AMD Launching Ryzen On 28th February During GDC 2017 - Vega GPUs

>>58456550

Cash ready awaiting benchmarks.

File: sample_0d0207407344d721c99d63a4abd2b64d.jpg (610KB, 850x1201px) Image search: [Google]

610KB, 850x1201px

>>58456550
THANK YOU BASED LISA
THANK YOU BASED KELLER
THANK YOU BASED RAJA

>>58456567
>>58456720
Are you ready for Another Massive Disappointment? :^)

File: certified shit wrecker.jpg (13KB, 381x286px) Image search: [Google]

13KB, 381x286px

>>58456807
What-what-what?

>>58456550
go fuck yourself, amd

i have been waiting for too long, and now you say two months more? gtfo outta here

>>58456720
Only idiot weaboos would fanboy this hard for such a shitty brand

>>58456550
oh hey
that's on my birthday

My body is ready
2017 will be Intel's last year of relevance.

>AMD stock is down from shortsellers cashing in
Pussies.
Should I buy now or will it sink below 8 before it skyrockets?

>>58457024
You should have bought it when it was at 2.

>>58457042
Don't remind me. I had $2000 I was ready to invest when it was at 1.3 and then life happened.

Meh, my i5 6600k arrives today, with an RX480 of course. Either way good luck with Ryzen.

>>58456838
based shitwrescker

>>58456550
>wccftech

>wont be able to do pcie GPU memethrough due to lack of integrated gpu on 8/16 cpu

meh, and single core performance doesn't look good either, i guess i will go with 7700k

>>58457499
>single core performance doesn't look good either
Says who? Everything official we have to go on shows IPC somewhere around Broadwell/Skylake/Kaby Lake level.

>tfw delayed until April 1st

>>58456550
>Polaris was aggressively priced against its Pascal based competitors.

the 480 was more expensive than the 1060 at release here in germany. that information is more a let off for me

>>58457554
Pajeet fucker lied about the price too, the 8GB 480 still hasn't hit $200 dollars.

>>58457579
4gb was $200. 8gb was like $240 or something.

File: drake_nvidia.webm (3MB, 1280x720px) Image search: [Google]

3MB, 1280x720px

>>58456550

>>58457579
everything I see you post is stupid

turn off tripcode cunt

>>58456550
I'm 90% likely to get a Summit Ridge build in March, but honestly I'm more interested in Naples.

Will there be 2S platforms for 16c/32t MCMs, or will 1S 32c/64t be the only option?
Zen seems to get decent clocks at lower Hz, but I'd rather stick closer to 3 GHz for a 32c total build, and I don't want to get capped by a single socket power draw limit.

File: snapshot_19.37_[2016.12.23_03.36.00].png (2MB, 1920x1080px) Image search: [Google]

2MB, 1920x1080px

>>58458074
remember that broscience bullshit few weeks ago?

>>58457125
It was never at 1.3. Don't lie

>>58458197
Please tell me about it :)

>>58457499
I already have my R9 380X for that.
>3.5W idle

Also, AM4 is ready for raven ridge APU's

>>58458176
32 core naples will likely cost 7-10,000$
Intel has no answer for it, both in memory bandwidth and just plain poreformance.

as for the rante, I would assume amd has at least a 16 core and 24 core opteron in the works, and potentially skews from 8 cores with quad channel (2 4 core dies) through to the 32 core monster, likely in 2 core increments, but possibly in 1 core depending how mcming the chips works.

basicly, through to the 24 core, amd is likely going to undercut intel, possibly only through 16 cores as more cores then that would either make it hex or quad channel for memory.

>>58458523
Say I had the budget for a current intel 10core, whats the likelyhood of amd allowing 2 ocable 8core chips on a single mobo?

Always regretted not going dual socket in the 920 days and while its probably stupid I just wanna do it with current tech.

>>58458523
>>58458604
I'm not sure if I can see AMD doing a 2S*8c platform despite the higher power/clocking ceiling.

Zeppelin can probably get much, much higher interconnect speeds on a shared MCM, and that trumps core clock speeds in a lot of tasks, since you can start striping/interleaving physical address space over all the memory controllers on an MCM instead of just the 2 on each die.

>Vega in May
>Probably won't be able to build a Ryzen + Vega build until August
Nah fuck it, going with jewtel and nvidia

>>58458693
Let me guess, your 2 year old system is just not providing that 2 minimum fps bump you need?

>>58458693
seriously. I'm an AMD fanboy, but May? that's ridiculous. it's a year after 1070/1080, and won't even perform as well as a 1080

my friend is literally waiting for Vega before pulling the trigger on a 1070, but with this news, he'll probably just go Nvidia

AMD needs to fix their release schedule, because they're losing to Nvidia and Intel in every category

>>58456550
>May 2017
>May two thousand seventeen

Woah Vega will be a ferice competitor to the GTX 2050 baby Volta

>>58458856
My build is from [spoiler]2007[/spoiler], the only thing I've upgraded is the ram and I put in a ssd 2 years ago. I don't play video games anymore really but I like to think that I would if I built a new one. Too busy/lazy/tired being a working goy

It's great to browse the web with though.

File: keep-calm-and-buy-the-dip-7.png (32KB, 600x700px) Image search: [Google]

32KB, 600x700px

>>58457024

>>58458905
AMD has been slow as shit lately, but it's not very realistic to expect that Vega 10 would be beaten by 1080s.

Vega 10 will have 60% more bandwidth and 40% more fp32 FLOPS, so the only way it will even be close is if the drivers are completely broken or yields are so bad a third of the chip has to be turned off.

I still think that GP102 will be better, but not even strong overclocks of GP104 should beat it.

>>58458678
So ill more than likely be looking at hoping opterons can OC and buying a higher corr chip? I game some but mostly im slowly getting into video editing so the idea I had was rendering out on 2nd cpu while still having my computer be useable.

Right now my projects arent huge but with family/friends knowing in doing it I have had hints at a lot of holiday/group trip footage heading my way.

>>58456807
>>58456879
>>58456904
>>58457499
Oy vey. 2 shekels for this lot. Good work chaver!

>>58458384
Okay I exaggerated. Was a little below 2 though. We're talking like 3 years ago or so.

>>58459046
maybe, but I can't see it beating a 1080Ti, and we all know once the Ti comes out, the 1080s will drop $150ish and be around $450. who's gonna spend $650 on a Vega when you can get an Nvidia one for a similar price? it's like Fury X again: if Fury X competed with 1080, it'd win, but since it had to compete with 980Ti, it lost

>>58459046
Dude fp32 flops =/= gaming performance. It's good for parallel computational tasks, namely shit like machine learning.
I don't know how well vega's gonna play games (probably still pretty good) but it sure as shit is gonna grab the attention of enterprise markets.

>>58459083
>So ill more than likely be looking at hoping opterons can OC and buying a higher corr chip?

Who knows?
Intel still sells 8c (1.6/1.7 - 3.2/3.6 GHz), 6c (3.4/3.7 GHz), and even 4c (2.6/3.2 - 3.5/3.7 GHz) dual-socket Broadwell models, so maybe AMD will throw a few similar SKUs out and see what sticks.

>>58458523
Isn't naples 8 channel memory?

>>58459003
Head over to pcbg, post your build and maybe we can find some cheap upgrades without a full new system

>>58459218
Deep Learning is increasingly using fp16 (2x rate on Vega and GP100), and it's more scientific simulation stuff that uses fp64.

fp32 is in kind of a weird space in the middle where some financial models etc. rely on it, but fp16/fp64 algorithms seems to be bigger growth areas.

>>58458905
Doesn't matter to me at this point if Vega is late compared to Pascal, I refuse to pay the GSync tax so until nVidia supports FreeSync I won't be buying their cards.

>>58459275
yeah, just got verified in slide leaks this week.
obviously only true for the 32c/64c 4-chips MCMs though.

>>58459246
With alibaba going for ARM -- and possibly giving other major players funny ideas -- and if AMD delivers on Naples Intel will have a bad bad 2017...

>>58459383
Even if everybody sticks with x86, Intel's enterprise Xeon margins are going to absolutely collapse if Naples is even remotely as good as it looks so far.

>>58457499
Enjoy your nogainz

>>58458678
>>58458604
Amd will make a dual or quad socket chipset for workstation/server use, but they may also come along with one for non server cpus or even ask allow non server cpus to work with that chipset.

there is quite a lot of use for a dual cpu even in a non professional setting, such as streaming, recording video, or just the overhead of doing something taxing. but this would likely be a mid cpu cycle thing, not an out the gate.

>>58459083
do not get more then one cpu for video editing, its one of the fun areas where the programs refuse to see more then one cpu.

>>58459315
depending on the chips changes how many channels there are,
1=dual
2=quad
3 (if possible)=hex
4=octa

it has to so with how each cpu die handles memory.

>>58457024
https://www.youtube.com/watch?v=tK_C8NAchko

>>58459650
Problem with that is sockets, I doubt they're going to have 2/3 de parts if the boards are all octo channel memory, which is why I only expected quad channel memory so 2/3/4 die opterons could be used for a top to bottom range, I mean intel's only getting into hex channel with their new lga3000 which is going to sit above the also new lga2066. It's like amd have hit a niche that intel didn't feel at threat with this 8 channel memory thing.

Another anon did mention the 16/32 opterons will be on a bga platform competing with intel xeonDs, wasn't the old 16core leak a low clocked part as well? would seem to tie in with that rumour.

Not falling for the usual retarded hype.

Reviews needed

>>58458523
>Intel has no answer for it, both in memory bandwidth and just plain poreformance.
http://ark.intel.com/products/93790/Intel-Xeon-Processor-E7-8890-v4-60M-Cache-2_20-GHz

File: ofnote.png (459KB, 756x620px) Image search: [Google]

459KB, 756x620px

>>58456550
funny post i saw in there.

So I just heard that Microsoft wants to reveal the Xbox Scorpio before E3, so before June. So maybe AMD and Microsoft are going to Launch their products together. AMD is already providing the hardware for a 4k Xbox.... I'd imagine Xbox will have some form of both Ryzen and Vega since AMD has been taking their sweet ass time.

thoughts on my asspull speculation?

>>58459930
only 24 cores
only quad channel

As the other anon stated, intel has no answer for Naples, at least not yet. They'll be releasing a 32 core Skylake Xeon sometime this year to compete with AMD's new Opterons. Though it will probably be clocked even lower.

>>58459986
as far as i know, the ps4 pro is using AMD hardware as well, maybe Microshill sucked AMD off for better hardware, we'll see though.

>>58460030
microsoft and sony use booth amd software for their consoles

>>58459986
AMD did have a little Xbox Scropio sign at their CES booth, and it did have the Ryzen logo on it. AMD wouldn't comment about it, but its pretty obvious why it was there.
The new Xbone will have newer hardware in it. It'll be Zen based.

>>58460047
I'm pretty sure that's what i just said.

>>58458856
a lot of us are 6+ year old systems and were waiting on intel to drip 6 cores down to not retard price points, if not 8 cores, but intel jewed us hard.

my sticking point right now is just the cpu, so amd putting it out within 6 weeks, I can easily get it and if vega sucks, I can stick it out for another year or so on my gpu.

>>58458905
why would they bother making the hardware if no one buys it? this is the point most people can't answer me in a good way on. the market has never, every time they were better price and performance and again when it was just better performance rewarded amd. Why go out of your way with catering to consumers? put some shit out at a sweet spot price/performance and call it a fucking day, the high end can be a byproduct of pro parts, and that's it.

I can't understand why people think amd should bust their ass to make a good part just so nvidia lowers their price for you to buy nvidia, everyone apparently wanted a monopoly and then bitch when they get it.

>>58459279
there are no upgrades that are at best side grades. assuming his pc is from that era, he would best be served with getting intel's 60$ 2 core 4 thread cpus.

https://pcpartpicker.com/list/kkRhYr

the case is a bit cringe, but get rid of the led fan in the front with something not glowing, and it would be great airflow for the cheapest price possible.

>>58459930
24 cores opposed to amd 32
quad channel memory apposed to amd's octa channel

currently intel has nothing to go against this. and when intel puts out their 32 core cpu with hex amd still has the bandwidth advantage, and no multi socket tax.

>>58456550
Seriously this thing won't be in stores until Christmas 2017 at this point.

>>58460047
>>58460076
Nevermind, AMD software?

>>58460076
you said ps4 pro and scorpio. but the xbox one and standard ps4 use them too

>>58460087
i'm retarded, i mean hardware

>>58460103
oh yes, i didn't know the normal xbone used amd but i was aware that the normal PS4 used amd as well.

I'm honestly interested in the specs of the 4k xbone.

>>58460122
That's what i read it as anyway.

>>58459986
>>58460030
>>58460048
PS4 pro stuck with gimpy Jaguar CPU cores and only really added 2x fp16 throughput to a doubled GPU.

Scorpio is anyone's guess, but it would stand to reason that it could have launched substantially earlier if it didn't include Zen and/or Vega tech.

Also, Rynen launches at the end of Feb, Vega will be in May. Microsoft can do all the backflips they want at E3, but Vega will have already (but very recently) launched.

>>58459046
amd is consistently beaten out by nvidia on cards with almost 40% less flops, and when they finally get their shit together on drivers, nvidia has cards with equal flops doing 10-20% better.

Its likely vega will beat out a 1080, but nvidia has the wiggle room to shipt prices down drastically, making a titan xp 500-600$ is not unrealistic but amd would have to force them to do it, but a 1080ti would likely satiate people enough they don't have to touch the titans price at all.

>>58459136
no, we all assume that is what will happen, realistically, if nvidia is not apposed, they will release a 1080ti for the 700-900$ price range and leave the 1080 alone. Hell, you want me to be smart about it

the 1070 will be apposed by a smaller vega, the 1080 will be opposed by a bigger vega and potentially a 1080ti will match a big vega, its all on amd to force those prices down, but if I was nvidia, unless vega was more than 10% better at a lower price point than any of my cards, i would not touch prices at all. because why the fuck would i? No one buys amd.

>>58457579
they raised the msrp and made it 200 for the 4gb, you fucking retard
>2 shekels have been deposited into your account, powered by novideo gimpworks

File: 1481484192769.jpg (36KB, 604x604px) Image search: [Google]

36KB, 604x604px

>>58456807

>>58458958
this is really assuming nvidia has anywhere to go architecturally, anyone know of any gpu techniques that came out that would increase performance on lesser hardware, this isn't an area I know much about?

its kind of assumed amd will mcm for navi, and nvidia would love to do that for smaller dies too, this is likely the next area that both try to hit.

if amd goes this way and nvidia doesn't, amd is going to win that generation for performance, 2 polaris dies without the crossfire tax would likely cost less than 400$ and would beat out a 1080
3 dies would beat out a titan for 560, and 4 would beat out nvidia's best for under 720$. This is assuming scaling is not an issue, and that there is no driver support needed, the mcm and talking is firmware and nothing can fuck it up through bad code.

But again, no one buys amd, so why would they cater to a high end market? going the mcm route, this allows them to make a great sweet spot gpu, and then scale it outside of the lithography, wasting no money or resources on niche products, it would allow them, once the gpu is done, to have an entire lineup at the same time, hell, you may see quicker gpu iteration, where there would be a 1-4 die range for polaris, then with vega another entire 1-4 die lineup ready to got.

This could potentially kick nvidia straight in the teeth if they are not trying to do this already. but again, no one buys amd for gpus.
cpus, even as shit as they are, people still get them, but give them an amd gpu that is half the price of nvidia and twice the performance, they still fucking buy nvidia.

>>58460220
Nvidia has more recently beaten out AMD cards by choking the geometry and ROP pipelines, but Vega supposedly has almost a 3x clock-for-clock boost in geometry throughput over GCN.

Nvidia literally cut ALU FLOPS from Kepler to Maxwell for comparable models but still did better due to the selective bottlenecking in addition to legitimate advances like their tiling rasterizer.

>>58458905
>it's a year after 1070/1080, and won't even perform as well as a 1080

That doom benchmark puts it between the 1080 and the Titan.

It's why Nvidia is delaying the 1080 Ti (a cut down titan) until Vega is out, to fuck them up the same way they fucked up the Fury X launch with the 980Ti.

>>58460346
>its kind of assumed amd will mcm for navi,

I had kind of thought so as well, but AMD quit tagging "scalability" as a feature on their marketing roadmaps a while ago, so who knows?

I just know that Navi is supposedly a 2019 product now though.

>>58459723
no, that would be retarded to not have 2 die parts at least, 3 is just an assumption they can.

imagine this

"We are amd, and we want back into the server market, here is out only cpu, 32 cores 8 channel memory"

They would be catering to the smallest segment of the market, the people who need 32/64 threads and need high bandwidth memory.

intel has 4 cores at the low end up to 24 cores at the high, amd will hit them along the line as not everyone needs more than 4 channels of memory, or their needs don't have more than 8 cores.

they would not give up a majority of the market just to have one sku above intel, unless this sku is under 2000$ so it fits everyone's need and budget, but that would be retarded to price it there if its better then everything intel has.

>>58460030
Microsoft unlike sony has the money for some major sucking (sonys game division is the only one making money). With Phil Spencer at the helm, they might be doing something bigger than expected since they have been pulling some moves to Merge PC and Xbox. especially since they have been expanding their Play Anywhere program and apparently a Game Specific mode in Windows 10.

>>58460378
AMD fucked itself with Fiji, and the 980 Ti at $650 was a preemptive overreaction by Nvidia, who apparently thought it was going to kick their ass.

Vega seems too similar for my tastes, a big chip with a lot of priority given to HPC customers at the possible expense of the gaming market.

File: 68371667.jpg (92KB, 400x358px) Image search: [Google]

92KB, 400x358px

>28th february
>may

Nigga I want to build me a new PC right the fuck now. My i4 4690k is getting old. Fuck waiting another month, I'm going Kaby Lake i7

>>58460566
Where did you find an i4?

>>58460378

It beats it in Doom which uses Vulkan which is an API that already favor d AMD. Kinda how like the fury x beat the 980ti in it, even though the 980ti beat it in 98% of the titles. Pretty sure Doom and Ashes of Singularity are the only 2 games I can think of where the fury x beat the 980ti. Chances are Vega will still be worse than the 1080 and definetly 1080ti, but at a more attractive price point. Forcing nvidia to drop prices.

>>58460590
Shenzheng Chyna probably

My first time I came to /g/ I learned that Phenom II x3's could unlock to x4's with a BIOS setting. Then I built my first custom PC with a Phenom II and a Radeon 4850.

Since then it's as if picking any other processor besides current unlocked i5 has been useless.

>>58459409

AMD has practically ZERO market share in the server space. They have nothing to lose and everything to gain. I'm 100% convinced that Zen and Vega were designed first and foremost as server/work station focused parts.
We have info that 8/16 Zen is just over 95W. Intel's 8c is 140w. Are you fucking kidding me? They have an efficient arch with comparable IPC and will most likely a better price/performance/watt. AMD can start and win a price war in the server space. They have nothing to lose and everything to gain. Whether Zen is a PC hit or not won't matter. If AMD even got 10% server market share, that's YUGE money.

I got some stock now but I wish I had more money to buy even more.

>>58460963
Part of the TDP delta is the fact that Broadwell-E(P) has 4 DDR4 controllers and the cores have much higher AVX throughput, but I agree with most of what you're saying.

>>58461008
Its not just their higher throughput potential but the circuitry itself.
If you have a high resolution thermal camera and look at a board from the back side you can see through it, and see the specific structures on die under load. The larger datapaths in the FPU are noticeable. The moment they switch on power consumption increases, even before they start processing instructions. That means that there is static cost in power to have them on. They're normally gated off, and the second they become active power draw per core increases significantly. They large structures, they account for a lot of transistors, and they're a lot of thermal budget.

>>58457262
>buying the literally cucked 290x 4 years after release
LOL

>buying a i5 after 2015
KEKED

>>58461572
good goy

File: 1481703908447.png (203KB, 500x420px) Image search: [Google]

203KB, 500x420px

>>58456807
Make AMD Great Again

>>58461605
amd are the opposite of jews you fucking faggot

>>58461623
>being so retarded as to misinterpret the Intel/nVidia shill

>>58461637
good goy is a insult you douche

>reading comprehension
learn it

>>58461664
Yes, I know, I was insulting the shill.
Do you have a bad case of autism?

>>58461675
Im the amd fag with a i5 4690 and a 390x

weird mix but runs pretty good

>>58461697
Well yeah, it should.
It's not a weird mix at all.
Both of those components are great value with excellent performance.

>>58456550
Too late to care. Fucking things should have been out months ago...

The moment of relevance has passed.

>>58461623
Okay, so they are Nazis then.

>>58461718
cheers anon

only game that kills it is Rainbow Six Siege but every patch since march has made the game run worse
and worse
and worse

now i barely get 60fps outside
>>58461733
SO FANNY

>>58461738
>Rainbow Six Siege
Yeah, that's not surprising. That game is horrifically optimized.
Should still be outperforming a GTX 980 though.

>>58461767
nah latest maps and patches killed performance even on the older maps

easily the worst running game ive played

>>58456550
>Vega in May
What the actual fuck?

>>58461794
Too late or early?

>>58461794
probably yield problems

fine by me anyway im broke trying to pay off debts and save until then anyway.

Vega seems to be just about right but Zen is overkill for basic gaming

>>58456550
Any prices yet?

>>58461824
i think he means too late

Guessing they are having problems with getting enough made to sell or binning issues idk.

Launching it an entire year after the rx480 is fucking dumb but if its a good card like the 290 was back in the day its worth the weight

>>58461826
More likely drivers.
Granted though a 500mm2~ die is fucking huge for 14nm.

>>58461926
does that mean shitter yields even with FinFet?

havent really been keeping up

>>58460590
i4 is a shitty engine used exclusively by shitboxes. Go with godtier i6 instead.

>>58459120
>>>
> Anonymous 01/12/17(Thu)17:48:45 No.58458384▶>>584591
i told my dad to buy when it was at 1.80 last year... and he didnt

>>58461953
The 14nm LPP process is pretty mature, yields of the Polaris designs are pretty strong, the Zen based Zeppelin dies should be in high supply as well. Though those are in the range 200m~ or smaller, Vega is considerably larger.
Without knowing the exact specifics of what a test die of a given size is yielding theres no way to estimate what a larger design would yield.

Though even the 600mm2 28nm HPP GPUs were pushing it.

>>58462018
so wait Vega is coming after Ryzen?

Fuck sake amd

I have been out of the loop.
Whats the specs on ryzen?

>>58462245
google

basically 4 core i5 level from 2015 enty and can compete with the 6 and 8 core i7's

>>58460220
I think nvidia would rather go for more aggressive prices on the Ti's than touch the Titans pricing scheme.

Titan is the rich man's card, I doubt they would readjust their strategies after they managed to brainwash peole into desiring that SKU line. IF Vega were to btfo Pascal current gen, I think it's a lot more likely for nvidia to launch the Ti's at the exact same price point they asked for the 1080, or even lower before considering to do anything with the Titans.

>>58463098
TXP is actually the poor man's Quadro P6000, currently selling for ~$5400 with all CUDA cores enabled.

>>58463188
Ok, rich gaymen.
You know what I meant.

And the p6000 is the idiot's card, since the new firepros are performing within 10-20% shy of them for a quarter of their price.

>>58463234
>thinking price matters for people who would actually consider a quadro

>>58463259
thinking that only rich fundie hipsters who are very talented photographers or graphic designers according to their socialite aunties buy workstation level graphics cards and not companies that buy them by the 100's or even 1000's...

File: compute-2.png (12KB, 626x367px) Image search: [Google]

12KB, 626x367px

>>58463259


http://hothardware.com/reviews/nvidia-quadro-p6000-and-p5000-workstation-gpu-reviews?page=2

800 dollars vs 5.900 dollars.

this price difference would make eve high frequency traders think twice about going for the absolute best.

File: 1476421564286.jpg (926KB, 3000x1687px) Image search: [Google]

926KB, 3000x1687px

>>58460389
now they didn't, so as nvidia
scalability is still there for volta/navi and both coming in 2018/2019

I have no idea what kind of ground work they have to do for it to work, 60% of GPU is driver support
how will it work no idea, certainly it won't be like cf/sli interface

the point is, volta not coming anytime soon
they wouldn't delay 1080ti if they had new gen it would be pointless and harmful for sales

>>58460963
they have 90% in thin clients
kind of server sphere

>>58464314
oh, and btw vega is pretty much on point with this roadmap, anyone who expected it earlier is stupid

>>58463259
It does, since they do workloads you can scale out.
And if it's a single workstation, you have $n to spend on a computer. Do you want the very best GPU, and an i5 or an i7 and the firepro?

>>58456550
>the price compared to the FX-series is 40% more
lmao AMD lost, they cost far more and previous gen, delivers less performance than current intel

>>58464360
https://www.youtube.com/watch?v=2fM9OfB4Jvk&t=713s

AMD IS FINISHED AND BANKRUPT!

>>58457375
T*P K*K

>>58464360
40% more over the FX series isn't an issue when it performs as well as haswell or higher.

My only concern is if they can reach high enough clockspeeds, 4.5 ghz on an overclock is needed to be competitive

>>58464479
It's an issue for AMD, a selling issue. When the major point of their team was to have a very low price and deliver a good punch to Intel. If the prices are unfair as they were with the RX cards, then you will see a loss in customers and etc. But drones like you don't know of it.

File: seen1.jpg (65KB, 500x519px) Image search: [Google]

65KB, 500x519px

Who Kaby here

File: Absolutamente Asqueroso.png (178KB, 288x415px) Image search: [Google]

178KB, 288x415px

>>58464521
you mean Skylake OC edition

>>58464497
>When the major point of their team was to have a very low price
It wasn't

>>58464562
>it wasn't
lmao

>>58464566
The major point was to have a competitive cpu again, and they did that, as long as they price it simmilarly or slightly below Intel's offerings, they will gain marketshare.

40% over the FX series isn't an issue when you see what prices Intel charges, but I wouldn't expect a dummy like you who cannot into basic calculus to understand that. Multiplying something by 1,4 is apparently beyond your abilities.

>>58464578
There is no competition when you're delivering only half of what intel has to offer and with 100% guarantee that it's future proof. Enjoy another disappointment amddrone, an EXPENSIVE disappointment

>>58464597
>There is no competition when you're delivering only half of what intel has to offer

Opinion discarded, you're not interested in a serious discussion

>>58464597
>100% guarantee that it's future proof.
>change sockets every 2 year
Good goyim

How the fuck did AMD managed to take one whole year to release Vega?

>>58464597
>0.1 rupee have been added to your Intel™ Extreme™ Pro™ account, Mr. Rasheesh

>>58464693
Who cares, nvidia released and thats what matters.

>>58464693
That's a pretty small turn around for going from Polaris to Vega.
Nvidia only has Pascal refresh this year.

>>58464597
Gee I wonder who could be behind this post

>>58463098
You would have to force nvidia to play their hand on a titan price drop, as in either touch it, or surpass it.

>>58464354
its at the december area of 2016 and february at worst on that map, maybe if the square ended where the g in vega is you may have a point.

>>58464578
someone did the math and it seems like instead of 40% it's closer to 55%, can't remember where or who but grain of salt.

>>58464903
yeah good boy suck that green cock

>>58456550
Way too late, AMD is fucked. They can't even make shit that competes with Intel & Nvidia's last generation.

>>58457499
Just add a $20 passively cooled PCI-e GPU to your system and use that as the host's GPU.
Alternatively, Raven Ridge, their Zen based APUs will be out this year too.

>meh, and single core performance doesn't look good either
It looks to be on average within a few percent of Skylake/ Kaby Lake.

>i guess i will go with 7700k
So, you'll go with almost half the CPU performance?

>>58465130
AMD hasn't released prices yet, so I wouldn't take anything like that seriously

I just bought a W541 laptop with the Haswell I7 (4810) for cheap. I see no reason to buy a memelake or baby lake part. I might build a couple more Sandy Bridge systems for gaming in the house. I'll end up with SB/IB/Has/DC depending on what's for sale with the K series. The i3 parts are the only thing new that might be worth a looky loo.

>>58465210
>t. pajeet

>>58465454
amd is the pajeet company
nice try rajeesh

>>58457499
>buying the refresh of a refresh of a refresh which runs 10% faster than the original
>but priced like it's future tech

>>58460566
And what hardware intense game are you going to play on that new PC? Oh yeah, that's right. There aren't any.

>upgrading from an i7 4690K

Okay, you're just an Intel marketer.

>>58465400
> for cheap
How much is cheap? I see them for ~$600 on Ebay.

>>58461784
Have you played arma 3 yet?

>>58465457
>t. envious ranjid

>>58460030
The PS4 pro still runs on a gimped bottlenecking laptop CPU.

>>58456550
>Vega GPUs To Launch in May
DAmn thats far

>>58467574
It's also a lie. Nobody knows the release date yet, WCCFtech is just pulling shit out of their ass so they don't have to put it in the loo

>>58467752
Yeah but there havent been any concrete news about the GPUs

Its most likely accurate

>>58467783
>Its most likely accurate
it can be anywhere from end of march to mid july
depending how many chips enterprise will buy
all production will go to them first

once raven ridge lands I might actually get a fucking laptop.

>strong iGPU
>strong CPU cores
>low tdp, long battery life
>smaller laptops with simpler board designs
>shared HBM memory (fingers crossed)

even if it doesn't rival the high end Intel stuff we might actually have some viable everyday laptops that can run video games and aren't total shit bricks

>>58467177
I bought mine for $500 burger bucks with ultra dock and 170w power adapter. Windows 7 was still installed which was perfect as the key was still valid for Win7 making it legit and not wasted on a shitty Windows 10 install.

>>58467177
>>58468217
Oh and of course it's still under warranty for two years almost.

>>58460566
How much you selling your 4690k for? I'd rather buy one of those tbqh.

>>58456550
Oh great Feb. Just somehow got more depressing.

>>58467899
This. Might sell my chinkpad for a Zen laptop if Raven Ridge is good.

>>58464597
Not him, but the only reason I ever bought AMD was because it was cheap

If they're both going to be ass blastingly expensive, Guess I'll just not upgrade or buy used. What's the point? My current parts does everything I need it too?
(And that's how the PC market dies)

>>58468693
The biggest misunderstanding seems to be that hardware sales are somehow relevant to PC GAMING. They are not. If better hardware is not necessary (current situation) or when it turns up is not utilized by devs for anything other than better graphics (situation since 2008 or 09) then the only thing that matters are game sales, and game sales have been steadily growing since forever. There is no crisis in sight on that front. Sadly no Crysis either.

>>58456550
raven ridge when

>>58467783
>there haven't been any concrete news
>Its most likely accurate

Would you like to insert another contradiction or revise your current one?

>>58464314
>certainly it won't be like cf/sli interface
early dual core cpus were mcm, they still shared the same ram and stuff

mcm gpus will probably mean some similar siamese twin shit where they're grafting together dies over some meme interconnect to work around the shit yields and expense of making massive single-die parts like GP100 and Fiji

>>58469162
No, you're mistaking MCM with having two single cores on the same die.

Fuck, I got the 480 a few days before Ryzen got announced. Did I dun goofed?

>>58469275
480 is for poorfags

>>58469275
Ryzen is CPU, the 480 is GPU. The new GPUs won't launch until May.

>>58469275
It draws 600 watts through your motherboard so yes

So as someone who hasn't been following the ebb and flow of AMD, how is ryzen shaping up? Would the 8c/16t chip be worth moving onto from a 6700k?

>>58456838
ABSOLUTE PUSSY DESTROYER

>>58469275
i dont get it
do you mean like you could save the money to put towards a better gpu?

>>58459689
Thank you based jewish repairman of shekel gains.

>captcha: material sion

>>58471467
Depends.
If you only use it for gaming, there's no reason to switch.
But if you need performance and your software can fully utilize the 16 threads, go for Zen.

>>58471467
The engineering sample competes with a 6900 so for gaming it would be absurd to switch. And even for high performance software aside from gaming I can't imagine it would be worth switching away from a 6700 any time soon. Why are people so fucking eager to buy expensive hardware when they're rocking almost top of the line components already? What are you hoping to achieve? One minute less rendering time when editing video?

>announce the supposed launch date
>no prices
>no actual benchmarks outside of their marketing team benchmarks
Oh look it's fucking nothing again

>>58472262
Hype?

>>58465302
it was based on the blender and handbrake demo, where they took the zen numbers as fact and ran the tests correctly on current amd cpus.

Like I said hard to take as gospel truth, however you can see intel cpus gen to gen, some applications gain negative% (said to be a bug) some gain less than 1% some gain 30%, the demo and why it's done.

For these two demos and their situation, the person found the uptick was in the mid 50's

File: 1451266951507.png (45KB, 383x385px) Image search: [Google]

45KB, 383x385px

Holding on my nutsack real tight while hoping for 200-235 dolla 6 core Zen.

>>58465457
seeing as intel apparently has more h1b1 employees then amd at a higher percent of total workers... kind of the pot calling the kettle black.

>>58467303
because when you actually thread, those cpus perform far above what we normally see.

as long as you aren't pushing retarded physics shit hard, there is little reason they can't do good.

>>58467752
H1 and amd typically releases at the ass end, for my shitty math, this is a bit sooner then we may expect.

>>58469118
all we got on vega is slides and some demos that could have been heavily gimped by way of tape. drivers. or clock speed because dev board.

really we got shit besides slides, and those slides are fucking meaningless due to not telling us a goddamn thing outside of buzzwords

>>58471761
most of us are holding out on sandy bridge or earlier, so any uptick here is a massive boost for us.

In gaming, games are starting to hit ram speed bottlenecks, so getting off ddr2 or 3 is a boost for us there.

Then you say render times, well... you have a video that takes lets say 1 hour to render on your i7 while killing your computer to anything but the render, the same on ryzen would be half an hour, or free up one core and its not 35-38 minutes, free up 4 cores and you now take an hour, but have enough free cores to play a game unimpeded. You like playing in photoshop, well now everything you do is even faster.

yea, there is clear performance to be gained across the board from both gamers and from people who do shit with their computers but found it hard to justify more then an i7's cost.

>>58474075
I would expect 6-core Zen to be priced at more like $300+ to position it against the mainstream i7s. AMD is still a tad bit behind Intel in IPC (probably looking at Haswell-tier on average), so offering an additional step up in core count at or slightly below each one of Intel's price points seems like a strategy they'd go for. This would also be in line with Lisa Su's stated goal of attempting to dispel the image of AMD as a budget brand, and we have already seen that policy with products like the Fury X (which admittedly was definitely not worth the money, though the BoM for that shiny AIO and bleeding-edge HBM was probably obscene).

Probably looking at 4C/4T being positioned against the 7350K, 4C/8T being positioned against the i5, 6C/12T being positioned against mainstream i7, 8C/16T being positioned against hex-core i7.

>>58474594
amd has hinted at the 8 core 16 thread at the i7 price now enough that if they dont do that, it will be seen as a failure

first was the public event
amd isnt stupid, they even said 'if we price it at 500, it wont effect the mainstream at all'
durring the event, for the public, and for the gamers, they showed it off, and the most telling was was the 6700 ryzen and 6900

then was at ces where they openly said 'we don't know, 500$ brings the price of performance down, but price matching the i7 changes the industry, still not decided'

Look, they know what they are pricing it at before the horizon event, but they are milking free press for it for as long as they can. because of them hinting at the 350$ range, toying with it, publicly saying its on the table...

Look, it was different when I looked at die size, saw what the 8350 and zen was in comparison and said they have a 100 to 800$ range for the price, a half year before we even seen demos, and sad if they arent shiiting us putting it at an i7 would be a compelling reason for even 6700 owners to consider it, that was speculation.

If amd never hinted and came out at 500-600$ or even just slightly undercut intel, I wouldn't grudge them for it at all. But all the fucking hinting to an i7 range... Look im buying it regardless now that I know the price is at most 500$, I may even buy a binned one for 500 if the base is 350, but if they price it at 500 after all the hints, because marketing decided 'lets go 500 but hint at 350, that way we get press and if we go 500 its still significantly less than an i7' amd will have effectively given up the home market.

6 core that won't clock to what a 4 core can won't sell to gamers, it's not a powerful enough jump forward as 8 core will be and the people on an i5 who would REALLY benefit from a 6 core 12 thread are put of the price range.

for mainstream success the 8/16 needs to be i7 range if not matched.

599 8c
429 6c
339 4c
149 2c
mobo will be priced same as intel

>>58477147
There are no 2 core Ryzens

>>58477147
>mobo will be priced same as intel
Ryzen is SoC, it already have all the things that a motherboard needed.

>>58478171
no?

>>58478171
It's not a complete SoC. It still has a dedicated FCH/SB for auxiliary SATA, USB, and PCIe ports.

>>58478188
well not all but the thing is the mobo is going to cost less than intel counterparts.

>>58478210
No, its a complete SoC.
The Promontory chipsets only add supplemental additional PCI-E lanes and IO.

>>58478214
>the mobo is going to cost less than intel counterparts.
because they're being made cheaply since none of the board partners expect AM4 to sell very well in the first few months. And this is on top of the delay in getting non-OEM AM4 boards out to the public, which caused them to lose potential revenue.

>>58478224
My understanding is that AM4 needs the FCH for the APUs. The current batch of APUs on the Excavator architecture requires an FCH in order to work, and it's likely going to be the same story for the Zen-based APU. Only Ryzen will technically be okay without the FCH present, but it still needs its own special "chipset" in order to work without the additional FCH features, hence the ITX-only A/B/X300 "chipset". Ryzen can not work without an FCH being present, thus it's not a true SoC like the Xeon-D and AM1/FT3 APUs.

>>58477147
this is autistic.
AMD would never leave such a large fucking gap in pricing between the 2c, which wont exist, and the 4c. They'd also be no reason to get a 4c, when Intel still has slightly better IPC from what we know. The 4c would lose to the intel offering, the i7, and would cost just as much.

>>58478267
And yet AMD refuses to let go of Bristol Ridge and keep its stupidly high OEM pricing in the way of cheaper 4-core Ryzens, which allegedly will have no HT.

>>58478283
>4 core
>no HT
If you're talking about Ryzen, and the prices i responded to are correct. No joke, DOA.

if your talking about Bristol ridge, that's a different beast. Isn't bristol ridge they're mobile APU?

>>58478283
>>58478317
think i just r
ealized what you meant
So it'll keep BR stupidly high by creating 4 core Ryzen without HT?

>>58478317
Bristol Ridge's flagship A12-9800 PRO costs about $140 to the OEM. If a consumer version of it comes out, it will undoubtedly cost at least $10 more due to the addition of a cooler and packaging. So if we do the math, the best case scenario for Ryzen's lowest pricing structure would need to be at least $30 more than the most expensive consumer Bristol Ridge in order to preserve its place in the market (ie Core i3 territory).
So that's a $150 hypothetical A12-9800K and a $180 low-end Ryzen at best. That's i5-7400 territory.

>>58474199
They give insight into the architecture.

Important things of note are it's the biggest change since the introduction of GCN (revisions have been unofficially called 1.0-1.3 thus far as opposed to gen 1-4 as amd calls them because so little changed besides scaling it up and making minor adjustments) hence being referred to as GCN 2.0.

also important is the focus on being seriously compute capable. GPU's have thinned the line of what differentiates them from CPU's for ages now and AMD seems to think it's the inevitable future for them. Their CU's are like really weak CPU cores with very limited resources.

CUDA is very similar in concept as well, actually.
What it means is we can see it's not meant to take advantage of how games are rendered traditionally at all. It's more like a computational accelerator than a graphical accelerator.

Basically it's more suited to the CAD and enterprise markets. Which is where the real money is made.
AMD has put the gaming market second with Vega. Which is a shame but securing a steadier cash source will allow them to expand more safely.

They've put off the short term benefit of a killer consumer part for the long term potential that comes with stable income.

I'm hoping it means better driver support continues and they can increase R&D budgets so they don't have to rely on one arch at a time.

>>58478382
>I'm hoping it means better driver support continues
GCN1.0 will be unsupported if GCN2.0 turns out to be that much of a departure, like Terascale from GCN1.0.

>>58456550
>vega
>may
nvidia won, amd lost.

>>58478259
>this whole post
wot
Don't post here any more.
Carrizo and Excavator are both full SoC as well. The "Fusion Control Hub" is on die.
Ryzen as well has this on die. The motherboard chipsets on AM4 are only supplemental, they are not required for anything.

They are full SoC, exactly as I stated.

>>58456550
AMD is shit
Zen will be shit
Intel is based
ARM is based

>>58478382

I don't think people realize how precariously close to the edge AMD is/was to becoming Matrox and VIA.

If Ryzen fails, there goes everything. IP, Patents, technologies, all sold or leased to the highest bidder as AMD becomes a custom embedded chip maker.

By all accounts Ryzen is a competitor, so now that AMD has a competitive piece of silicon on all fronts, they can now focus on professional markets and finally turn all that capital they are spending in to profit. Pay back loans and become not just competitive, but dominant.

If Lisa Su pulls this off, she will have pulled a rabbit out of the hat, and completely turned AMD in to a monster tech company in the next 20-50 years.

>>58478518
>Carrizo and Excavator are both full SoC as well
Except not all of them are, idiot. Even the FT3 Excavators aren't true SoCs in the sense that it needed additional controllers from the OEM partners in order to complete the computer/tablet/laptop, but I'll let that one slide ;^)
The whole of the FCH is not on-die, but most of it is. There are dedicated PCIe lanes for filling out the rest of what's missing and that's where the additional controllers come in.
Ryzen has MOST of this on die, but not everything that is necessary to boot without an FCH present. Hence A/B/X300 "Chipsets" like I mentioned earlier.
Never post here again.

>>58460030
>ps4 pro is using AMD hardware
So? The Nintendo Switch is using a Nvidia SOC.

>>58478569
>I'll just lie and talk out of my ass: the post

>>58460139
>oh yes, i didn't know the normal xbone used amd
Wow, newfag much.

>>58478587

Can't weight for my 4 hours of battery life!

>>58478601
>Can't weight

You can't grasp spelling, either.

>>58478591
You should really do your fucking research before you pull shit out of your ass yourself. Carrizo was a first step towards a true SoC APU outside of the Cat Cores, but it still did not have the full support and features that Bolton or even Hudson had. Thus, AMD left it up to the OEMs to fill in the blanks with their own controllers to complete the missing FCH elements.

>>58478607

You don't like your 720p nvidia shield console?

>>58478621
I don't own a single console.

>>58478601
It's actually 2.5 hours.

>>58478627

Not yet anyways. You have to weight to be disappointed in the Nvidia shield console.

>>58478642
>weight

>>58478616
>still proving you're talking out of your ass
Everything essential to the system is on die, you low IQ autist. Additional chipsets are supplemental.
They are full SoC.

>>58478649

We know you're fat, what does wait have to do with it?

>>58478663
I'm skinny and black.

>>58478661
No they're not, you complete and utter moron
Jesus fucking christ, your mother must have drowned you in alcohol as a fetus before stuffing you back into her uterus
You're so mind-bogglingly retarded AND dense you couldn't possibly even type from lacking the mental capacity to lift your fingers
Is you tech-illiterate handler posting your gurgles and groans for you because you're literally too stupid to combine letters and breathe at the same time?

>>58456879
This.

I went kaby lake and upset the waitfag autists on here.

I have been burned twice due to amd's shitty cpu's.

>>58478563
Funny enough Matrox is still around. Their biggest asset was their superior multi-screen capabilities back in the day and they've stayed alive by offering low power options to businesses who just want many monitors combined together (think like at an airport where all the screens do is display text). Also funny is their latest options use GCN parts with custom drivers.

Ati and Nvidia beat out the others because they were quick to make revisions (sometimes several dozen within a year time span), stayed ahead of the curve, and very very offen just got lucky as hell.

Look at nvidia and ati both. Their first 3 offerings or so in the 3d space were "okay" at best and complete shit at worst. Voodoo was king and only fell because they gambled on buying a company that would allow them to make their cards in house and fucked over the lead they had over the competition.

Nvidia and ati both had instances where they made massive gambles that didn't pay off, but they also had strokes of genius or just lucky breaks that kept them going.

Zen and Vega put the professional markets first. Having amazing consumer parts is meaningless in a shrinking market that is fairly niche. But demand for servers and workstations will always be huge and those sectors mean contracts and steady income.

>>58478480
Its still likely that gcn will benefit from 2.0 development.

>>58478563
their position is... weird at best.
They couldn't withdraw from the cpu market entirely, and would likely get propped up if they fail hard, but they could fully exit the gpu market because it doesn't matter, consumers don't buy their shit, and even when programs are ram dependant, nvidia held ram back and amd didn't, people still didn't get their pro cards either, even though they have to be certified to not crash and pass.

Quite frankly, I could see amd bow out of the gpu market even if they get big funds if navi fails, keep in mind, we are all thinking this will be an mcm with the absolute bare minimum needed to function as a single die, that they will put 1-4+ gpus in one socket. Its the sweet spot strategy on crack, 1 gpu needs to be made, and it has a range from low end to the extreme high end, hopefully this will pull proffit large enough that even when no one buys it it makes enough for them to continue.

>>58478382
No, they really didn't tell us shit, they showed us what may work or what could happen, but they gave us nothing beyond "here is our new never before seen tech" and then showed it off to a fairly unimpressive demo all things considered.

someone said it was a dev board so maybe it was clocked so low when they put it in production it doubles in speed. wishful thinking at best, but i'm thinking maybe 20% better than its current showing, because amd has yet to go, look at our product, its meh, then release it and its awesome, they tend to show it off in best case scenarios.

>>58478752
nvidia is where it is today due to luck, where as amd is where it is today due to innovations and luck.

nvidia had a good, 4 card run where amd offered better at lower cost, and the only way they could compete was work with devs and lock code to their gpus.

on the other hand, amd is such a massive company that even when they fail and get dragged through mud in debt that people keep giving them chances cause if they ever get their foot in the door, those chances pay dividends.

>>58478787
Nvidia is where it is today mostly because of good marketing.

>>58478480
Not exactly. Polaris is still GCN 1-based and carrying over optimizations is fairly simple. Likewise they have a robust hardware task scheduler that makes their "brute force" approach possible. They have no real incentive to provide support now with a full lineup coming, but I guess we'll see.
I've accepted the driver support we've seen is the result of GCN 1.0-based still being sold until mid last year and it's very likely it'll be dumped like any other arch.

But what I really meant was hopefully we'll see the same level of attention to drivers in general, keeping up on new features and game ready drivers.
Being able to alocate more resources and manpower to drivers would mean they can make more unique cards and possibly a vastly different architecture that is more traditional without overburdening their already small driver team.

Vega is a bit of a compromise to the consumer as its strengths are in compute. Low level api's and drivers can improve base performance an insane amount, but it's never an assured thing.

>>58463353
Is it realy a thing to buy Pro card for something else than 3D modeling, movie montage, photoshop and co ?

>>58478787
Nvidia really had a lot of great innovations of their own. The GeForce 256 was the first card to offer a hardware T&L engine which is bigger than you might think.

You may consider their coining of the term "GPU" at this point a gimmick, but it's actually an important distinction (nevermind the arbitrary stipulations they suggested on what a GPU was that read like a list of specifications only their card met) between what "video accelerators" WERE at the time and what they quickly BECAME and ARE.

Prior to this, graphics chipsets were very "dumb" devices, but this marked the fist time they could not only hand complex compute tasks, but also potentially do so better than the CPU.

It took a while before the hardware t&l strategy paid off, but it marked a turning point for the matket. Moving forward, offloading once CPU-only tasks to the GPU became as important as adding feature support and increasing pixel output.

Flash forward to today and we could nearly remove the CPU from the equation if we so desired.

On top of that nvidia was much more clever with how they filled in price segments. Ati's early attempts were most often bios lock-out's overclockers could easily bypass while nvidia was doing some primitive "binning" and revising old chips to fill in lower segments.

They also cleverly made their cards easier to support on the driver side and dedicated an entire facility to providing hands on assistance in optimizing game code for their products.

These days they've been pretty acummy about how they go about things, but as a company they've proven their metel in many ways.

While Ryzen might be good hardware AMDs marketing strategy is quite absurd , first they unveil 3 years or so they are making a great CPU, hype it up , then let the hype die. Now they are trying their best, not knowing that winter time is great for selling CPUs with all leftover sales having gone to Intel's kaby lake. Maybe they don't target end consumers anymore , also their latest Ryzen ad was stupid , the one with this drum player. Can't wait for Intel's response.

>>58458054
I don't get it, did Drake do a commercial or something?

>>58478259
>since none of the board partners expect AM4 to sell very well
msi plans 20 motherboards alone

>>58475770
Kek I can't believe people think 8c/16t chips will be anywhere near $350.
Another
Massive
Delusion

>>58456550
2017 GUARANTEED AMDOMINATION

JEWTEL/NOVIDEO ON MAXIMUM SUICIDE WATCH

>>58471516
Yeah, I didn't expect Vega to come after Ryzen, AMD is fucking around with much money

>>58456550
>AMD Launching Ryzen
> someday, eventually
> we rilly rilly mean it This Tiem

Whatever.

Call back a month or three after the product is actually on the shelf. I'll check out it's performance with reputable reviewers then.

A lmost
M aybe
D eployed

>>58461794
Yeah they're missing the crowd that wants to upgrade for Mass Effect

>>58479979
>reputable reviewers
but anon
they don't exist.

>>58480057
Linus

>>58480039
>upgrade for Mass Effect
If it's console-compatible it just cannot be good enought to require an update.

>>58456550
GTX 1080 Ti will come out and make AMD embarrassed like always.

https://youtu.be/C4BUb6wSSXk
Is he right?

>>58480310
Yes and no.
Vega is clearly focused at enterprise markets where high compute matters most, but it doesn't mean AMD is "giving up."
Winning share in these markets has mucb higher payoffs than the niche high performance pc gaming market, but securing steady cashflow allows them to increase their R&D budget and thus release and support a wider range of products.

They've been forced into a position where they had to make a modular arch they can make small revisions to in order to reduce costs associated with multiple chipsets and still provide a full product line.

We saw the limitations of this tactic with the fury where the shaders were bottlenecked by the pipeline, leaving many unutilized.

Vega is a compromise but it's a calculated risk.

>>58480310
Sort of. We know nothing about clocks. And Vega is aimed at compute market anyway. Lotsa flops for reasonable price, what's not to like?

>>58480310
never heard this kind of pajeet accent

>>58460410
sony imaging is making money.
sony finance is making lots of money.
stop spouting shit.

>>58461697
that's not a weird mix.
a weird mix is fx8350 + nvidia.

>>58479176

>being this clueless

>>58479771
probably because of the new horizon demo where they hinted at it, and ces where they out right said it.

Marketing could be be fucking horrible once again and fuck them by hinting and saying 8/16 at 350 could happen, and then they push it to 500$ because 'that still less then intels 8 core' and kick all the enthusiasm out of ryzen form the get go.

Its not going to be a bad cpu, it will likely REALLY take off in the enterprise market because it has skus where intel has no answer for it, and it will have skus where its lower power and if the 6900 is anything to go by, potentially faster too, as the base clock while still hitting the 95 watt envelope is going up. but it could end up being a wash for amd.

>>58480408
it's called irish.

>>58479795
cpu wise, fuck yea, gpu wise, it looks like its another 'sow the seeds for future gains' deal.

>>58480310
Fanboys have been worshipping this guy since his "amd masterrace" videos like 2 years ago. He seems to be super dedicated to everything amd related and if he's saying it's going to be shit it probably will be. His logic is better than all the other retards trying to prove otherwise though like that channel "Good Ole Gamer" who tried to use math to calculate the performance based on older gpus.

>>58480124
if you buy it after 3, you are retarded to boot.

remember, pee bee is the name of a character that will be said often.

>>58479771
yep the 8c/16t is at least 500+ because AMD marketing people were hard at work trying to bring people to this idea at ces, they dont want to be the cheap option anymore like lisa said.

>>58480521
https://youtu.be/Vl7r-zLq7wQ

Any chance of a dual core Zen?

>>58480636
Nah.

File: Bait of next-gen.png (120KB, 625x626px) Image search: [Google]

120KB, 625x626px

>>58464521

>>58480592
This same logic would suggest the fury would be massively superior to the 390x and we all know that's not true.

Compute performance =/= gaming performance.

Better drivers will definitely help a lot though. I imagine they used fiji drivers as they already include optimizations to play to the strengths of HBM, but obviously the arch is different in so many ways small modifications to the pre-existing drivers aren't enough to make the most of it.

We'll see on release but this kind of speculation is silly.

>>58480663
It's not impossible. They could disable 2 cores on a quad core part if they feel the need to offer such a thing.

>>58480740
There are no quad-core SKU, it's not fucking Intel. Quadcores are cutdown octacores.

>>58458693
>student loan comes in in September
i can wait :^)

>>58480723
Ironic how the guys on that channel were making videos only a couple weeks ago saying how everyone shouldn't create unnecessary hype over amds products so there isn't any backlash if the products aren't as good as people had hoped.

AdoredTVs video comes out saying vega is basically DOA and now they're super butthurt and have gone on full deluded fanboy mode and making rash claims. This will end up biting them and amd in the back if it catches on (which overhype like this actually does) and we'll be back to square one. It's the fx series release all over again.

>>58480759
I know that but there should be no reason they can't cut down further to create a 2 core. They could even just disable 2 cores on a perfectly fine 4 core to fill the spot. They've done similar stuff in the past.

>>58480838
Why would you disable 2 cores on the perfectly fine quad-core?

>>58480849
The same reason the first run of 4gb 480's were 8gb cards with a different bios, the same reason some 6/3 core phenom parts were perfectly fine 8/4 core parts with perfectly fine cores disabled and the same reason ATI took fully functional GPU's and underclocked them just to provide more options: to meet demand from every corner of the market provided it sells.

But I didn't say it was a good or realistic idea, just that they could.

>>58480838
The chances that at most 3 of the 8 cores are functional are pretty low.

>>58480379
amd DOES NOT SELL TO NORMAL PEOPLE, at least in the gpu market.

Amd made the best mid range card we have seen in what... 8 years? what should be the highest selling card, and more people bought 1080's 1070's then they bought that gpu, even 1060's which arguably was better depending on location.

Amd doesn't sell to normal people.

If I was amd, I would just fucking abandon consumer gpus entirely, They are not worth it, and focus on the prosecutor, with your research there fueling what trickles down to consumers, at least till you get mcm gpus figured out.

>>58480592
jesus christ this guy is such a fucking idiot, his first video on vega just had to try to prove adore wrong by saying 'wait giz, this is only at half its real clock and fucked drivers, it's going to double the 1080s performance easy'

going by math, this gpu will be fucking amazing, going by the same math, the 290x should be equal to a 980ti, but it rides bitch at around 60% as powerful at stock, drivers have closed that gap significantly, but its still riding bitch by a significant margin.

>>58480792
adored is right and wrong at the same time, it may not blow a titan xp out of the water, but price is a VERY important factor. amd is capable of selling chips bigger then this for 300~$ and because same die for business and consumer, consumers will be riding bitch on binning like with polaris... this means the price for these chips that are completely fucked for enterprise could be cheap just to stick it to nvidia and get market share as a pseudo loss leader.

I don't believe the clocks or drivers for the chip are nearly as fucked as everyone defending amd thinks they are, but I look at this realistically, get get the chip out there to bug fix on a somewhat mass scale, and because these are scraps, they don't lose much if anything but can gain.

>>58480838
The chip is made as a 8c with a fairly mature 14nm process that has already been producing excelent yields (see Polaris 10), so there won't be all that many half the cores not functioning silicon to make 4c products out of. Not in the volumes needed to meet a pleb tier priced 4c demand at least. That's why they have Raven Ridge 4c, 4c+igpu and 2( or 3)c+igpu lineup priced at $50-200 coming later this year, and why the Summit Ridge is going to be HEDP only lineup with only 6c and 8c processors offered at the 250-600 price bracket

>>58480908
>The same reason the first run of 4gb 480's were 8gb cards with a different bios
nope, just few cards send to press were like that as they didn't have the final 4gb version ready

>>58480581
and then they put out fury x at a VERY reasonable price when they had between a 980 and a titan x to price between.

Im takeing 'we don't want to be the budget brand' as meaning 'we don't want to be seen as shit you get when you can't afford the good one, we want to be the go to option'

That doesn't mean that they need to be expensive, that just means they don't want to make shit products, and so far, nothing they made after that statement has been shit.

>>58480759
the apu will be a quad core, thats where you will get a dual core sku from.

>>58480838
amd is only making 8 core dies, to make a 2 core they would have to completely fuck a 8 core up to the point it useless, even a quad core requires a half fucked 8 core.

likely the 4 core 8 thread sku will be 150~$ and later on the apu will have a fucked gpu sku that goes down to dual core for sub 100$

>>58481113
first run 4gb were 8gb subsequent ones were real 4gb

>>58481082
It honestly depends on how the cores can be deactivated, because we could see an entire range of 4 5 6 7 and 8 core cpus if they can go individually.

>>58481113
There were people who literally got 8gb 480's with a sticker on the box covering the 8gb number.

It was only on reference cards from the first run but there were quite a lot.

You're missing the point though, AMD, intel, and nvidia have all had cases where a part that is normally a binned version of a higher tier part has high enough demand to warrant just selling the more profitable part with functional parts disabled.

It's a fairly rare thing and I doubt they'd bother but it's still "possible."

>>58480663
there is going to be 2c APU though

>>58481428
That comes later.

>>58481070
vega is probably around the same perf as 2x rx 480, we can at least reasonnably assume that

>>58479176
Newfag

Another
Massive
Dissapointment

File: bane.jpg (79KB, 716x768px) Image search: [Google]

79KB, 716x768px

>>58481442

>>58481886
>retards spam their unrealistic expectations
>other retards parrot it
>other retards listen to the vocal retards
literally the only way you'll be disappointed tbqh senpai.

>>58481177
>Im takeing 'we don't want to be the budget brand' as meaning 'we don't want to be seen as shit you get when you can't afford the good one, we want to be the go to option'
Ah, so the samurai and pajeet wants to be the jew now.

>>58478365
But wont RavenRidge replace Bristol?

Any pricing leaks? I'll finally be able to replace my FX 8 core going to be so happy.

Replacing with a kaby lake i5 would run me around $500 hopefully AMD offers a lot more for the money

>>58482875
From what I'm hearing in this thread, they won't. This could be an incredible chance for AMD to damage Intel in the market, but apparently they'd rather pay off some debts quicker so they can get right back to getting fucked by Illegal Intel deals.

>>58456807

Well kind of.

The cpu seems legit. But the GPU is barely beating the 1080. Which is laughable at best.

File: 1484357505875.jpg (291KB, 863x752px) Image search: [Google]

291KB, 863x752px

>>58482929
Reminder that it essentially in an box with no airflow and still has 5 months till launch as well

>>58482852
Raven Ridge launch is estimated Q4 2017.
>>58482939
And it was an engineering sample. When was the last time NVIDIA showed engineering samples?

>>58480408
>>58480515
he's Scottish you blithering retards

>>58482875

for 500 Id think you could get a top end mobo and the 6c/12t Ryzen, or a top bin 4c/8t with extra cash to spare.

>>58482033
>amd
>not mostly overhyped dissapointment

k

>>58480408
I fucking love scotish accent. I love it so much that I even try to mimic it, even though I'm not a native english speaker. This way I can say the "r" like in my native language and don't need to gay things with my mouth.

However what he says makes sense but I hope he isn't too right. Who wants to buy nvidia after their NVIDIA-Spy shittalk at CES.

>>58483438
I think I'll end up going ITX or waiting for Ryzen APUs for my itx rig

>>58485330
>I fucking love scotish accent
eww

>>58457024
>mfw bought a shitton of $11 Jan17 calls for dirt cheap then turned them around in 45 mins for 100% profit

>>58461738
>>58461767
>>58461784

I play Siege on 1080p@120Hz with an i7-2600k and an RX480 8GB and I consistently get 100+ fps.
You must have shitty settings (hint: use only FXAA, if at all, no AA other than that; set shading to medium or low; set Postprocessing to off)

Will it at least = 1080 at a cheaper price so I can grab a freesync monitor?

Have there been any plans to put Ryzen in laptops?

>>58489735
Raven Ridge, google it.

>>58489735
The Raven Ridge APU due late this year is a 4c Zen plus RX460+ level Vega, which should be great for high-end laptops.
2c-3c binnings might end up in the channel too.

>>58482826
jew means they give you a shit price, amd may go shit price at some point but that's much further down the road. they want you to chose them not because they are cheap, but because they are the better option that also happens to be cheap.

at least by actions so far that's the only way i can take it, even if they put the 8 core to 500, it would still be following this, just they would be fucking themselves due to marketing.

>>58481465
If it wasnt that would be pretty shitty given size.

>>58481203
That was a joke image
"Amd solf you 4gb, you get 8, nvidia sold you 4, you got 3.5"

>>58481428
apus are either first half, or calendar year, I forget which one.

>>58483151
When was the last time nvidia showed performance on anything not finished and already out? they will hold shit up with wood screws before they risk a crash on stage, thats not really helping amd right now though.

>>58482875
good motherboard looks to run 200$, and cpu will be between 300 and 500 for the 8 core, they fucking go 500$ for the 8 core then the 6 core is pushed to i7 price bracket. and the 4 core 8 thread is pushed to the i5 bracket.

>>58482928
You are listening to morons who think amd has to charge a few$ less than intel for equivalent parts when amd themselves have all but said the price match i7 with 8/16. marketing could be thinking this is a great move and then shit it out at 500 which would be retarded, but they gave their range, price match i7 to 500$ is all they are looking at for the base 8/16

>>58482939
put up or shut up and amd still isn't putting up.

File: 5e0a010ff11fbb9dc26404bb3b1a57224cfa4dd1_hq.jpg (26KB, 370x398px) Image search: [Google]

26KB, 370x398px

>>58490358
>(You)'d 3 times in one post
I'm finally getting good at this

>>58490275
>>58490293

>APU

Yuck... Never again