Where are the custom vega models? tf they need so long time for?

where are the custom vega models? tf they need so long time for?

Other urls found in this thread:

anandtech.com/show/11724/samsung-sk-hynix-graphics-memory-prices-increase-over-30-percent
youtube.com/watch?v=KD9CRQp2YpA
twitter.com/SFWRedditImages

they use them for mining at the factory before they ship them to customers

>making custom PCBs for some of the most advanced technology in the world
If anything they're relatively quick at shitting out custom cards.

ur brokem english is cuty :)

AMD is probably staggering the launch.
Maybe they want the software to be ready before AiB cards launch.

are u a grill?

kek

Because this is the most retarded launch ever

Nothing beats nv30.

>Maybe they want the software to be ready before AiB cards launch.
This would be the first time that AMD gave any fucks about having software at launch.

Well, that's nu-AMD.
Just look at EPYC launch - like every vendor, both software and hardware, present.
Blatantly unfinished gaymen drivers is mostly lack of resources.

...

great, what's this shit

Some dude undervolted Vega.
AMD is infamous for overvolting cards stock.

is this a custom partnercard?

nope

Basically my RX Vega 56 matches an OC'd GTX 1080 (there are higher OC's with better scores of course) at only slightly more power usage (typically GTX 1080 OC'd uses over 200 watts).

No, default shitblower.
That's the limit of GCN.

BTW HBM2 OC'd still uses the same amount of power at around 20 watts so it's always worht overclocking the memory. Total for both GPU and memory will be about 250 watts.

Yes partner cards should be even better performance with decent cooling. I put my blower at 2950 max and it can get a bit loud but not leaf blower loud. I put mine to 4900 rpm for shits and giggles and was able to overclock really high but the noise was unbearable. The screengrab is about my sweetspot.

Always takes a bit. How many weeks has it been?

>Yes partner cards should be even better performance with decent cooling.
They will have marginally higher clocks.
The main advantage is noise levels.
Better software will give better performance.

There is a dead lul with the drivers right now. They are obviously working on them but also don't forget the new pro WX 9100 GPU's is coming out October so they will be working on making sure the gimping is balanced so they can keep their SKU's well separated from each other.

Both WX9100 and SSG are early September.
Guess why consumer drivers are pure shitfest right nao.

it's been 14 days user. two weeks. what do you expect them to do in two weeks?

Yep! They are working on those drivers. Also bear in mind the WX 9100 and SSG cards will get the cream of the crop in terms of silicon. Only the best dies will do. Gamers will get the shittiest silicon that does not cut the mustard. Which is why production is limited right now. Also take into account the HBM2 ramp up time for production. It's a real shitshow right now. Plus memory prices on GDDR5x and mining etc mean more folk will flock to Vega as time progresses. It's why I jumped on the £380 reference 56 now.

The reference PCB is top notch. No need to create custom PCBs. All they need to do is slap a cooler on it that can handle the 300W of power it consumes.

Get ready for assrape

anandtech.com/show/11724/samsung-sk-hynix-graphics-memory-prices-increase-over-30-percent

GDDR=/=HBM

do they need to buy reference cards at release? wtf, thats why they are sold out everywhere

I'm pondering getting a Morpheus II heatsink and strapping on some cool and quiet fans (not sure which fans though yet).

Slapping whatever custom cooler on reference PCB since reference PCB is fucking stellar for both 56 and 64.
>Also take into account the HBM2 ramp up time for production.
Samkike and SK Kikenix are already cutting G5 production (guess why). Hell, Samshit is ramping both 4-Hi and 8-Hi production.
And i think crapple will get top bins for Vega10, Pro cards will get decent ones, and gaymurs will get trash (that is still overvolted for no reason, Vega10 can hold DPM6 at 1.1v, 1.2v is totally overkill).
That's G5.
HBM2 is going to get cheaper once Samshit rams us the production.
The problem is packaging.
2.5D is S L O W.

I said GGDR5x. HBM2 might not rise we shall see. If HBM2 works out cheaper more people will buy Vega (once drivers are sorted it will be a no brainer).

I thought about getting a reference Vega 64 when it was for sale for 505€ at mindfactory and putting an alphacool AIO water cooler on it. But then I read several reviews for their CPU coolers and a lot of them said that their new pumps were shit and they would start to make loud grinding noises after two months.

I decided to wait for AIB cards instead.

Good luck. I may still sell my reference on and get a third party if I can get one cheap enough. Otherwise it's the Morpheus II hybrid mod for me.

Asus did just that. It's not good, they simply removed power limit and added 2.5 slot cooler.
I'd like them OC v56 memory to make it v64 and add 2.5 slot cooler, but AMD won't let them.

No, ASUS is what, 12+2 compared to 11+1 for ref?
PCB is also taller.
Definetly a custom one.

>like every vendor, both software and hardware, present.
>Blatantly unfinished gaymen drivers is mostly lack of resources.
I blame Guy Kawasaki for this. His "talks" are very influential, and he basically memed that it as okay to released unfinished buggy products.

Do we know a release date for any of the custom cards?
I have three weeks of vacation coming up and I wanted to buy a Vega and spent it playing vidya. Please tell me we'll get custom cards tomorrow.

They launched both FE and RX Vegas to please shareholders.
That's it.
No.
Heck, only ANUS showed custom designs.

Don't expect third party cards until end of this month at the earliest.

How good is the reference liquid cooler? Is it worth getting for 700€?

Vega 64 is not worth it period right now. If you want to take the risk that drivers will fix it's inherent flaws then go for it. But it is a risk. Right now (especially if you flash Vega 65 with 64 BIOS) is the performance sweet spot. Matching a GTX 1080 with some undervolting and overclocking.

>How good is the reference liquid cooler?
Dunno.
>Is it worth getting for 700€?
Right now?
No, the performance isn't there.

>Vega 56

>How good is the reference liquid cooler?
Tiny radiator. You're better off with the big heatsink, and fans.
>Is it worth getting for 700€?
My opinion, no.

why would a manufacturer spend time ans money when they know miners will gladly eat shit at twice the price for the blocky, simpler stock design

What are the chances that the 3rd party cards will have better stock performance because they'll take advantage of undervolting by default?

Miners won't touch Vega, period.

Slim. They will still rely on AMD supplying them chips that have a large margin of error (not golden chips). More likely just quieter cooling with slight overclocks and let users do the undervolting.

Not yet. Some dude gets 43mh on his so it is improving. If Vega becomes cheaper to buy than other GPU's expect demand to rise.

Oh fuck.
Goddamit ETH going PoS when.

Sorry I mean to add 43Mh at 130 watts (he's keeping how a secret right now though).

I heard they have put off PoS for a year. Fuck miners!

>43Mh at 130 watts (he's keeping how a secret right now though).
Underclock and undervolt the core, OC the memory.
That's for ETH at least.

I think there was more to it than that. Something about preventing it from crashing due to HBM2 being starved or something. One suggestion was he was running another app that keeps HBM2 voltage from dropping. I dunno it's a bit too technical for me.

Wait, this is an undervolted Vega56 roughly matching an OC'd GTX-1080??

Holy shit.

Yes.
That's the limitations of previous GCN iterations.

oh silly user, you didn't know that? watch this, the RX Vega56 is literally faster than the RX Vega64 right now. All the Next Gen Compute Units doing absolutely nothing. AMD fucked up bigtime

youtube.com/watch?v=KD9CRQp2YpA

>overclocking something to hell makes it faster
Whoa, Steve.
Now please rope yourself.

>4 0 0 W A T T S

asus says early september but who knows
probably early october for everyone else

That's with power limit removed.
You can do it to any nVidia card too (but you need to hardmod it).

I think pascal can't do that, we'd see some 500w numbers if they didn't hard lock it

Who cares about the others. Asus make top notch cards with great performance and outstanding looks.

You can, you just need to hardmod it.
And no one but LN2 dudes would ever hardmod a GPU.

suuuure.

>t. ANUS shill

I'd advise AMD do the same so retards stop "muh housefire" spouting, but it would kill undervolting and we can't have that.

I forgot to add in the fan powerdraw so on my undervolt of 230 watts (not sure if that includes 20 watts for HMB2?) it probably is somewhat higher than that. I should plug in a kill-o-watt and test this out really.

It won't kill undervolting, but every xXxtremexXx overclocker will rant for ages about that.

back when asus didnt care

Whether a miracle or disaster, AMD never fail to surprise me.

Weren't partners saying maybe in October?

nobody said anything except asus.
all that clickbait rumor mill worth nothing

They have even more suprises coming soon.

...

ASUS made pretty silly things back in the days.
I still remember 3850X3.

they said it's not widespread available until october. hope they didn't meant vega is available in october when the customs arrive

the 5870x2 was actually working quite well shitting on anything nvidia had by a large margin

last ares that actually was doing something

The last one was what, 7970x2?

>le undervolting gives better performance maymay
>T-Trust us please Vega isn't shit
Just take the L like a big boy

t. Jensen

Upvote!
You sure showed me

t. jacketman

>h-he wears a leather jacket
>let's call him leather jacket man
Never understood how this was an insult. Any AMDumbfucks wanna extrapolate?

t. Huang

t. Chunky grandma

t. Jensen
Shitposting aside, both mommy and jacketman are doing good job handling their respective companies.

True
I agree on that one

Raja is holding her back though

Raja is doing more than fine.
Happy shareholders > happy mandchilder.

He's just doing fine
She could find a much better employee

Better than lead architect of every good (or even decent) Radeon starting with R300?
I doubt you'll find someone more competent to lead not-ATi than him.

I undervolted and overclocked my fury.

I wish I could do the same. My Fury Nitro can barely undervolt, and crashes on a +50 MHz core overclock.

Me too.
It's stable with a -96mV undervolt and 50MHz overclock.

can't overclock or undervolt mine, crashes no matter what

umm no sweetie

most partners strip the boards down to save costs. that is why you only see 1 or 2 brands in xtreme OC videos.

>most partners
MSI is not "most partners".

>he thinks AIB's only start getting their chips when the product launches

I'd like to think you were just pretending to be this stupid, but it's far more likely you really never realized that AIB's get their chips months in advance so they have time to actually design PCB layouts and source parts and supply.

In the future, could you try using your brain for ten seconds?