Where were you when VEGA won?

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

Other urls found in this thread:

hardwaresfera.com/la-amd-vega-687c1-destroza-la-nvidia-gtx-1080-sombra-la-nvidia-gtx-1080-ti-compubench/
3dmark.com/spy/1544741
techreport.com/news/31822/rumor-rx-vega-time-spy-benchmark-score-leaked
hothardware.com/news/amd-radeon-rx-vega-3dmark-timespy-benchmark-allegedly-leaks
3dmark.com/spy/1544741)
twitter.com/SFWRedditImages

Not this shit AGAIN

Fake news

Alternative facts

>3ip's
>4posts
:thinking:

It's fake like all the other "leaks"

FAKE!!!
Nvidia won't let this happen

VEGA YASS!
JEWVIDIA BTFO!
Based red team.

AMD shills so desperate they literally start making up benchmarks

post link or fake news

NO NO NONONONO FAKE DELETETHIS

$500.

Nail. Coffin. Volta. Kek.

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.

Oy-vey, delet this goy.

That silence must feel like a chokehold to Nvidia

Nvidia has been spooked for a long time.

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

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

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.

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.

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

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

>no link

FUCK OFF AMDRONE

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

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)

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

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

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?

>it's the board pierro

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

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

>breathing joke
Whatever the fuck that means lol.

The answer lies in the mirror.

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.

TDP AMD 300
TPD Nvidia 250

That's over 20% , not 9%

no link

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.

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.

People don't appreciate AMD doing anything.

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

>Ignoring idiots
And only 5% of Sup Forums left.

Sup Forums 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.

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

YAAAAAAAAAAS AMD SLAAAAAAY

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

...

>4096 fucking ALU
You fucking bet.

and here the real results

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.

...

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.

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

THREAD/

THIS IS FAKE

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

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

3dmark.com/spy/1544741

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

3dmark.com/spy/1544741

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

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

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

This is the same leak from 3 months ago.

>make a thread with a fake benchs
>not amdrones

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.

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.

thanks for the kek, user.

NO, THIS CAN'T BEEEEeee....

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

And 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

3dmark.com/spy/1544741

It's fake...

This is the real one:

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

And yes, it's a massive disappointment

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

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

>posts link to another fake

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...

>he posted it again!

I don't get it, can you elaborate?

3dmark.com/spy/1544741

see

>...and AGAIN

Really zaps the synapses

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.

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

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

So this link (3dmark.com/spy/1544741) is fake?

It's not Vega, so yes

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.

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

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

Why?

Why it's fake?

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

>why it's fake
see and stop posting.

>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

What is '''fermi drivers'''?

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!!!

R-r-r-r-right...

LINK OR GTFO NIGGER

Believe what you want fanboy

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