too dangerous.

>>60197445
I think people don't appreciate what AMD did with Fiji, it was basically two Tongas glued together with the same frontend as a much smaller Hawaii while still on 28nm, but a lot more efficient that it was only slightly behind Maxwell that was the biggest rewamp since Fermi.

Vega on the other hand is AMD's Maxwell from the arch side, so a huge jump should be expected.

>>60197596
People don't appreciate AMD doing anything.

>>60197619
Ignoring idiots, there are still plenty of technically minded people here who can actually diagnose a whitepaper or core layout.

>>60197639
>Ignoring idiots
And only 5% of /g/ left.

>>60197658
/g/ isn't a small board, so that's enough for a proper discussion.

Didn't your mom teach you not to talk to idiots? You'll become an idiot yourself.

>>60197671
I don't talk to idiots but said idots usually derail threads into shitfests with no possibility of proper discussion.

YAAAAAAAAAAS AMD SLAAAAAAY

>>60197596
Fiji was rushed, it was supposed to end up on 20nm but we know what happened there, and cutting out the frontend was their choice when they had to go back to 28nm.


Honestly I find it amazing that AMD is spending all its new transistor from 14nm finfets on the frontend, they probably want to make it futureproof for a while, there's clearly more than enough ALU power in Fiji

File: 1489475219786.jpg (11KB, 259x194px) Image search: [Google]

11KB, 259x194px

>>60193320

>>60197735
>4096 fucking ALU
You fucking bet.

File: Radeon-RX-Vega-TimeSpy-Benchmark[1].png (320KB, 1223x724px) Image search: [Google]

320KB, 1223x724px

>>60193320
and here the real results

File: LRM_EXPORT_20170411_145552.jpg (1MB, 1312x1730px) Image search: [Google]

1MB, 1312x1730px

>>60197191
Kill yourself you perma-BTFO'd tripfaggot

I often imagine that the technical jargon spewing AMD shills are like car salesman trying to convince someone that their product is worth more than it is.

Who knows what motivates these autists, they are like those people that obsess about train models.

>>60197774
>>60197756

>>60197735
Not just on the frontend, for the IPC increase needs a longer pipeline, so the ALUs will be larger, then there's TBR that needs a complete overhaul of the L2 cache and ROP system, there's a good amount of xtor budget going there.
I won't pretend I know how much is going for the new command scheduler, shader primitives or variable precision, but it's probably notable.

>>60197825
>clock increase
Not IPC, I don't know how they're getting IPC up on the ALU side without counting in the the others.

>>60197755
THREAD/

THIS IS FAKE

>>60197755
Looks like a number was shooped out in front of the 5

More :C1 revision benchmarks.

Performance all over the place of course, I wonder what that C3 one is

https://hardwaresfera.com/la-amd-vega-687c1-destroza-la-nvidia-gtx-1080-sombra-la-nvidia-gtx-1080-ti-compubench/

File: screenshot-www.3dmark.com-2017-05-03-16-32-11.png (435KB, 1086x2346px) Image search: [Google]

435KB, 1086x2346px

>>60197954
http://www.3dmark.com/spy/1544741

This is FAKE. Vega is slower than 1080. AMDrone are in full of damage control because Vegay is shit.

http://www.3dmark.com/spy/1544741

all this amd supremacy influx on /g is new meme / some advertising agency moved here or what?

>>60197774

Kinda like all you Novidya kiddies not understanding why their card is shit at direct X 12 and just cry about AMD optimization.

>>60198004
Bruh I ain't even an AMDrone, but if anyone's in damage control, it's you

>>60197975
This is the same leak from 3 months ago.

>>60198095
>make a thread with a fake benchs
>not amdrones

>>60198155
I didn't make the thread, I came for the tism

Whatever Vega is, it should be best paired with a Ryzen, or at least I hope AMD had enough sense to make the CPU take advantage of their driver stack.

That infinity fabric and HBCC might only be on Ryzen as well.

>>60198004
That benchmark was done with the card both underclocked 25% AND the CUs running at half speed per clock. rofl.

You're retarded if you think AMD is going to make such a massive die card and sell it for $600-$700 when it's slower than a 1080. That was a stolen, gimped, engineering sample.

File: HNI_0080.jpg (21KB, 404x402px) Image search: [Google]

21KB, 404x402px

>>60196131
thanks for the kek, anon.

>>60193320
NO, THIS CAN'T BEEEEeee....

>>60198911
The point is the OP pic is fake....

>>60199185
And >>60198004 is equally fake, retard.

If you take a 1080Ti, underclock it 25%, and disable half the cores, and use Fermi drivers, that's not a real result. Fucktard.

what if it was real tho. imagine the nvidiot anal pain

>>60199212

http://www.3dmark.com/spy/1544741

>>60193320
It's fake...

This is the real one:

http://techreport.com/news/31822/rumor-rx-vega-time-spy-benchmark-score-leaked

And yes, it's a massive disappointment

>>60199468
>this is the real one
>posts link to another fake

>>60199468
Yes, yes, Vega is Fiji with a overclock, good show.

>>60199532
>posts link to another fake

http://hothardware.com/news/amd-radeon-rx-vega-3dmark-timespy-benchmark-allegedly-leaks

Also the OP pic is a 1080ti because he can't post the rest of the pic or the bench link...>>60199429

File: 1492981927265.jpg (150KB, 700x1120px) Image search: [Google]

150KB, 700x1120px

>>60199637
>he posted it again!

>>60199731
I don't get it, can you elaborate?

http://www.3dmark.com/spy/1544741

>>60199809
see >>60199212

File: hm.png (37KB, 375x360px) Image search: [Google]

37KB, 375x360px

>>60199809
>...and AGAIN

Really zaps the synapses

>>60199248

We can only hope. I nearly considered a 1070 before I saw Vega was on track to launch this quarter.

Thank you AMD. Give 'em hell.

>>60199731
Oh boy...I'm done with amdfags...

>>60199731
You are so stupid jesus christ. Op will never realese his like because his 3dmark score is fake

>>60199846
So this link (http://www.3dmark.com/spy/1544741) is fake?

>>60199890
It's not Vega, so yes

>>60199890
Holy shit. What don't you understand? How are you this stupid? Yes, it's fake.
Why the fuck are you even posting shit when you have literally 0 understanding of hardware and benchmarks.

File: 1488054924552.jpg (42KB, 499x636px) Image search: [Google]

42KB, 499x636px

>>60199212
>use Fermi drivers
Literally what???????????????????????????????

Also the 1080ti have 11GB of memory, not 8.

>>60199928
>>60199923
Why?

>>60199928
Why it's fake?

>>60199931
>literally what?????????
exactly my point, you fucktard. Can you not read?

>>60199947
>>60199946
>why it's fake
see >>60199212 and stop posting.

File: 1455420638036.png (443KB, 462x469px) Image search: [Google]

443KB, 462x469px

>>60199212
>If you take a 1080Ti, underclock it 25%, and disable half the cores, and use Fermi drivers, that's not a real result. Fucktard

Amdrone

>>60199976
What is '''fermi drivers'''?

>>60200015
Outdated drivers, he's saying that if this is indeed Vega, they're playing it safe by cutting down the total usage and still working on optimizing the drivera

>leaked info about vega is bad
Another Major Disappointment!!!

>leaked info about vega is good
FAKE NEWS!!!

File: 2017-05-03-18-36-00.png (17KB, 789x450px) Image search: [Google]

17KB, 789x450px

>>60200032
R-r-r-r-right...

>>60193320
LINK OR GTFO NIGGER

>>60200059
Believe what you want fanboy

>>60200059
>1200MHz
>CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT = 1
r
e
t
a
r
d

>>60200045
>leaked = shoop

>>60200080
>post pic of the 'fermi drivers'
>belive what you want fanboy

>>60200080
Now, post the pic or the link of the OP leak

File: 1493742333660.png (1MB, 1440x2392px) Image search: [Google]

1MB, 1440x2392px

>>60193320
seems legit

>>60200143
>>60200145
Never had it because I'm not OP, I'm just not being a blatant Nvidia shill like you two faggots

>>60200174
what are fury x benches for Doom at those settings though?