ZEN FX Processors in Late Q4 2016

Go amd go you can do it fucking poo Lords

Is this a whole new socket? I want an amd mitx build and am3 doesn't go on mitx for some reason

File: IMG_20160713_195942.jpg (187KB, 1024x1024px) Image search: [Google]

187KB, 1024x1024px

Haswell-E for half the price here we come
Amada for good luck

File: 1468231342302.png (631KB, 858x856px) Image search: [Google]

631KB, 858x856px

>>55705232
>not posting the original logo

>>55705271
It doesn't go on mITX because it has a discrete northbridge. You can't fit all that shit on mITX without making some serious sacrifices. Zen will have a new socket.

File: 1464750240878.jpg (107KB, 604x531px) Image search: [Google]

107KB, 604x531px

>>55705302

>>55705271
Zen will be using socket AM4, it'll be a new chipset and it'll be shared between both the APUs and the dedicated CPUs. Can't say anything for certain at this point but it's almost guaranteed that there'll be mITX boards.

>>55705232
Polaris wasn't a disappointment. They promised the previous gen's high-end gaming and recent VR at a 200$ price point. They delivered it. The fact that nVidia has a comparable offer doesn't take away that Polaris wasn't by any means a disappointment.

>>55705310
>>55705346
dandy, thanks gang

AMD please don't fuck it up...

Zen will be an even bigger disappointment than Polaris

>>55705346
AM4 boards don't have a chipset.

Phenom II was awesome right, maybe it'll be another phenom II kind of thing

>>55705232

Bulldozers was shit.

Fury is a gem and shows its power now in DX12 and Vulkan.

Polaris made Nvidia shit their pants, forcing them to do 2 paper launches. It's still an EXTREMELY good choice, and I'll upgrade to it from my GTX580 giving me almost twice the performance for 250 euros. (I'll be picking the Nitro+ which shits at any version of the 1060).


When you spend 500-600 dollars/euros on a card, you don't spend it to keep it for 6 months. It's prolly a card that will remain in your system for 2-4 years. The fury X demolished both the 980 Ti and Titan in DX12 and Vulkan and shows that it was a far better choice for futureproofing,(unless you are one of those retards that don't believe in futureproofing). Also power draw is a fucking joke. If you have the money to spend for such expensive GPUs an additional 50 euros annually in your electric bill won't matter. I'm not a fanboy. I frankly don't give a shit about either company. But currently I see that AMD offers me better choices in the GPU department for the money. The 1080 costs 950 euros here. That's around 1100 dollars. I can easily get a used Fury X for around 350-400 and get 15% less performance with better futureproofing.

>>55705425
They won't have a northbridge but they'll have something vaguely resembling a northbridge as well as a few other necessities. It'll still be a chipset, we're not quite to the point where a mobo is just a break out board.

>>55705232
AMD claims an IPC improvement of 40%.
If they follow their usual behavior this means IPC performance decreases in reality by 4% and over all performance gain is achieved by extreme power draw and high stock clocks with per core performance still being on the level of a stone.

>>55705476
I think the nbridge is on the CPU itself now

>>55705453
>unless you are one of those retards that don't believe in futureproofing
But future proofing is generally speaking retarded. There will always be some new something preventing your system from keeping pace with the new ones to the point where you might as well buy mid range and stay happy with that for ~3 years.
>an additional 50 euros annually in your electric bill won't matter
Are you running furmark 24/7 while powering your card with a diesel generator?

>>55705501
This tbhfam, Zen is another bullshit 'module' design its IPC is gonna be lolz

>>55705517
Yes, you're absolutely right and I'm an idiot for typing
>something vaguely resembling a northbridge
I meant to say southbridge. Apologies anon.

>>55705547
Intel boards have southbridge too so nothing's unusual there. Well, it's not really called a southbridge when there is no northbridge.

Intel calls it the PCH. God knows what AMD will come up with. HyperTransportHub maybe. HTH

File: amd-soc-and-promontory-chipset-io-coverage-1024x724.jpg (195KB, 1024x724px) Image search: [Google]

195KB, 1024x724px

>>55705476
Northbridge and Southbridge are on die.

Socket AM4 has a supplementary chipset code named Promontory that will differ based on the segment a board is targeting. Premium boards with the most expensive chipset will have added SATA, USB, dedicated NVMe, and other I/O.

>>55705529
Totally independent cores.
Shitpost elsewhere.

File: promontory.jpg (105KB, 560x750px) Image search: [Google]

105KB, 560x750px

>>55705476
AM4 boards don't have to have a chipset. Southbridge is on-chip - secondary Southbridge (Promontory) is optional.

>>55705346
Test board Myrtle is mITX.

>>55705581
Here he is, the AMD shill pretending to be smart!

I can already foresee Zen being equal or slightly above current competing Intel chips. Like with the 480 and nvidia answering with the 1060, Intel will have their finger hovering over the big button to push out a product soon after that will do better but not by much.

>>55705501
>If they follow their usual behavior this means IPC performance decreases in reality
I'm trying to follow your logic here but you're not making it easy. An increase in IPC is an increase in IPC. That's completely independent of clock rates. As for power consumption, rumours have consistently pointed towards a TDP of 95 watts, which means it probably won't be too power hungry.
>>55705529
>Zen is another bullshit 'module' design
You very clearly have not been paying attention.
>>55705565
Exactly why I said "vaguely resembling", AMD has referred to it as a Fusion Controller Hub on some of their APU boards but they dropped their "fusion" branding so god only knows what they'll call them now.
>>55705581
>Socket AM4 has a supplementary chipset code named Promontory
Again, vaguely resembling a southbridge. A rose by any other name and all that.

>>55705453
That delusion lmao. Fury is a gem ayyy.

https://www.youtube.com/watch?v=UXT6I2--cJ4

If true, the OC'ed AIB 480's beat out the best OC'ed AIB 1060's in most DX11 titles, and all DX12/Vulkan titles..

Very impressive

File: Zen_Summit_Ridge_First.jpg (73KB, 369x676px) Image search: [Google]

73KB, 369x676px

>>55705599
>get called out for lying and shitposting
>keep shitposting

Look at the little 14 year old /v/tard So cute.

>>55705623
Performance per clock will land somewhere around Ivy Bridge and Haswell, might excel in certain metrics, but wont compete on the FPU front.
I don't think they'll have much OC headroom though. Probably a nominal clock around 3ghz, maybe a few highly clocked boost states to help out in more serial workloads. Might be interesting if they could clock each core complex independently. Gate one entirely off and push clocks higher on one to make use of the chip's full TDP on 4 cores.

8 core Summit Ridge will probably fair well against Haswell-E, might under cut the mid range Broadwell-E on price and total value.

lol AMD is shit drop dead pls

Can't wait until it comes out and surprise surprise it's major piece of shit with serious constraints here and there and then the fanboys and shills start backpedalling on the hype and the promises trying to rewrite history and pretend nobody was hyping the thing at all.

They did this exact same shit with bulldozer, vishera, the 300 series that wasn't supposed to be just rebrands, the shitty fury lineup including the shitastic "overclocker's dream" and now poolaris that is just mid tier crap whereas the enthusiast market is 100% dominated by nvidia.

This is why after falling for the bulldozer meme I never cared for any of these shitty AMD promises again. It's ALWAYS the same bullshit and they never deliver on anything.

>>55705694
100% fake AMD is incompetent
Only ashill would believe that

>AMD


LOOOOOOOOOOL

>>55705523

Tell me what features I'm not enjoying with my 2500K at 4.5Ghrz, 8 Gigs of ram and GTX580.

Because for the whatever gaming I do, and pc requirments I have.. I'm TOTALLY satisfied. I'm not one of those mongoloids who begs to play AAA games on EXTREME at 4K with hairworks or whatever and DEMAND 60FPS.

Can I reach high end ranks in any of the 4 games that I'll list you with my PC?

>Hearthstone
>LoL
>WoW
>CS:GO
>Dota
>Any emulated old school Znes or Playstation 1 game


Can I render a video or a movie in a satisfying enough time? Does my PC responds instantly on everything and boots in 10-15 seconds? Is the experience smooth overall?

The answer to everything is yes. A big fucking YES. And I can say this now even after 4 years of using this PC. So tell me exactly WHY should I upgrade, and WHY should I spend fucking 950 euros for a double in my performance just on some visual clarity in 1440p which I don't even use. What are the so called features which I miss currently from a pc using a 1070 and a i7 6700K with 16Gigs of DDR4.

30 years company and it has never won against INtel or Nvidia and it will never ever win because AMD is a huge shit

File: 1435465150291.jpg (889KB, 4800x4200px) Image search: [Google]

889KB, 4800x4200px

>>55705694
>1420 Mhz RX 480 beats 2100 MHz GeForce 1060 in Witcher 3
>LEL

Everybody knows amd doesn't have enough megahertz to deal with Nvidia

>>55705754
AMD used to dominate Intel... Just saying. Think Pentium 4/Athlon 64 era.

>>55705814
Because Intel let them :)
Intel could have at any time beat them into the ground

It's competitor is Kaby Lake which is Skylake +3% at best.

File: ImjrboO.jpg (236KB, 1280x960px) Image search: [Google]

236KB, 1280x960px

cant wait for zen to flop and were still using quad cores for mainstream and dual cores for entry level for another 10 years thanks papa intel

>>55705891
Summit Ridge is not a mainstream desktop chip. Kaby Lake is not its competitor.

>>55705915
Yeah sure retard.

>>55705351
He's not wrong.

AMD did good.
Nvidia just happened to have an ace (and ridiculous marketing cred) up their sleeve.

>>55705232
piledriver was awesome fag

File: firefox_2016-07-22_16-39-06.png (35KB, 801x628px) Image search: [Google]

35KB, 801x628px

>>55705232
>disappointment
what?

>>55705780
Kek'd legitimately

>>55705698
but what's in the cores? im half expecting "40% more ipc" to be made up in needless fucking integer units and it won't even be better for single threaded tasks than an i3-2100

also is that a 4-way shared cache? l2 or just the l3? WHATS IN THE CORES? WHY IS IT A SECRET WHAT KIND OF DISAPPOINTMENT ARE YOU HIDING AMD REEEEEEEEEEEEE

>>55705590
That's micro atx, dingus.

>>55705753
This

>>55705698
>>get called out for lying and shitposting
>>keep shitposting
>Look at the little 14 year old /v/tard So cute.
I'm the anon you replied to in both posts mentioned in >>55705590
I'd like to point out that I'm not >>55705599
and I'd like to reinforce my previous statement. Promontory is southbridge-esque enough to fit the definition of a chipset as I've understood it.

by end of 2017 rx480 going to be 20% faster than 1060/980 due to amd performance upgrades and nvidias intentional downgrades.

>>55705985

the secret AMD is hiding is that zen cores can only be disabled in groups of 4 and they clock like shit compared to bulldozer and skylake.

zen A0 4 cores = 3.8ghz overclock on a golden chip at most. you won't break 4ghz or 5ghz OC.

>>55705985
That's L3 cache. I don't really like the look of that die, it looks like two 4 core processors on 1 package.

File: Zen-Architektur+Core+V0.3.2.png (115KB, 640x629px) Image search: [Google]

115KB, 640x629px

>>55705985
>I don't know about linux kernel patches
Pay attention, no one is hiding anything, and the Zen core architecture is detailed in full at HotChips next month.

Its a 10 wide arch, 4 wide general int core. It doesn't use the FlexFPU from the Bulldozer family. Each core has a private L2, and they share an L3.
The L3 appears to have a beefier version of Jaguar's L2 unit that maintains coherency.

>>55705931
Summit Ridge is enthusiast class. They're competing against intel's i7E line.
They are not mainstream desktop parts.

Stay in /v/ with the other children if you want to behave like one.

File: 480 Asus STRIX OC.png (44KB, 500x1210px) Image search: [Google]

44KB, 500x1210px

>>55705351
>14nm
>still hotter and draw more power
>low OC even the AIB

ZEN ES - 2.8/3.2(turbo) @ 95W

>>55705985
>WHATS IN THE CORES? WHY IS IT A SECRET WHAT KIND OF DISAPPOINTMENT ARE YOU HIDING AMD
Have you been living under a rock? There have been block diagrams floating around for months now. If you haven't been paying attention, why are you posting?

>>55705440
>maybe it'll be another phenom II kind of thing
>60%-70% lower floating point performance than intel offerings of the day like the i5-750/i7-920

please god no

>>55705663
>I'm trying to follow your logic here but you're not making it easy
It is pretty easy.
AMD has claimed such shit befor and never delivered.
Instead the IPC performance even decreased in some known cases and an overall performance gain to a prior generation was only achieved by high clocks and power draw.
>An increase in IPC is an increase in IPC. That's completely independent of clock rates
As said before claim and reallity regarding IPC are really different things for AMD.
And I am fairly sure this was easy enough to understand the first time.
>>55705663
>As for power consumption, rumours have consistently pointed towards a TDP of 95 watts, which means it probably won't be too power hungry.
That is why they have TBA written on it.
Also note that TDP=/=power consumption.
For comparision last gen i7s with a TDP of 90-95W resulted in an actual power draw of roughly 120W on full hard load.
The FX-8350 with its TDP of 125W is known to achive a up to 200W on full load, the monster and abomination that is the FX-9590 withs its 220W TDP achieved a real world power draw of 400W on full load in tests.
TDP=/=actual power draw
TPD=thermal power design

This is why many board manufacturers removed official 8 core and 125W TDP support from many of their mainboards unless a big top blow fan is used that supports the board.
AMD chips are known to put a typical board under a lot of stress if they are on load.
It is also a very common problem that I see in my part time work.
People buying the AMD chips because of muh 4GHz-5GHz and 8 core and go on to cheapskate on the motherboard and often just want the cheapest one that has like four phases and no heatsink.
Best is when they also want like the cheapest PSU.

File: GCN vcore-clock.png (37KB, 1171x703px) Image search: [Google]

37KB, 1171x703px

>>55706034
You should at least have a basic understanding of ICs before you try to critique architecture.

It is two core complexes, yes, but their physical proximity or lack thereof is not indicative of anything. You are looking at the logic on the front end of line. All signaling and power delivery is done through the back end of line. They are physically connected to one another no matter where they are on the front end.

>>55706056
Polaris 10 is intentionally clocked significantly out of the architecture's sweetspot for reasons obvious to everyone with an IQ over 90.

>>55705833
Whatever hepls you sleep at night

>>55706060
How much cores?

>>55706100
8

>tfw i bought the 9590 literally as a collectible so that i can proudly say i own an unopened unused version of the hottest most power-hungry out of the box x86 cpu that has ever and most likely will ever be produced by either amd or intel

i bet this thing will be worth a fortune in 20 or 30 years for that fact alone

>>55706115
So 8 half cores again? LEL AMD DOA

>>55706130
kek

>>55706060
Who would believe that shit? 95W and 8 cores? LOL, even intel needs 140W and they're 10 years ahead of AMD

Half cores ala Bulldozer confirmed

>>55706082
>AMD has claimed such shit befor and never delivered.
Please, show me where they have made such claims. They claimed that the FX series would perform well in multithreaded workloads, which it did.
>Also note that TDP=/=power consumption.
You have absolutely no reading comprehension. From my post;
>rumours have consistently pointed towards a TDP of 95 watts, which means it probably won't be too power hungry.
Lets break that down
>Rumours
I specifically mentioned that they're rumours, no need to go off on your pointless tangent.
>TDP of 95 watts, which means it probably won't be too power hungry.
Since that sentence isn't clear enough for you, I'll make explicit what was just implicit; the use of the word "probably" here very clearly indicates that I am NOT implying that TDP is power consumption and you are a mucous eating oaf for thinking it I am.
>The rest of your post
Nothing other than the rantings of a feral idiot.

>>55706127
Might be a good investment.
Doubt anyone will ever redo this garbage.

Best part was that they originaly intended to sell this in the price range of the big Sandy/Ivy/Haswell-E consumer/semiprosumer chips.
Even better was that there actually existed idiots that bought the chips at those prices.

>>55706196
Zen cores are smaller than intel's latest Core I arch, and they don't have anywhere near the die area dedicated to FPU logic.

Zen absolutely will not compute with intel on FPU performance.

>>55706196
I have 12 cores at 105W. That's on the 22nm node. You can probably get 8 cores at pretty good clock rate with 95W on a 14nm FinFET.

>>55706216
How is that any different from what I said you stupid shill?

>>55706234
They aren't CMT "half cores" you shitposting child.

File: 2f7.jpg (28KB, 601x508px) Image search: [Google]

28KB, 601x508px

>>55706216
>Zen absolutely will not compute with intel on FPU performance.
>because of how much die area is dedicated to it

MAYBE THE ARCHITECTURE IS JUST THAT EFFICIENT

>>55706240
Why are you so mad at being wrong shill?

>>55706216
Compete with what? 512bit AVX?

File: 1466843939604.jpg (20KB, 655x360px) Image search: [Google]

20KB, 655x360px

>>55706240

>>55706245
Theres no way to get around larger data paths.
The Zen FPU has two 128bit data paths, intel uses significantly larger native 256bit.

>>55706263
Any vector workload, anything else that heavily uses FPU logic, like X265 encoding where Skylake is 80% faster than Excavator per clock.

>>55706211
Yeah i remember the initial pricing it was IMMENSELY high, one of the greediest things since the latest broadwell-e line from intel, because they were literally giving people nothing more for that money than an oc'd 8350. I don't know why people defend AMD morally after shit like this.

File: cckpWx6.jpg (3MB, 3413x1440px) Image search: [Google]

3MB, 3413x1440px

>>55705946
I wouldn't call the 1060 an "ace" - it's comparable to the 480 in dx11, but struggles in vulkan/dx12, and it can't SLI, whereas the 480 gets significant benefits from CF in vulkan/dx12.

And on top of that the 480 4GB is also $40 cheaper than the 1060 3GB.

That's not AMD being BTFO or countered by some "ace up nvidia's sleeve"; not even remotely.
The only thing nvidia has going for them is the marketing/lugenpresse/shills.

AMD's card is not only futureproofed by both vulkan/dx12 support and better CF support that the 1060 can't match, but it's also the cheaper option, AND if AMD's past driver updates are anything to go by, will increase in performance over the next year or two until it's 5-10% ahead of the 1060 even in dx11 games.

And nvidia is going to really be hurting if AMD price drops the 8GB 480 to $200 at some point in the near future - which is very likely, considering the 4GB version is literally just an 8GB model with 4GB disabled by the bios. At that point it's a $200 8GB 480 vs a $250 3GB 1060 - that's even assuming nvidia's able to ramp up production to end the supply issues with the 1060 so it's actually selling at $250 instead of the $300+ it currently is.

Nvidia's literally up shit creek if, two months from now, it's AMD's $200 vs Nvidia's $300+ for +5% performance in dx11 and -20% performance in vulkan/dx12; without any SLI options on top of that.

The only people buying the 1060 right now are those who can't see through the nvidia shilling, and/or haven't done their research.

>>55706130
muh modules do

>>55706298
>The Zen FPU has two 128bit data paths, intel uses significantly larger native 256bit.
How long have they done this? Is it every chip since x generation or only higher end ones?

File: Llano_vs_SandyBridge_vs_Westmere.jpg (501KB, 1024x1600px) Image search: [Google]

501KB, 1024x1600px

>>55706326
Its been that way for quite a while. FPU throughput has always been intel's strong suit going back a decade now. They've made considerably large cores, with an utterly massive portion of total core logic dedicated just to FPU logic.

http://www.realworldtech.com/haswell-cpu/4/

People literally return 3 AMD GPUs for every Nvidia GPU, and Nvidia has 4 times the marketshare.
Think about that a little.

>>55706198
>Please, show me where they have made such claims. They claimed that the FX series would perform well in multithreaded workloads, which it did.
Maybe look into the archive, there was I believe more than one thread the last few days that talked about AMD having done this.

>You have absolutely no reading comprehension. From my post;
Since you complain about reading comprehension, let us take a look back to this
>>55705501
>over all performance gain is achieved by extreme power draw

It was you that went from power draw to TDP and also said this
>rumours have consistently pointed towards a TDP of 95 watts, which means it probably won't be too power hungry

The one that made direct correlations of TDP to power draw was you and you alone and that is fucking wrong and thus I have to doubt your reading comprehension.
Hell sure we could leave it at that if you just used the correlation and said rumors, but you especially had to say consistently.

You can try to down play your shit with ranting on about using probably, but you tried to imply a correlation here.

>Nothing other than the rantings of a feral idiot.
Also the fact that you simply discredit known facts does not make your case look better.

Probability of shill 40%
Probability of AMDfanboy 55%
Probability of simply being retarded while still having the mental capabilities to write on /g/ 5%

Probability of ranting angrily 100%

>>55706308

There's no 3gb m8 (not yet at least). There's a bunch of 6gb aftermarket cooler options for $250 on newegg

And i'm assuing rx480 AIB cards will be around 199-230 for 4gb, and 239-269 for 8gb, which would make the nvidia cards very competitively priced.

Like the sapphire nitro, its good looking as fuck and has a backplate, I really doubt that it will be priced at 239 like the reference cards were

>>55706370
Why isn't AMD taking steps to improve floating point performance? Seems to be the most important metric these days. Is it the insanely lower RND budget they have that just makes it too hard?

>>55706430
Price drops won't happen until after the AIB's get in on the launch price gravy train.

They'll get a month or two and then AMD will cut prices in order to gain marketshare vs nvidia.

>>55706308
>struggles in vulkan

kek

>dx12

show me a few non AMD-optimised games where it struggles

>1060 3GB

doesn't exist

>AND if AMD's past driver updates are anything to go by, will increase in performance over the next year or two

This is a myth you stupid AMDrone. AMD GPUs have got faster because CPUs have got faster, lessening the performance penalty AMD's high overhead has. If you want to enjoy this "increased performance", then have 400 dollars ready to support Shintel every few years,

File: far-cry-primal.jpg (33KB, 736x464px) Image search: [Google]

33KB, 736x464px

>>55706462

>>55706462
>show me a few non AMD-optimised games where it struggles

show me a non rise of the tomb raider title where it doesn't

Shall I buy an i5 now or wait for the Zen?

I'm about to give up on AMD completely after the shitty 480 I had high expectations for.

File: shadow-of-mordor.jpg (34KB, 747x487px) Image search: [Google]

34KB, 747x487px

>>55706462
2

>>55706056
>it's 3% worse
OH WOW STOP THE FUCKING PRESS, BLUNDER OF THE CENTURY
Thanks for proving my point, retard.

>>55706477
Every Far Cry since Far Cry 3 has run better on AMD.

>>55706485
If you think the 480 was somehow a bad launch, or a bad card, you're either drinking the nvidia koolaid or completely uninformed.

>>55706446
Larger datapaths will suck down an ungodly amount of power, and require a whole host of IP specifically for them to keep them under control. Intel's latest chips actually have two separate clock states depending on what size vectors they're processing. Simply flipping them on to crunch larger numbers consumes a stupid amount of power that eats away at TDP. Thats why they stowed the latest AVX for consumer chips.

Everything comes down to cost, and if you're a company like AMD who has a finite budget, you spend it where it'll do the most good. They simply can't afford to try and meet intel in every metric.

File: gears.jpg (34KB, 697x447px) Image search: [Google]

34KB, 697x447px

>>55706462
3

>>55706485
keep in mind there's 2 entirely different teams working on gpu's and cpu's that are thousands of miles apart.

but anyway i regret buying my 4790k because it seems the age of quad core mainstream may be ending very soon, it definitely isn't going to be the modern future-proof equivalent of the 2500k i was hoping it would be if Zen ends up not sucking balls, we've at least got news that 6 cores will be the new standard in 2018 due to coffee lake, but that may be made even sooner if Zen kicks ass and it might be more than 6...

What's your current cpu? If it's just total garbage crap you may as well upgrade.

AMD should just get its caches latency in fucking order, especially the first two.

File: 14nm vs 16nm vs 28nm.png (344KB, 1988x930px) Image search: [Google]

344KB, 1988x930px

>>55706499
>14 nm

>>55706462
>show me a few non AMD-optimised games where it struggles
Every single one he would show you, you'd say it was an AMD-optimised game.

>>55706525
>lost the argument on performance
>moved to power
Yes, those 10W more are a huge issue for me.

>>55706503
It's a horrible card. They lied about everything - the performance, the price, the power, the voltage. Basically everything.

I have 20 year history of AMD/Ati brand loyalty and my first pc was a K6 with a Rage VR. But the bulldozer and the 480 killed it for me completely.

>>55706523
As I said, it's not about the card itself, but about their greed and their false advertising.

File: perf_oc (1).png (29KB, 500x570px) Image search: [Google]

29KB, 500x570px

>>55706308
>it's comparable to the 480 in dx11,

how is it comparable? the 1060 is 10% faster than a 480 at 1080p, and can gain 15% perf from a trivial OC. that puts it in range of a ref 980ti/fury x. meanwhile an OCed 480 can't even beat the stock 1060.

I am hoping Zen is good. Not because I am going to buy it, but because I want it to light a fire up Intel's ass so they actually make improvements to newer generation processors rather than the same old boring smaller nm size and 5% speed increase.

>>55706561
They really didn't lie about anything.
I bet you're that retard shitposter who thinks the Polaris 11 demo total system power was supposed to be Polaris 10 TDP.

>>55706561
What did they lie about?

am I liek the only one who really hates AMD with all their soul?

>>55706560
40-6=34 not 10
39-6=33 not 10
163-116=47 not 10

>>55705453
>4gib
>future proof

>>55706562
>oc'd 480 can't beat stock 1060
Source: muh butt

Also, see
>>55706510
>>55706496
>>55706477

>>55706596
Yeah I hate em too reee REEEEE REEEEEEEEEEEEEEEEEEEEEEE FUUUUUUUUUUUUUUUUUUUUUUUUU REEEEEEEEEEEEEEEE

>>55706564

Unless you are literally an intel engineer you should want Zen to succeed. Everyone, whether they prefer Intel or AMD, are a general consumer or an enterprise, will benefit from Zen being good.

File: AMD-Uprising-5.jpg (75KB, 755x492px) Image search: [Google]

75KB, 755x492px

>>55706570

>They really didn't lie about anything.

if all the marketing doublespeak and insinuation by AMD isn't lying, then what is?

not to mention the fact that AMD employs 1/9th of it's workforce using the H1B visa and has an openly socialist agenda (pic related)

AMD is shady as fuck pushing this bernie-tier agenda, not to mention the fact that some of their investors even have ties to groups like ISIS.

>>55706632
Not me, I only buy the best of the best beause I'm not poor, AMD is better off dead so Intel can have complete control, we'd get Intel-only optimized software and more performance without AMD forcing everyone to use generic codepaths and settings just because they exist.

>>55706537
http://imagescdn.tweaktown.com/content/7/7/7789_200_nvidia-geforce-gtx-1060-review.png
>muh struggling

>>55706579
>>55706570
150 watt card - AIB cards way over 200, nitro+ goes over 300 watt.
6 pin connector they bragged about turned out to be an engineering mistake.
Performance was advertised to be close to 980, no such thing.

And the main reason - $239 MSRP. I calculated the euro VAT added the usual $25 AIB branding tax and set aside 260 euros to buy the card.

Turns out the non-reference cards start at 299 euros. Nvidia manages to sell their cards at the exact MSRP +vat price.


It's the second time, actually. The first time I wasn't careful, trusted their hype and upgraded to an FX-6300, the worst piece of shit I ever owned, my old Phenom was better.

>>55706650
>I only buy the best of the best because I'm not poor

You don't realize that your money will go farther if intel is pressured into making large leaps in performance? You're pretty stupid. Are you even running an 80 thread xeon motherboard? Then you still have a lot to gain from more performance on the enthusiast/mainstream either way.

I forgot to mention idiots also don't want Zen to succeed.

>>55706672
>I'm proud of being poor

All you commoners can do is bark bark and bark
And I'll be here, laughing at you from above.

>>55706650
>AMD forcing everyone to use generic codepaths and settings just because they exist.

wat? it's all x86 my man, 'code paths' are handled at the hardware level

guess im buying kaby lake now, gg

>>55706561
Performance claims: It'll be comparable to a 970, 2x 480s in CF compete with a 1080
Reality: It is comparable to a 970, sometimes also a 980, depending on the game. It can also compete with a 1080 in crossfire.

Power claims: It'll have a tdp of 150W
Reality: It has a tdp of 150W

Voltage claims: There weren't any

Price claims: It'll be priced at $200
Reality: The 4GB model is priced at $200, the 8GB model is priced at $240

You're talking out of your ass; I don't know where you get your "news" from, perhaps nvidia PR reps?

>>55706691
Back the fuck off nigger, you don't know what you're talking about

>>55706646
If you're going to throw around the ISIS boogeyman, you have to implicate every single investor in the world. Anyone who invests in hedge funds for retirement is party to government corruption, drug smuggling, human trafficking, war profiteering, and everything else under the sun. If you're putting money into Wall street, supporting international banks, then you're colluding with them as well. Every single person with a dollar invested is guilty.
Ooooooh scary. Big bad people with their evil money. You faggot.

>>55706669
>Shill retard doesn't know how average power draw works
Every single thing you posted can be immediately disregarded on this fact alone. You're a shitposting child, and you belong in /v/.

Average power draw for the reference RX 480 is 150-160w. Peaks in line delivery do not matter.

>>55706686
>calling me poor
really just seems like projection now. if you are too clueless to see the benefits of intel offering more for less at every single price point from low end to the highest, i can't help you. Keep browsing facebook and playing candy crush on your 6950x or whatever it is you do, that is the only reason i could see this mindset developing.

>>55706720

ISIS supporter detected

>>55706669
>150 watt card - AIB cards way over 200, nitro+ goes over 300 watt.
Overclocked cards have higher tdp; No shit?

>Performance was advertised to be close to 980, no such thing.
Advertised as a 970, and it gets 980 or better performance depending on the game.

>>55706561
>They lied about everything
Show me a statement they said that was factually wrong.
>I have 20 year history of AMD/Ati brand loyalty and my first pc was a K6 with a Rage VR.
Here we go: you were always the world's biggest AMD fanboy, but now you're like sooo disappointed that you're unfortunately going to go with nVidia. Why do shills always repeat the same shit over and over?

>>55706562
>the 1060 is 10% faster than a 480 at 1080p
Wrong, even the Techpowerup summary showed the 1060 was 8% better, but of course you had to round that up.
Also how many DX12 titles did they test again? 2?

>>55706728
ISIS was literally created by the CIA. They're a puppet organization used to push a globalist agenda.
Go back to your Huffington post comment section.

>>55706692
yep

very sad article. i've been waiting just long enough.

provably will go skylake even

this thread is falling apart

>>55706646
I refuse my AMD is so tainted now, what the fuck

>>55706744

so you support globalism, Ahmed?

no wonder AMD gives money to ISIS proxies then. they need more H1B visas.

>>55706694
>Reality: It has a tdp of 150W
You are talking about the reference card. The Devil, the Nitro and the XFX Black have TDP ranging from 200 to 300.

>Voltage claims: There weren't any
They overvolted to hide the ASIC spread of the chips.

>Reality: The 4GB model is priced at $200, the 8GB model is priced at $240

Only in US.
In Europe the reference models cost the same as the GTX 1060 partner cards, non-reference available for preorder at the price of R9 390.

>>55706744
Do you know what the word 'literally' means?

>>55705833
>Intel could have at any time beat them into the ground
But they were to busy blowing up their pipelineand ramping up to 10 gigglehurts without requiring users have their own nuclear generator in their backyard

>>55706732
>Why do shills always repeat the same shit over and over?
You are one of the reasons I'm considering switching to intel. The new underage fanbase is horrible.

>>55706714
Okay, what instructions are coders being 'forced' to use because of AMD being super fucking out of date? Keeping in mind that both intel and amd are way ahead of the curve in most including in SSE where people are still not even moving past SSE1/SSE2 for software not because of AMD but because of the PREVALENCE of OLD hardware such as, for example, the i7-920 and even older chips that still see wide usage.

don't talk to me like im a child, nigger

>>55706790
>The Devil, the Nitro and the XFX Black have TDP ranging from 200 to 300.
And when did AMD ever make claims about cards from other vendors? Do they own XFX or Sapphire?

>They overvolted to hide the ASIC spread of the chips.
Again, where did they lie?

>Only in US.
OH WOW
So when AMD made a claim for a price in US dollars they actually meant it for the US market? Holy moley!

>>55706790
>ASIC spread of the chips.

WHAT?!

>>55706805
To be clear, I'm not throwing around /g/'s latest meme word. I truly mean it: these threads are made by people paid by companies to shit on the opposition. We've had people admit to doing it and posting proof (internal scripts they were supposed to use, messages from superiors regarding what strategies to employ). The most obvious sign are fuckhuge posts with obvious bias and copy-pasta content (always the same photo, text or both in the OP).

If it puts you at ease, this is an obvious shill post, although from the other team: >>55706308

>>55706525
>actually has components on the pcb

>>55706824
Well, the nvidia managed to do all that. Somehow they are a $249 MSRP card, but EVGA cards sell for the same 279 as the reference 480.

>>55706870
I'm not being paid by anyone, I'm just used to AMD as a cheap mid-range alternative.
Then I look at the prices, compared specs and realized that upgrading to Intel/nvidia would actually be way cheaper, because of the retarded promises and price policies.

Now I have a choice of getting a 8370, a new 990fx board, a 480 and a fast DDR3. And I will have to buy a new PSU too. 750 euros alltogether.

Or I can get an i6 6500, a simple motherboard, DDR4 ram and a 1060 to get the same performance and I get to keep my PSU. That's 620 euros alltogether.

Since when AMD stopped being the cheaper choice?

>>55706870
kek, I'm just posting the actual facts of the matter because this board has been overwhelmed by nvidia fanboys and their marketing team and I'm sick of the constant lies being spread. I've only got involved in this at all because I hate shills.

The 480 is a pretty good card, and a better long-term option than the 1060. In a year people will be laughing at 1060 owners like they're laughing at 970 owners today. "haha, you really bought a 1060 expecting dx12/vulkan support!?" The shills make it seem like the 1060 is good competition to the 480, but it isn't - they're only comparable in dx11, and that is going to swing more and more AMD's direction with driver updates. Meanwhile the 480 has advantages literally everywhere else.

nvidia gets what they pay for, and what they pay for isn't better dx12/vulkan performance. It isn't pushing the envelope of game development or graphics. It isn't establishing new technologies in the industry. No, what they are paying for is marketing, and this thread, and almost every thread about AMD since ~2 months ago, have been non-stop anti-AMD shilling.

It wasn't even this bad at the last AMD launch.

>>55706790
>The Devil, the Nitro and the XFX Black have TDP ranging from 200 to 300.
their tdps are 175w

>>55706947
>same performance
except for immensely higher single core fp throughput which is everything for gaming

>>55706947
It's curious how you hop between currencies whenever it's convenient.

>>55706972
200-300 with upped power limit. Which will give enough boost to be on par/a bit faster than 1060. Of course I would want that.

>>55707014
Why in the fuck would you claim that the rx480 has a greater than 200W tdp and then later post that it's only true when it's an upped power limit

This isn't regular shilling, it's incompetent shilling.

File: NVIDIA-GeForce-GTX-1060-Official_PCB[1].jpg (4MB, 3433x2332px) Image search: [Google]

4MB, 3433x2332px

>>55707014
it's your choice to up that power limit anyway. considering the pcb population, polaris is more efficient than the 1060 anyway.

lest we forget the last time nvidia tried hardware solutions (fermi)

>>55706828
He is very poorly describing the ASIC quality rating that CPU-Z will read. The GPU has a little fuse bus that it flashes a number into that indicates roughly how well the die performs at a given voltage level, and a few other things like clocking and leakage.

To ensure AMD had shitloads of volume for the launch of the RX 480 they set their margins for a passing die pretty wide, and set default voltages for the cards accordingly. They didn't "hide" anything. That poster is just a tech illiterate retard fanboy desperate to spread FUD.

File: 1469226727539.jpg (303KB, 800x1077px) Image search: [Google]

303KB, 800x1077px

>he actually believes AMD will deliver

File: AMD-Radeon-RX-480-8GB-GDDR5_PCB_54228[1].jpg (239KB, 1006x600px) Image search: [Google]

239KB, 1006x600px

>>55707026
gee i wonder why one card uses more power than the other

anyway why the fuck has this zen thread devolved into gpu discussion.

>>55707045
I don't understand, just from looking at those photos, what's the matter. Can you explain?

>>55707045
/g/ - Graphics Cards

>>55707025
>everything I don't like is shilling

>>55707026
The only advantage is DX12 support, which nvidia realistically does not have desu and I'm not sure I will ever want to upgrade to botnet, I'm fine on my windows 7/gentoo.

>>55707065
one is incredibly overengineered and sold for $199 the other cheaps out on components as much as possible and sells for $299

>>55707078
nvidia won't be able to support async until they can make an actually power efficient architecture so when they implement ACEs the cards aren't housefires again.

>>55707089
Which is which?

>>55707078
More like, you're incompetent as fuck and can't even do your very simple job right.

How much do you even get paid? It's fucking sad.

>>55707108
i'll let you take a wild guess

>>55707089
Ye, the problem is that async is only realistically present in a fucking botnet. There aren't any games running on vulkan in the foreseeable future.

>>55707123
I genuinely have no idea. The AMD one seems to have more stuff, so I'm guessing it's the overengineered one? But how can you tell the components are cheap?

>>55707125
>only present in botnet
>What is Mantle
>What is Vulkan

do your own API for fuck sakes.

>>55707089
That has nothing to do with power consumption. Having more power phases or an over-engineered VRM does not change how much power a card draws.

Also, AMD is actually kind of stupid for over engineering the VRM like that. Know why? Because it's expensive. They aren't a charity, their goal is to make money, and putting more of these components on the board costs them money. They paired the RX 480 with such a shit cooler anyway, the VRM is basically wasted unless you put a 100 dollar waterblock on that...
a 100 dollar waterblock on a 200 dollar card, huh. Sounds like a plan to me.

File: 1467064173666.png (309KB, 420x420px) Image search: [Google]

309KB, 420x420px

As long as Zen is cheap and a significant upgrade from my Phenom II X4, I'll be a happy man.

>>55707125
parallelism is one of the main features of dx12/vulkan, force it through a single pipeline and it's essentially dx11 again.

>>55707133
the 6phase power delivery on the 480 is better than the 5phase power delivery on the 1080. the 1060 is $299 and the 1080 is $799, just let that settle in.

The current Clocks are low 2.8-3.2ghz

but thats A0 spec the end spec will likely be 3.4-3.5 and maybe some OC room up to 3.8-4.0

Its gonna be Haswell adjacent or slightly better and its going to be made appealing by being agressively priced and having More cores and Hyper-threading by default

So you get a 4core/8thread equivalent to the i7 4790k for the price of an i5 instead or maybe even slightly less.

And they price their 8core behemoth for around the price of an extreme 6-core intel i7
Also the Zen chipset could have more Express lanes than midrange intel as well but dont quote me on that.

basically its more for less. and if they sell i7 4core variants for much less than intel then everyone building gaming PC's will buy AMD instead and ever look back since there are games that perform drastically better on 8-threads and the 4thread i5's have issues.

>>55707141
are you actually criticizing a company for not completely ripping you off? what a world we live in.

>>55707141
So any aftermarket 480 with a better cooler would solve that issue basically?

>>55707158
What I mean is, why pay more for 480 if I won't really be using it?

>>55707160

wishful thinking: the post

AMD can't respin a new stepping at this point if they still want to reach the october launch date, so the clocks will likely stay the same.

>>55705453
>He thinks power draw is all about the electric bill
Please educate yourself on energy conversion.
All that power the card draws goes somewhere. If it's not going to additional performance, guess where it goes.

Spoilers: It becomes heat.

The more performance per watt you can squeeze out, the less heat you'll generate for the same amount of performance.

>>55707174
you're paying $100 less for the 480

>>55707177
>Spoilers: It becomes heat.
depends on the components used and how efficiently they use them.

>>55707170
I'm criticizing their business model. I don't have that card nor do I ever intend to buy it. It has a shitty cooler, so they would already be ripping me off for selling me a card with a shitty inadequate cooler.

There is no merit in mismatching components like AMD has done with the reference board. Yea the VRM is impressive, but what's the point?

>>55707173
Sure, but if you're an AIB building an aftermarket 480 why would you use the reference board? Most AIBs use custom boards with their high end coolers.

>>55707221
>It has a shitty cooler, so they would already be ripping me off for selling me a card with a shitty inadequate cooler.
remember when there was a stigma about people buying ref cards being stupid? what happened to that? why are people suddenly buying ref cards now?

>>55707221
>Sure, but if you're an AIB building an aftermarket 480 why would you use the reference board? Most AIBs use custom boards with their high end coolers.
Then I'm completely lost, what's the problem exactly? AIBs always use their own boards.

>>55707125
>There aren't any games running on vulkan in the foreseeable future.

Valve has been working on Vulkan support for over a year now. You can currently run Dota 2 under Vulkan, although it is buggy.

Dota 2 runs on Source 2. That means that the Source 2 port to Vulkan is almost done. Which means any Source 2 game will run under Vulkan.

Valve has a lot of clout with game developers and does not want DirectX to be the dominant API anymore. It is in the best interest of Valve that ALL video games run on as much hardware and as many operating systems as possible.


Ignoring Valve for a moment, it becomes quite clear from a development standpoint that Vulkan is hugely beneficial to the developer, as they can develop for the consoles, for PC, for Mac, for Linux, EVEN FOR SMARTPHONES all at the same time.

Vulkan is much better positioned than OpenGL ever was to become a significantly used graphics API. You'd be stupid not to see that, and it has nothing to do with AMD, Nvidia, async compute, or anything else retards on g like to argue about.

File: cards.jpg (165KB, 1046x573px) Image search: [Google]

165KB, 1046x573px

>>55707184
NO, I'M NOT. THAT'S WHAT I'M TELLING YOU.
RADEON IS WAY MORE EXPENSIVE

>>55707264
>way more expensive
>Actually same price

Nice, good job.

>>55707282
Are you blind or stupid?

>>55707300
I saw the site and I know this is cherry picked.

>>55705754
>it has never won against Intel or Nvidia
Uh, Athlon 64 days AMD shat all over Intel.
ATi used to be neck and neck with Nvidia as well, until Nvidia got the jump on them with scalar architecture and ATi made the mistake of trying to beat them with more stream processors instead of cooking up a new arch to compete with.

>>55707261
Dota 2 Vulkan runs using a wraper, it is far from being finished.

>>55707235
No idea. I still think it's stupid, the reference coolers are almost universally bad.

>>55707311
Yea I wouldn't call it "shitting all over them" any more than I would call what nvidia is doing to AMD right now that. Intel had competitive performance, but their shit consumed a lot more power to deliver it.

>>55707306
It's not. We are talking about non-reference cards here. See for yourself.

http://www.mindfactory.de/Hardware/Grafikkarten+(VGA)/Radeon+RX+Serie.html/listing_sort/6

http://www.mindfactory.de/Hardware/Grafikkarten+(VGA)/GeForce+GTX+fuer+Gaming/GTX+1060.html/listing_sort/6

>>55707341
blowers in general are loud and only good for multigpu setups

File: no cherries here.jpg (138KB, 783x436px) Image search: [Google]

138KB, 783x436px

>>55707343
You sure showed me.

>>55705453
>demolished the 980Ti

Enjoy your 4GB VRAM

inb4 Nvidia fanboy. I personally use a pure AMD build. It was a completely asinine move to launch a flagship with only 4GB RAM, even if it is very high bandwidth RAM

>>55707170
>are you actually criticizing a company for not completely ripping you off? what a world we live in.
It makes sense if he's being paid to criticize AMD no matter what they do.

>>55707390
>4096mb

Try again.

Your stupidity just helped me decide for intel/nvidia. Thatnk you.

So with all the shilling from both companies... When should i truly upgrade my 4.5 ghrz 2500k. With proper oc i can get it to 5 ghrz contesting a stock i5 6600k now.

Even a gpu like a 980 or aib 480 would give me like 50% real world performance over my 580.

I truly hope that zen turns out well so i can get a octacore zen which would give me 2-3 times better performance overall. I also hope by then the gpus of 300-400 dollar range can give me 3-4 times the performance of the 580.

>>55707336
Since I can find literally no source for what you're saying, I advice all anons to ignore it.

Again, Valve has been working on this for a very long time, and not just for Dota 2. It's Source 2 that's being ported, not Dota 2.

>>55707410
What is problem with 4gigs?

None, I still run 3gigs card, you didn't decide anything, you already had intel/nvidia and you are trying to cherry pick prices when clearly they are exactly the fucking same.

>>55707416
I upgraded now, but I'll probably upgrade again next spring. It'll be a great time for upgrades. Zen only has to be decent to shake up processor pricing, and AMD/Nvidia will finally have HBM cards on the market.

>>55707429
Some games don't even run maxed out on 1080p with 4GB. Both the 480 and 1060 have the performance to run the kind of gfx settings that take over 4GB.

File: specccc.jpg (33KB, 552x224px) Image search: [Google]

33KB, 552x224px

>>55707429
>What is problem with 4gigs?
Doom and GTA V already eat up to 6 gigs of vram. What's the point of an expensive card if you can't play fullhd on ultra settings?

> you already had intel/nvidia

On my old PC, because I dissasembled my newer one to clean the dust, but here you go. I have another one on athlon x3.

>>55707482
Post a webm using your phone to record you launching speccy and include a timestamped paper in the video.

>>55707482
I run Doom on Ultra @ 1200p, 45-60fps dops down to about 30 in some heavy shadered areas with loads of glass.

I have no idea where you get idea that Doom eats 6Vram.

>>55707507
DOOM literally requires around 5GB for nightmare mode. Won't even let you enable it without that.

File: catalyst.png (211KB, 1528x736px) Image search: [Google]

211KB, 1528x736px

>>55707495
This should be enough for you.

>>55707507
I've seen it in benchmark videos. Check them out on youtube, the vram gets almost full.

>>55707525
yea but he said ultra settings.

>>55707148
>still using his Phenom II
jesus man i caved in a year ago because i just couldn't anymore, how do you do it?

had a 965 @ 3.8

mostly played cpu intensive (strategy, mmo, etc) poorly optimized games that run on 1-2 threads though

>>55707581
I only play CSGO, TF2 and GTAV these days do it does the job.

There's no current AMD chips worth updating to and Jewtel prices are absurd in Australia so I decided to just wait for Zen. Hope it was worth it.

>>55707679
Buy your chips from ebay. China and Singapore are right next door.

>>55707450

I also wonder when will the prices for 1440 monitor drop to an acceptable level in europe.

>>55707688
That's not how these things work. Ebay prices are the same as any other retailer.

>>55705453
>falling for the future proofing meme
The $200-$250 cards of this generation are equal to the $550 cards of 18 months ago, and this generation's mid range cards have minor feature improvements that you couldn't get on last gen high end hardware at all. You're much better off being on a mid range, more frequent upgrade cycle than you are spending tons of money on something that's going to be outclassed by budget parts in under two years.

>>55707733
>You're much better off being on a mid range, more frequent upgrade cycle than you are spending tons of money on something that's going to be outclassed by budget parts in under two years.

only applies to high-end nvidia, you can still rock high end cards from 2012 onwards from amd for dx12 because they actually support their cards unlike anything before maxwell from nvidia that is from even a year later like the gtx 780.

same pattern as always, gtx 260 was a dumb ass buy because within a couple of years dx11 blew its ass away but people who bought hd5xxx cards at the same month were set.

>>55707730
You're buying the wrong thing then. I can get a Xeon chip that shits all over your Phenom 2 for like 60 bucks while consuming less power. Some of them are bad because they support nasty sockets, but yea.

File: IMG_20160713_195910.jpg (190KB, 600x649px) Image search: [Google]

190KB, 600x649px

I want to build a pure Amada system!

>>55707760
I don't want an old Xeon with outdated instructions, features, platform, RAM and no upgrade path sonny boy. Not to mention the motherboards for those old "bargain" Xeons are wicked expensive.

>>55707679
>Jewtel prices are absurd in Australia
it's pretty bad here too, well over $200 for dustbin entry level non-hyperthreaded last gen quad cores like the 4690k

>>55707759
No, it applies to AMD just as much. The 480 is about equal to an r9 390 for half the price at launch. Just because you can still use your old card doesn't mean buying high end cards for "future proofing" isn't retarded from a value perspective.

>>55707775
Who says you need an old Xeon. There are ones for 2011-3. Although those are gonna be Haswell ES processors.

>>55707769
>gpu's suck at making draw calls or offloading any kind of work from the cpu so you need intel to shoulder the burden to make them worthwhile

>cpu's run hot, slow, and are only good at multithreaded integer workloads even though FPU performance is almost everything these days

you think you want it, but you don't.

>>55707802
Oh yeah, 2011 chips. What a bargain. I'll just drop my plans to get a cheap Zen quad-core and go buy an Intel™ Extreme Edition™ i7-6800K™. Motherboards starting at only $499!

File: GhfTuGn.jpg (159KB, 1280x720px) Image search: [Google]

159KB, 1280x720px

>>55707820
>I can predict the future using dank memays I hear from /g/

>>55707794
the difference is you can get the 390 far far earlier and enjoy it then AND in the future. gtx 780 is still a power house in terms of raw horsepower and would do fine if it had the support from nvidia it needs for current/future dx12 titles. But it does not. On the other hand, if you bought an hd 7970 a year earlier you're still fine.

That's why it only applies to nvidia. AMD high-end cards get much closer to their actual end-of-life in terms of performance before they are no longer able to run games without a launch error.

This isn't about comparing value between newer cards and older cards, it's about being able to go way longer without needing to upgrade. That is to say, getting your money's worth out of that high-end gpu. You could have bought a mid-range nvidia card 2-3 gens in a row, or one high-end AMD card several gens ago and be just as well off compatibility wise.

>>55707850

history repeats famerino, i want Zen to succeed but im also being realistic

>>55707840
I got a mobo for mine for 230 EUR. They aren't that expensive. About 2x the price of a mid range LGA 1151 board.

>>55707873
Cool. And Zen will probably be cheap as fuck with <$100 motherboards. Fuck Intel.

>>55707855
>This isn't about comparing value between newer cards and older cards
Yes it is you retard. I'm showing you that spending $200-$250 every year gets you a better VALUE than spending $500 every two years. You will still have a very capable card every generation. Your autistic fear of upgrading is costing you money.

>>55705232
AMD delivering 2012 performance in 2016

>>55707872
If history repeats and Jim Keller btfo Intel befre then he can do it again

>>55707889
<100 motherboards are shit. Doesn't matter what processor you get them for.

>>55707917
Weird, my $60 board has been going strong for the past 6 years.

>>55707045
Less components=less components that can fail.
In science we say the simpliest (too late her so forgive my spelling issues) way is most likely the right one.
The universe stifes for the highest simplicity.
An approach that is overly complex is most likely the wrong one or will fail at some point.

>>55707893
Why do you need to even buy every 2 years if you bought a high-end AMD card? hd 5870 came out in 2009 what is that... 7 years ago? Still on par with a 750 ti in terms of compatibility and performance. 7 years of still being able to launch games that are being released.

hd 7970, my other example, 4 years old now and still going to be viable for several more if you aren't a settings whore. What you're saying to me is still only applicable to nvidia cards, or for people who simply can't stand less graphical fidelity as time goes on.

>>55707581
Phenom is doing fine, sitting on a 945 now with low-end 6 year old video card.

It runs all the paradox games and witcher 2 at high settings.

>>55707141
Those are active components, not passive ones, if I remember correctly, thus they affect power draw.

>>55707341
>Yea I wouldn't call it "shitting all over them"

Athlon64 at 2.6GHz outperformed every Intel CPU on the market unless you took the extreme edition and OC'd it to 5GHz.

>>55707939
my parents use an ancient Asus P5GZ-MX that has been on most of how long they've had it. Motherboards are mostly priced on features/expansion slots.

>Su confirmed no desktop Zen until Q1 17 in latest earnings call

JDIMSA

>>55707983
Motherboards are mostly priced for the chipset and the VRM.
Try overvolting on a cheap motherboard and you will fry the whole thing, including the CPU.

>>55707939
Yea and so has my 80 dollar board that I bought in 2009. Doesn't change the fact that it's shit. Only 6 SATA2 ports, a 5+1 phase VRM, the socket is placed unfortunately close to the top PCI-E slot (mATX board), shitty sound chip. Basically no good features at all.

File: 42bcee5009c7a523f2c07972e178b1f6.png (186KB, 1014x887px) Image search: [Google]

186KB, 1014x887px

>>55707990
if ANY zen hits the market in q4 2016 we'll at least know what were in for because they'll be benchmarked to hell and the desktop variant will be on par aside from having slightly higher performance all-around core-for-core

>>55707944
>Why do you need to even buy every 2 years if you bought a high-end AMD card?
You don't, you can change my example to any time frame using the same ratios. My point is that "future proofing" is retarded from a value perspective because technology moves on so fast. I'm genuinely confused as to why this is so hard to understand. Why buy a high end AMD card when waiting a year will get you better performance for less than half the price? Why? I don't understand why you'd want to be able to run games on high for a year, then run them on medium to low for five years after that when you can buy an x80 card from AMD every generation or two while spending less money overall and consistently being able to run games at medium-high. You should never buy a card with future proofing in mind. You're also forgetting that who makes the better, more future proof cards changes from generation to generation. Nvidia used to make power hungry, housefire cards. Now the tables have turned, and the tables can turn on anything in the future. If you're buying a card because you believe the tables won't turn, you have no right to say you aren't a fanboy.

>>55708014
>>55708002

>having your VRM's on the motherboard
>buying AMD after 2014
cant make this shit up

>chipset doesn't support l33t noscope overclocking oh noes
damn this matters for lifespan im sure

>>55708014
>Less features
Yes this is the correct reason why they are cheaper and why you buy a motherboard based on your needs not based on price alone. Some users need more sata, pci, fan headers, DIMMs, whatever, and that's what they pay for, that's what they get, an overall beefier system.

>>55708075

>chipset doesn't support l33t noscope overclocking oh noes
Overclocking is actually the only way to bring the AMD to a modern performance standard. If you use an AMD you MUST clock it to 4.7 ghz.

>>55708057
>You don't, you can change my example to any time frame using the same ratios.
Not for Nvidia. Remember the aforementioned Tesla cards? You could have bought a 295 for $750 when it came out you still would have received a steaming log of shit a few years later thanks to dx11.

>>55705293
Fuck off

>>55707990
That didn't happen, and thats not what was said.
Initial release is end of this year, full volume availability is q1 2017.
They have been saying the exact same thing for over a year now.

Stop spreading FUD.

File: c3w_1920vh.png (16KB, 582x368px) Image search: [Google]

16KB, 582x368px

>>55707391
>pure AMD build
You just love making horrible decisions, don't you.

>>55708095
>Not for Nvidia
Holy shit first you claim you aren't a fanboy, then you pull this shit. Nobody can predict the future. Sometimes AMD loses a generation, sometimes Nvidia does. Who knows, the 11x0 series might be more future proof than AMD's next cards, this shit happens all the time in the tech industry. But fine, if you want to throw your money away buying products that might be subpar because of brand loyalty, go ahead. It's your money, not mine.

File: let-down.png (609KB, 854x640px) Image search: [Google]

609KB, 854x640px

G-g-guys this is going to be good right?

P-p-p-please

>>55708057
>If you're buying a card because you believe the tables won't turn, you have no right to say you aren't a fanboy.

It's funny because I've always bought nvidia still use one, I just take notice of how, in the past, my cards have refused to work with newer technologies that older AMD cards have no issue with, and it bothers me. i DO upgrade regularly because i literally have to. That's the nature of the planned obsolescence business practice nvidia touts.

>>55708133
but for the past 10 years nvidia have consistently been shitty about support for new techniques on old silicon. History. Repeats.

>>55708153
And now you're completely disregarding the main point of my argument, I even used only AMD examples to show that buying a card for "future proofing" is a terrible value. Fucking respond to that point.

>>55708120
The FX probably wasn't overclocked desu. They don't work, unless overclocked and tuned. They should've made a chart with 9590 at 6ghz.

>the 480 is bad
>amd cpus wont be good

well memed

>>55708181
Are you trying to kill us all.

>>55706040
>shared L3$ per module
>not shared between modules
I smell another disappointment coming up. AMD has had shitty L3$ indexing and latencies since 2006.

Funny how all the RX480 people are busy playing games or out getting wasted/high/laid with all of that money they did not throw away. while Intel and Nvidia seem to have an endless supply of virgin tears over here for us AMD fans to read up on tomorrow

Boohoo... AMD is not fair... Boohoo... I dont even have 200 bucks.... Sustain us with those sweet sweet endless supply of Nvidiaintel tears

>>55708239
>pulling things out of your ass

The L3 is not tied to one CCX

>>55708172
because you don't spend less money buying a x80 every generation that the high-end would have been fine in. 7970 had an MSRP of under $500, these x80 cards are consistently around ~$200. That's 2 and a half generations you skip, and by then you're still playing things on medium-high. All gens after that are just freebies assuming your card doesn't kick the bucket. And you'll be able to get away with at least Low settings for many generations longer than a mid-range card from the same time because of that premium you paid back in the day.

This is where AMD keeping support for older cards is such an advantage if you're buying something high-end and actually want your money's worth rather than just flushing it down the toilet to a company that wants you to buy the next shiny thing asap.

>>55708236
There is a 10 page guide to overclocking an tuning an FX and the NB, after which it performs at i5 level.

>>55705232

>wccftech

GTFO!

>>55708075
I bought that board in 2009 based on my needs. Guess what happened. 6 years passed and my needs had changed.

File: maxresdefault.jpg (40KB, 1280x720px) Image search: [Google]

40KB, 1280x720px

WHEN'S VEGA

>>55708255
Look at the diagram closely. It is. Cores in one module will not be able to access the L3 cache in the other module and vice versa. They can only access the cache that is local to their module.

And this is coming from a company that hasn't been able to improve their L3 cache performance for over a decade.

>>55708292
That "diagram" is something made by Dresdenboy based off of data in a linux kernel patch, you tech illiterate retard.
The L3 is not tied to one CCX. There is an enormous slab of logic in the center of each L3 unit, and they are connected to one another.

>>55706308
>it can't SLI, whereas the 480 gets significant benefits from CF in vulkan/dx12.

Wrong, 480 is not using CF on vulkan dx12 but the standard way (multi-adapter) and 1060 can't sli but can use multi-adapter too so you can use two 1060s on dx12 vulkan if the game supports it (i've seen ashes using two 1060 and it works great)

>>55708312
>There is an enormous slab of logic in the center of each L3 unit, and they are connected to one another.
Proof? Because even the shilliest of shills over at SemiAccurate were able to determine that the L3$ were distinct and not inter-accessible.

>>55708266
>7970
I used the r9 390 in my example. The last few generations for AMD have been weird for the purposes of this argument because of all their rebranding.

>All gens after that are just freebies
No, they aren't. You paid for them up front.

>if you're buying something high-end and actually want your money's worth
Why buy something high end when it is going to be outclassed by the next generations mid range for much cheaper? The price/performance of high end cards is always worse and you need to keep them longer to get your "money's worth" compared to mid range cards.

>>55708315
To be fair, any game should 'run great' with $500+ of gpu power behind it, at this point it's not about which competitor runs it great, it's which one runs it better, because it's all impractical penis measuring nothingness rather than actually playing the game.

>>55707733

GTX580 went for 550 euros. The only card that finally matched it's performance was the R9 270X which sold for around 250 euros here and only the non reference models clearly beat the GTX580 (around 5-10%). The 270X was released 3 years later. And someone with an 270X can enjoy decent gaming even today.

So a 6 year old graphics card still holds up today for decent gaming.

Wasting 550 euros for 6 and perhaps 7 years wasn't bad at all.

The added bonus was that for a few years my card played everything maxed out

File: anime pic.jpg (71KB, 400x465px) Image search: [Google]

71KB, 400x465px

>Thread about Zen
>90% GPU shitposting
>Only 1 mention of Jim Keller
I'm disappointed, /g/.

>>55708323
>Why buy something high end when it is going to be outclassed by the next generations mid range for much cheaper?

Because even after your card has run through its high-end cycle and you've pooped yourself out on all the games it can max out, you've still got a mid-range card equivalent without spending a dime. Typically yes i would say keeping it for longer is how you get your money's worth, and for AMD at least it's a valid decision to make when you buy it, unlike Nvidia, but there's also the fact that it keeps its value afterwards more so you can re-sell it at a much better price, Nvidia cards age like milk left out in the sun.

So yes, being able to resell your card to someone who wants compatible mid-range current gen performance for a bit less because it didn't immediately lose all of its value when the latest api or whatever got announced is another benefit to buying AMD.

>No, they aren't. You paid for them up front.

Freebies as opposed to upgrading as regularly as you're suggesting, I mean.

File: 18.jpg (544KB, 1600x911px) Image search: [Google]

544KB, 1600x911px

>>55708320
The structure in the center of the L3 shown on Summit Ridge here: >>55705698
Is similar to the L2 unit employed in the Jaguar quad core compute unit.

Zen's core complex is incredibly similar to the Jaguar CU, except it is the L3 that is shared. Look at the Jaguar CUs used in the PS4 and Xbone APUs. Each of them as fabric connecting them together making the shared L2 coherent across all 8 cores.

>being such a tech illiterate retard that you think you can magically see BEOL looking at the top of an IC

Outstanding.

File: serveimage.jpg (660KB, 1600x1200px) Image search: [Google]

660KB, 1600x1200px

Does zen still have pins?

>>55708387
God i hope not. The one advantage it has over LGA namely that it's easier to fix bent pins on the cpu than on the mobo is NOT worth all the other bullshit that comes with it

>>55708267
Nice, but electricity is not free in my country.
We pay roughly 0.30€ per kWh and the next raise is just around the corner.
And I do not really want a CPU that draws roughly 0.5kW by itself.

>>55708387
>>55708402
Socket AM4 is PGA

>>55708412
Just don't leave prime95 on permanently and you should be fine, friend.

>>55708387
>>55708402
Imagine if it was BGA and you had to stay on a fixed mobo and CPU setup.
Worse the heatsink and fan are not replacable.

>>55708418
>AMD server sockets are LGA
>consumer sockets still PGA
As always, the non-enterprise pleb gets shafted.

>>55708357
Jim Keller left the aborted fetus in a dumpster along the Ganges River. Then some poo-in-loo picked it up and started sprinkling it poo-dust to bring it "back to life".
Let's get this clear
>dual channel only memory, where one channel is dedicated PER MODULE (nigga why)
>HyperTransport 3.0 can't even support the same bandwidth as DMI2.0, let alone DMI3.0
>NO NATIVE USB3.1 CONTROLLERS IN THE YEAR 2016
>expecting SoCs to fully compete with CPU+distinct southbridge
AM4 is barely competitive with LGA1150 Lynx Point chipset, let alone an LGA2011-3 and X99 chipset.

>>55708386
No direct proof though. There's absolutely zero sign of modules being able to share their L3$ into one massive pool and all signs point to each bank of L3$ being tied to the cores directly around it. Just like the Core 2 Quads.

>>55708451
>not buying an opteron so you can enjoy the bliss of opening a single instance of dota 2 at 50 fps and then being able to open 15 more with the exact same fps

what are you some kind of enthusiast intel pleb?

>>55708469
Yeah, sure. AMD is going to sell 32 core 4 die MCMs with ridiculously high bandwidth GMI links but none of the CCXs can read each others L3. Those links are totally superficial, just for show. That huge cache interface? Its there for placebo.
Sure.

Tech. Illiterate. Retard.

>>55708326
What i was trying to say is that new apis has a multi-gpu standard with better scaling and less bugs.

Both sli and crossfire are prone to have bugs, latency problems, bad scaling or it's just not working at all on some games.
Both vendors needs to adjust and create profiles for every game so it can "work" so probably they'll be happy to remove support for this as soon as more titles start using dx12/vulkan.

>>55708495
No, I have 24 threads ;)

>>55708502
>Those links are totally superficial, just for show.
DING DING DING
You hit the nail on the head. Just like the Core 2 Quad before it, the total cache size IS FOR MARKETING PURPOSES.

>>55708451

LGA is far worse than PGA in pretty much every way unless you need 2000 pins like socket g34 does.

>>55708383
>you've still got a mid-range card equivalent without spending a dime.
No, you have to pay for that privilege up front.

>Freebies as opposed to upgrading as regularly as you're suggesting, I mean.
But mid range cards cost less up front, I literally do not understand your point at all.

I don't get you at all. You're clearly an AMD fanboy, and I'm saying the rx 480 is a good buy, and you're arguing against me.

>>55708526
How's it worse? PGA processors are annoying as fuck to handle. Every second I feel like I'm gonna bend a pin or something.

File: memory.png (25KB, 528x307px) Image search: [Google]

25KB, 528x307px

>>55708469
>>dual channel only memory, where one channel is dedicated PER MODULE
this can't be real
tell me this isn't real

even an entry level garbage can cpu like pic related gives each core full reign over the memory controller

now you're telling me amd is going to be cutting their bandwidth in H A L F and sharing it between F O U R C O R E S

it's finished, this is going to be a worse architecture than bulldozer.

>>55708426
You know that there are people that actually do shit aside from gaymen on their PC.

>>55708551

because intel's LGA design is meant to be as cheap as possible and often breaks pins just through normal cpu insertion.

>>55708526
LGA has advantages in electrical signaling that only a select few nerds in the world care enough about to study. It isn't a big deal, and neither one of them is really better or worse than the other.

>>55708551
Bending a pin does nothing unless you break contact with the package. You can always bend it back.

>>55708555
Stop listening to a tech illiterate shill.
The Memory PHY are located around either CCX, but the two CCXs are fully coherent.

>>55708591
It isn't intel's design. Both AMD and intel buy them from a 3rd party.

>>55708528
im only arguing against high-end nvidia in favour of high-end amd.

if you want me to pick between mid-range nvidia and mid-range amd it's less of an issue because those cards (usually) run out of steam for newer games before the technologies they support go out of style. I'd personally choose the rx480 just for all the displayports.

>>55708591
I've abused those sockets a bit and the pins definitely don't break easily. Inserting the processor and locking it down is so easy, at least it was for LGA 2011-3.

>>55708595
I'm a flipper handed freak. I'll not only break the pin, I'll snap the processor in half while doing it.

>>55708595
>but the two CCXs are fully coherent
Jesus fucking Christ, do you even believe your own drivel? How much is Pajeet paying you?

>>55708599
>im only arguing against high-end nvidia in favour of high-end amd.
Me and you are fundamentally arguing against different things then. Never did I ever support Nvidia here, I tried to be as neutral as possible.

>>55708613
Pins depends on the manufacturer really

>>55708613
Well if you're going to knock a cpu around like a football i would definitely say from experience Intel has far more durable offerings. I used to flip around an old core 2 duo in my hand almost habitually and after several weeks of that it still worked.

>>55708625
>tech illiterate shill just desperate to spread FUD

How does it feel to spend your Friday night on the internet lying about a company because you have an autistic grudge?

>>55708555

the modular design itself is super retarded as it is. what was the benefit to grouping 4 INDEPENDENT cores into modules?

>>55708647
>grudge
kek, you should really start noticing those blatant red flags during Zen's development. If you really were tech literate, you should have at least noticed two regarding the interconnect between both Zen modules, one from the Summit Ridge specification leaks last summer and one for the speculative HPC Zen back when AMDrones were shilling "OMG 32core APU WITH HBM!!!1!".
Here's a hint: LAZY ENGINEERING WORK as expected of an all-star team of underpaid poo-in-loos

>>55708628
i think we just had different ideas on future proof, for me personally, if i already got several years of performance out of a card i dont care if i have to go Low settings or even a lower resolution, the medium-high meme dream was never something i chased, I just have to upgrade from time-to-time with nvidia, but if i just dropped one lump sum into a nice AMD card i would have been able to leave it in there for so much longer each time. DirectX errors are fucking annoying.

>>55708684
Cheap-ass scaling.

>>55708687
>even more tech illiterate FUD nonsense
I get it. You're an mentally defective autist who does nothing but shitpost with childish fanbodyism. You don't need to keep pushing yourself.

>>55708684
Same reason Intel dropped 2 Pentium 4 cores into a single module and cranked out the Pentium D while AMD were flushing out their assholes with a hot sauce filled hose.

Desperation.

>>55708697
and this

>>55708708
Stay mad. Zen will underwhelm and disappoint because AMD cheaped out. Screencap my posts if you want, it will just make the paper launch all the more sweet on my end.

File: download (2).jpg (4KB, 200x144px) Image search: [Google]

4KB, 200x144px

>>55705232
>POOING IN THE LOO

>NOT POOING IN THE STREETS LIKE A GOOD PAJEET

>>55708728
He doesn't even work with AMD CPUs, reeturd

>>55708728
Rajeet didnt make Zen, Zen's meme man is certified shitwrecker Jim Keller

>>55708728
i mean it doesnt really impact my opinion of AMD as a company because i don't think any of it matters beyond pure meme value but i wonder if this guy, when he wasn't an AMD engineer making six or more figures, has ever actually just popped a squat on a designated street and let loose a long brown steamy snake of curry shit into the mile-long pile with everyone elses.

>>55708733
Autism detected.

>>55708684
Thermal management, orienting all IP blocks for scaling in an MCM, simplified binning with defective dies. There are dozens of reasons for it.
Theres an entire area of IC design for this specifically.

>>55708739

zen wasn't designed by keller, the zen designed started before he joined and was led by suzanne plummer.

keller mostly worked on their new ARM and x86 embedded stuff.

>>55708774
Keller was the department head who oversaw both Zen and K12. As a department head he likely didn't work hands on with anything.

>>55708724
>>55708708
Also, I can't wait until you start DEFENDING AMD's lazy and Jewish tactics of offering less for equal amount of shekels.
>but you don't need 8MB of L3$ for all cores
>4MB is large enough, why do you need more?
>PCIe Gen 2 is still relevant in today's market
>so what if ASMedia USB3.1 controllers suck shit through a cow's ass with massive latency issues?
>you don't need TRUE dual channel memory, one per 4 cores is enough
>just buy higher speed RAM, idiot

>>55708774
Keller gave AMD alien scrolls he found from Area 51 back in 2006. He promised every 10 years he would help them design an Intel killer cpu by deciphiring the scrolls.

File: john madden.jpg (60KB, 700x700px) Image search: [Google]

60KB, 700x700px

INTEL ARE ALL JEWS WITH THEIR OVERPRICED ENTHUSIAST BROADWELL-E GARBAGE AMD IS ALL ABOUT THE VALUE ITS THE VALUE

>https://web.archive.org/web/20130719043005/http://www.newegg.com/Product/Product.aspx?Item=N82E16819113347

>>55708757
I really hope he has anon. Respect.

>>55708788
>even more tech illiterate childish shitposting

Its 8MB per CCX actually, Samsung SRAM cell size is a known quantity, child.
AM4 supports PCI-E 3.0
The integrated FCH supports USB 3.1
The CCXs are coherent, its dual channel

Its hilarious how you behave exactly like the childish Nintendo fanboys on /v/ arguing about why everyone else is trash.

>>55708788
>PCIe Gen 2 is still relevant in today's market

Realistically it is, there aren't many things you can plug into those things that would fully saturate it.

I agree though, less for more is bullshit.

>>55708843

>The CCXs are coherent, its dual channel

[citation needed]

each memory controller is physically linked to the module, to make dual channel quad cores AMD would need to disable half the cores and cache in each module instead of packing one module and two memory controllers on the die.

>>55708864
>tech illiterate retard yet again thinks he can tell how the BEOL is connected based on how the FEOL is arranged

Hilarious. Adorable really. You're the cutest little autistic shitposter I've seen in a long time. Almost as cute as Tony Dickey before his lawsuit got laughed out of court.

>>55708843
>The integrated FCH supports USB 3.1
THIS IS LITERALLY FALSE
>Board designers are reportedly having to use additional retimer and redriver chips to get acceptable bandwidths over such ports, and in some cases even entire USB 3.1 controllers, eating into the platform's PCIe budget and escalating costs.
THEIR INTEGRATED CONTROLLERS ARE AN UNMITIGATED FAILURE AND ONE OF THE REASONS FOR THE AM4 LAUNCH DELAYS. THEIR ONLY SOLUTION IS TO DISABLE THE CONTROLLER IN THE IFCH AND USE THE EXACT SAME CONTROLLER SOLDERED ONTO THE BOARD.

>>55708887
Posting to the wrong anon, sperglord.

>>55708887

>personal attacks with no source cited

you're projecting kiddo

File: giphy.gif (1000KB, 500x381px) Image search: [Google]

1000KB, 500x381px

>>55708908
>kiddo

>>55708469
It has native usb 3.1. It just has signal loss issues requiring a 5$ booster chip. It`s intel that lacks native 3.1 in YORL 2016.

>>55708555
There`s zero confirmation, and the only thing people are basing it on is the leaks of 16 core zen with quad channel, and 32 core zen with 8 channel. The scaling of this, and AMD`s modular design preference paired with the SOC construction leads people to jump the gun yelling that each module can only handle a single channel of memory.

HOWEVER, Kabini is incredibly similar to everything we currently know / assume about zen. And that legitimately is single channel for 4 cores. While it seems unlikely that they would do something as ridiculous as having an independent single channel memory controller on each module. Don`t put it passed AMD to do something fucking silly like this.


What would seem more likely is that literally everything they`re doing with Zen is based on improvements in the opteron tech combined with all their effort put into SOCs and cat cores. But I`m just pulling shit out my ass, and nobody really knows. So there.

>>55708967
>It`s intel that lacks native 3.1 in YORL 2016
And that's being directly addressed with the 200-series chipset coming out at Q4 this year. At least Intel didn't waste precious development time, budget, and die space by putting a fucking useless controller onto their chipset and ordering their partners to buy the exact same fucking controllers in discreet form from ASMedia again in order to have functionality.

>>55708967

seems more like zen is primarily meant for the server market and the HEDT/mainstream/mobile market is basically an afterthought.

>>55708967
>>55708907
I had not seen this before posting. Got a source on board manufacturers having to install complete usb 3.1 controllers? I had only seen articles on them needing signal boosters, and had even just did a quick google search. Maybe I`ll look harder.

>>55708907
>shit posting tech illiterate child going full blown autismal rage mode because he can't handle reality
>yet again exactly like the Nintendo toddlers on /v/

Its adorable that you take WCCFtech rumors as gospel when they're convenient. As for AM4 delays? What delays. There are none to speak of. Early on this year ASUS and MSI reps stated that they hadn't even begun manufacturing AM4 boards yet. The only *rumor* about AM4 board launching early came from fanboys who thought desktop Bristol Ridge would be available this summer.

There was no delay. There is no issue with their integrated chipsets. The supplemental Promontory chipset is literally nothing like their integrated I/O.
You're not even capable of making sense at this point.

Go calm down and play some Pokemon GO, autist.

The CCXs are coherent. The cache is shared between them. Its a dual channel IMC, location of the PHY literally mean nothing. PHY connect to the IMC before they connect to the cores.
Your delusional shitposting is the funniest thing on this board I've seen in a long time. Its like Lisa Su personally raped you as a child, and you can't rest until your vendetta is fulfilled. Really dramatic stuff.

>>55708998
If an architecture is server worthy in terms of performance wouldn't it also be PC-worthy by proxy with just a few features added/removed? Unless the cores are complete shit again.

File: top kek.jpg (412KB, 950x604px) Image search: [Google]

412KB, 950x604px

>>55709009
>As for AM4 delays? What delays. There are none to speak of.
>What delays. There are none to speak of.
>What delays.
I'm fucking done with this shill
Holy shit, this delusion