Is it going to be a meme or not?

Is it going to be a meme or not?

Why are they behaving Apple-tier secretive this time?

Deliberate sandbagging to avoid Nvidia further undermining marketshare by dropping 1080(Ti) prices?
Or Fiji-tier Fury X disappointment?

Other urls found in this thread:

bitsandchips.it/english/52-english-news/8049-almost-80-of-amd-driver-team-is-working-on-vega-drivers
twitter.com/AnonBabble

Sandbagging, they've done the same with Zen and it sure worked out.

Ask father time as I'm sure he has the answer

i fucking hope it's worth the wait. i've been waiting for so long. i have to plan carefully a trip the the US to get this shit among other shit, but there's no launch date

i've been waiting for the laptop ux410uq since october and there's no launch date on sight.

living here is suffering. pic related. it's cheaper to travel and buy shit than to pay corrupt assholes

They are polishing drivers. Just Wait™.

bitsandchips.it/english/52-english-news/8049-almost-80-of-amd-driver-team-is-working-on-vega-drivers

The launch day drivers should be near max performance.

Given how 80% of the driver team is working on Vega launch drivers and the fact they're using HBM2 I think we're in for two GPUs that will either be competitive with the 1080/Ti pricing model or they'll outright beat it.

$400/579 with the AIB over clocked models hitting $599

Oh, so they are actually trying their hardest not to make another Fury launch. My apologies to king poojeet.

600 dollar GPU trading blows with the newest titanXP.

they were quiet about ryzen too, yet it's great

It's going to be like Zeppelin/Ryzen:

AMD is trying to build a single solution for both consumer and enterprise sectors, so it won't be the absolute faster gayming card out there, but they'll be able to snag a lot of enterprise market share with fat margins by being able to pull things off that their competitors can't.

Vega is throwing in a bunch of logic to use HBM as L3 as well as some virtualization stuff that won't impact games in the least, but all of a sudden they can start selling $600 or whatever GPUs for a few grand a pop to companies.

Vega is already confirmed to have xGMI link (100 GB/s each, supposedly 4 total) that will allow heterogeneous MCMs/APUs to be made with Zeppelin dies, which is something that neither Intel nor Nvidia can convincingly pull off.

FWIW, I think that Vega gaming will be surprisingly good (just not astonishing), I just don't think that gamers are the real focus at all anymore.

>Vega is already confirmed to have xGMI link (100 GB/s each, supposedly 4 total) that will allow heterogeneous MCMs/APUs to be made with Zeppelin dies, which is something that neither Intel nor Nvidia can convincingly pull off.
>HSA stopped being a meme
What a weird time we live in.

>xGMI link

sorry, GMI links (= inter-die/cross-interposer)
xGMI = cross-PCB @ ~20GB/s for 16 lanes.

yeah.

I'm honestly also suspecting that Zeppelin and Vega will be able to talk to each other through a PCIe slot/connector in native Infinity/Data Fabric protocol, cutting out the PCIe controller and IOMMU latency as well as allowing core-to-core communication not limited to standard PCIe messaging.

It wouldn't be of value to most consumers unless Vega is full virtualized in a way we haven't seen GPUs yet, but it could allow interesting things in enterprise and HPC.

Oh, thank you AMD for making tech interesting again.

I don't think there is a "real focus" at this point

AMD is spread extremely think compared to its predecessors and they simply can't afford to give up on the margins of the business market and the mindshare of the consumer market, which generates sales in the former

chasing halo products is just worth it for them either, which means we'll have decent GPUs with Vega, but no victory over the Titan line, and thus sadly more Sup Forums shitposting

Wait™

>more Sup Forums shitposting
If only mods were alive.

Ad hom. You got fuckin rekt, deal with it.

You have no idea how much cloud/web hosting sales are ready to jump ship to Naples after years of getting fucked by the Xeon cock.

Although it's obviously critical to RTG and important to the AMD as a whole, every other segment they're involved in will seem small in comparison.

Mid-low end quiet low-power cards when?

What?
Navi.

2018 if you're lucky, but it's more likely that Polaris hangs around in rebrand form longer than is tasteful.

2018 for the Vega 11 rebrand.
AMD doesn't bother making GP107 style new releases for low end cards, since they're already selling toaster tier graphics in APU form.

>what?

Holy shit fucking kys newfag

Fuck.
GP107 is not exactly what I had in mind, 120-140W is fine too.
Allegedly nVidia does not want to play nice with Ryzen so I can't decide whether to sidegrade my 960 now or Just Wait™ some more. Performance increase of 470 does not seem worth it while price is retarded, but 570 would probably be a bit too hot/loud for me.

>What?
I want these cretins out of my Sup Forums!

570 is the same gpu.. Just stock oc
Undervolt or underclock it and it'll run like a 470 but more efficiently