1st AMD Zen/AM4 Motherboards Coming In October?

>>56331282
>>>/v/

>>56331436
Processors aren't the same thing as gaming videocards you dolt

Learn your maymays or don't fuck around

File: 13938537_10209983596448848_4748243332306719645_n.jpg (11KB, 200x217px) Image search: [Google]

11KB, 200x217px

>>56331503
both are shit either way

>>56331065
what about the cpus do they come out at the same time as the mobos?

>>56331503
>>56331539
>>>/v/

>>56331539
Don't be a fanboy

You've not even fucking seen the processor and you're shitting on it

http://seekingalpha.com/article/4002864-amd-zen-likely-good-enough

>>56331065
But the chips come out 2017?

File: AMD-Zen-ES-AotS-Benchmarks.jpg (704KB, 1256x750px) Image search: [Google]

704KB, 1256x750px

>>56331553
im sure it brings "competition" to the table

>>56331613
>WCCFTECH game benchmarks
Kys

>>56331541
>>56331588
Bristol Ridge is on socket AM4 as well.
Some OEM systems have been on sale for a while now.

Chips will go on sale for the builder market before too long.


>>56331625
They're not even from WCCF, they're pulled from the AoTS online bench result catalog.
Someone stole a Summit Ridge demo system from AMD's Computex event in Taipei and they were playing around with it. Someone searching through the bench results saw the new processor ID and it was determined to be a Summit Ridge engineering sample.

All the curry crew here did was compile the results and make their own graph.

>>56331613
Dis they stated a cost for that Zen processor?
If it's around the price of the 4670K i see no problem

>>56331588
>>56331659
Basically 14nm versions of current gen APUs + DDR4

Hope they don't waste much silicon on them

>>56331689
Bristol Ridge is not a 14nm part. Its 28nm HPP exactly like Carrizo.

>>56331700
They shouldn't be wasting any money on it then

>>56331724
Its just a respin of Carrizo in a different package.
They're selling units until they hit their projected return.

Not Zen, at least not now. Zen reaches the market in 2017 the chips coming this year are APUs based on excavator or something

>>56331659
>They're not even from WCCF, they're pulled from the AoTS online bench result catalog.
From an alpha version engineering sample running at 2.8ghz.

The real Zen runs at 4.5ghz.
Intel is on suicide watch.

>>56331915
t. AmdIDF

>>56331966
A more recent engineering sample beat 6950X in cinebench.

And it's still an engineering sample. The real thing has 10% more performance than Broadwell-E. At the price of Skylake.

Remember when competition was like, "Oh hey this is the Athlon 64 it is objectively better fuck you" and then "This is the Core Duo. Eat shit, friend"

Now it's just Intel being better across the board in every aspect except performance-per-dollar in certain areas, but it's mostly Intel has offerings and AMD simply isn't competing in all except the budget and budget-enthusiast space.

I feel like Zen won't change it. If AMD really had a "Get outta town, Intel, this is our territory again" chip, they wouldn't be keeping it as quiet as they are. I'm worried this is going to be another 480, just a "Well I guess it beats out the competition specifically at that price point but I have $500 to spend I want something better", AMD just can't compete.

It's actually pretty upsetting.

>>56332134
> "Well I guess it beats out the competition specifically at that price point but I have $500 to spend I want something better"

Thats exactly what Summit Ridge is, except the competition from intel is $400-$1000.
Its not a competitor to mainstream i5s and i7s. Its competing with the i7E line for content creation workloads.

will it support ITX though
no point if i cant put it inside a sexy Ncase

>>56332339
Whether or not manufacturers offer mITX boards is entirely up to them, not AMD.

>>56332365
Not really, AMD's fault that AM3 was a fucking housefire and thus couldn't support mITX.

>>56332365
AM4 is a unified socket though so if manufactures make AM4 boards they will support everything from the lower end APUs to the high end parts. Only thing not using AM4 is the BGA packaged parts

File: 1470700051264.jpg (87KB, 468x248px) Image search: [Google]

87KB, 468x248px

Im putting off a CPU upgrade in hope for AMD to unleash their shitwrecker powers.

Sure im "falling for" the AMD waiting meme but the Intel waiting meme is having to wait 5r fucking years for a processor thats more than 10% faster than your current one.

>>56332397
Nope, not even remotely right.
Mobo manufacturers could put enough VRM on a board to supply a stable 300w CPU. Tons of enterprise hardware pulls power like this without issue. Its only a matter of how much money they decide to put into their power phases.
Manufacturers skimped on AMD board, meanwhile some entry level intel boards had better VRM that high end AM3+ boards.

>>56332408
There could likely be a tremendous amount of segmentation among AM4 boards. Its not about what AMD does, its about where the manufacturers feel they can price something to move the most units. If AMD systems aren't going to sell as well they aren't going to build them with nicer components.

>>56332339
sauce

Why did Jim leave?

>>56332590
He seems to do that. He did it after designing K8 and he also did it after designing the A4/A5 at Apple. He leaves as mysteriously as he comes. Probably looking for better opportunities.

>>56332590
>get hired on to oversee two projects
>the creation of the high level arch for two new cores
>task is completed
>some other company offers you millions of dollars to work with them
>take the offer

Work on the Zen+ core was underway by May of 2015. Keller was reported to have left AMD in mid September of 2015.
He wasn't going to stay there and do nothing, and hes not a low level technician who does grunt work either. He wasn't brought on for a permanent position.

>>56332646
Keller was lead architect on K8, but he didn't design it. K8 was a slight evolution of the K7 core arch he worked on, but both were someone else'e brain child.
The work he did on Apple's ARM core arch is speculative as well. Apple acquired a vast wealth of IP from buying another firm.

>>56332134
That's exactly why AMD is keeping Zen quiet. AMD has made a bad habit of hyping the shit out of their products and then having them fall just shy of expectations, which is a huge let down for everyone.

>>56332853
Do you live under a rock?
Seriously.

>>56331991
Sauce?

>>56331991
>And it's still an engineering sample. The real thing has 10% more performance than Broadwell-E. At the price of Skylake.
hahahaha no
no way amd would price it that low

i say $500 MINIMUM for the broadwell-e competing cpus (which is still dramatically cheaper than what broadwell-e is).

>>56332470
because theres no competition
they are waiting for amd to release something worth buying

>>56331503
I've always read it as meem.

>>56332470
I'll probably wait for Zen+.

I'm using a laptop from 2005 right now. I'll probably be fine until 2018.

File: att.gif (2MB, 500x386px) Image search: [Google]

2MB, 500x386px

>>56333527

File: 1471799406511.png (79KB, 901x500px) Image search: [Google]

79KB, 901x500px

>>56331065
>wccftech.com

>>56331436
AMD is shit go fuk urself

File: 1443610333932.png (121KB, 663x430px) Image search: [Google]

121KB, 663x430px

>http://wccftech.com/amd-zen-cpu-performance-double-fx-8350/
>This means that a single Zen core is in effect equivalent to two Piledriver cores in performance
>TFW looking to build a new computer with one of AMD's current high end FX processors before I'm forced to use one of their new botnet enabled Zen processors
>TFW my botnet free build will be obsolete months after I complete it
Should I go for the full retard FX-9590 just to get a little extra performance or is it not worth it over the most powerful non full retard tier FX processor? It supposedly has the same performance as 2 of the highest end pre-IME botnet Xeons (and consumes an equal amount of power).

>>56333622
Stick to 8xxx and wait.

>>56333622
Just wait for Zen retard

and there'd be no way for it to double a piledriver core in performance, at most it's 50%

>>56333622
9xxx was a joke, if anything wait for zen or be a lesser fool with the 8xxx

>>56333639
Wait for what? I don't see things getting any better.

>>56333640
>>TFW looking to build a new computer with one of AMD's current high end FX processors before I'm forced to use one of their new botnet enabled Zen processors
Why the hell would I wait for Zen?

>>56333677
>new botnet enabled Zen processors
When has this been implied?

You're getting a new AMD CPU in months that will make FX look even older. Either low ball with FX-8xxx Or just get an unlocked Haswell chip or newer.

>>56333677
Every time I become intoxicated, it seems many other people's output turns to gibberish as well. Very strange, it's as though we're connected, or I'm generating you in some sollipsistic-esque way.

Either way, you're saying some dumb shit. You don't even know what you're using that might have a hardware backdoor, nor do you have full knowledge of the capacity for bugging a given device remotely than spreading it throughout a network and infecting your firmware. Most asinine is that you assume the FX 8/9xxx is with certainty, not backdoored.

Just get Zen, stay with what you have, or stop using computers. You can only fight the botnet with awareness at this point.

File: dissappointed housefire.png (165KB, 578x691px) Image search: [Google]

165KB, 578x691px

wccftech claimed that pascal was going to have hbm and be a huge upgrade over maxwell

theyre literally the least credible source on the planet but /g/ keeps posting their meme articles

>>56332652
>and hes not a low level technician who does grunt work either.
He is a certified shit wrecker

>>56333622
>Should I go for the full retard FX-9590
Only if you live in the artic

>>56333736
>When has this been implied?
Zen has AMD's new Platform Security Processor, which is a court ruling away from being a potential undetectable hardware back door (like Intel's Management Engine) and the US government is actively pushing for that court ruling (and at least one of the current presidential candidates openly supports such a ruling).

>>56333739
I'm not talking about some theoretical potential back door inserted when the chip was being designed. I'm talking about the very real chance of companies being forced to sign malicious firmware produced by the US government. With the access that AMD's PSP has to your system it would be the perfect hardware back door in such a case.

>>56333784
>pascal was going to have hbm and be a huge upgrade over maxwell
It does, and it is, the new TITAN X destroys everything on the market right now

>>56332590
Its how he works
Keller doesn't need AMD, they need him. They give him the dough and he gets the job done. It's how a certified shitwrecker works.

File: ipc uplift.png (10KB, 582x162px) Image search: [Google]

10KB, 582x162px

>>56333640
Zen is 60%~ higher IPC than Piledriver
Bulldozer modules all carry a CMT penalty meaning that per core performance suffers from the cores sharing resources. The throughput scaling is about 80% because of this. A single isolated core per module has a theoretical throughput of 1x, two cores per module have a throughput of 1.8x instead of 2.
Zen being an SMT design has no CMT penalty
Zen being an SMT design has an additional logical thread per core, and each of those threads brings about 30% the performance of a physical core in a well threaded workload outside of some freak niche instances

It would be disastrous for AMD if 8 core Summit Ridge didn't offer about double the multithreaded throughput of 8 core Vishera. They need that level of performance to compete in the segment they're aiming for in a ton of metrics: http://www.anandtech.com/bench/product/697?vs=1317

>>56333851
Pascal is just a rehash of Maxwell on a smaller process
It also doesn't have HBM.

Yeah the Titan XP is the most powerful GPU but it is no shitwrecker.

>>56333898
>It also doesn't have HBM.
http://arstechnica.com/gadgets/2016/04/nvidia-tesla-p100-pascal-details/
>Pascal is just a rehash of Maxwell on a smaller process
>Yeah the Titan XP is the most powerful GPU but it is no shitwrecker.
http://www.tomshardware.com/reviews/nvidia-titan-x-12gb,4700-5.html

I would consider that shitwrecking

>>56333677
You aren't even worth the government's time, stop this retarded ass meme

>>56333945
Hard to shitwreck when the best CPU can barely feed it. Unless you're buying 8k rigs diminishing returns hit fast.

Fuck APUs sideways.

Where are my new FX chips?

>>56333956
If the US government gets the court ruling they've been pushing for then every level of the government will have that ability right down to your local police.

>>56333969
>Hard to shitwreck when the best CPU can barely feed it
Isn't that the definition of shitwrecking when the best hardware is incapable of keeping up? I would say so ......

>>56333982
Well ain't that nice, it's probably already happening if you have a tin foil hat, so essentially, none of it matters

>>56334006
>concern about legal precedents that the US government is openly trying to set is tinfoil

>>56333956
It depends how much time we are talking.

File: 1291431487380.jpg (153KB, 960x720px) Image search: [Google]

153KB, 960x720px

After Bulldozer.. I can never be hype about an AMD cpu again.

My heart broke that day, still hasn't healed.

I miss team red...

File: BulldozerHotChips_August24_8pmET_NDA-6.jpg (240KB, 1500x844px) Image search: [Google]

240KB, 1500x844px

>>56334223
Bulldozer hype was 100% from delusional fanboys on forums and here on /g/.
AMD's official statements sure as fuck weren't misleading. The only exception was that marketing guy JF-AMD doing damage control on OCN.

At Hotchips 2010 AMD made it incredibly clear that there would be serial performance regression vs prior Phenom II chips. This was a full year before the launch of the FX 8150 when everyone on /g/ shit their pants as the reviews came out.

File: 1466806563635.jpg (76KB, 572x501px) Image search: [Google]

76KB, 572x501px

Post yfw Zen releases and it's absolute garbage and mid tier Intel CPUs beat the more expensive Zen offerings.

>>56334278
>much less area and power

wat

>>56334479
Not compared on ISO process undoubtedly.
Phenom II X6 die was 346mm2 in 45nm.
The Zambezi die was 315mm2 in 32nm.
Power consumption between the FX 8150 and X6 1100T wasn't that far apart either.
I'm sure they were looking at power per core/module there to try and make it look more favorable as well.

They're flat out stating that their Bulldozer design offered 80% of the throughput of a Phenom II per core.

The BD hype was just fanboy delusion. AMD spelled it out plainly for everyone.

Does AMD expect literally anybody to buy any of these fucking Bristol Ridge APUs? Raven Ridge will replace it all in 6 months at the absolute latest..

>>56334571
>>56334278
fanboy hype got to me good...

>>56331666
This is an engineering sample at lower than final clocks.

>>56333945
>tesla
irrelevant to end users

>>56334633
We will see, its more or less a placeholder and an excuse to get the AM4 motherboards out sooner

>>56331812
Limited availability is possible in late 2016.

>>56333861
>Bulldozer modules all carry a CMT penalty meaning that per core performance suffers from the cores sharing resources.
Actually, that's not what is happening. Bulldozer has tiny cores, that's why performance suffers. An 8 core bulldozer CPU has approximately the execution resources that a 4 core 8 thread Intel CPU would have (it's slightly less on the Intel, but whatever).

CMT is a half-assed fucked up implementation of SMT, where only a tiny part of a 2 core module is shared between threads while other resources are dedicated per thread - giving you 2 threads and two weak-ass cores per module. Weak ass, because as part of going for a huge core count meme they reduced each core in size to keep costs and complexity down.

Zen fixes it by sharing all resources on a single core between two threads. Sharing resources between threads is a way to increase performance. AMD just half-assed it with their meme architecture. There is a reason why CMT is an AMD exclusive meme. Nobody else is doing it.

File: 1471813742942.jpg (61KB, 1000x800px) Image search: [Google]

61KB, 1000x800px

Who /deneb/ here? Finally AMD will have a chip worth upgrading to. Piledriver was a mess, a waste, a big fat mistake.

>>56332365
You can't fit am3 on mitx it's literally bigger with its dual chip configuration than the X99 platform, and only asrock have been mad enough to try X99 mitx.

>>56333576
Zen probably won't be really shit wrecking for the HEDT market, however it will put AMD back into the competition and will rape the server market
Zen+ looks like what will put AMD in parity with Intel in the HEDT market
>>56332695
>Apple acquired a vast wealth of IP from buying another firm.
Part of what Apple acquired with PASemi was Keller, he was working there at that time >>56333622
>Should I go for the full retard FX-9590 just to get a little extra performance
No, the 8xxxe could overclock better than the 9590
>>56333945
All the wccftech articles were about the consumer lines, talked about the
Tesla/Quadro lines
>>56335044
>CMT is a half-assed fucked up implementation of SMT
CMT isn't SMT retard, learn about what you're talking about
CMT isn't why Bulldozer was bad, the leaky as shit process coupled with betting on high clocks was a disaster, AMD was expecting way too much from GloFo
Also, it was rushed, Bulldozer was redesigned quite a few times because management was pants on head retarded
>>56334633
The APU's have gained quite a bit of traction on the AIO and laptop market, Bristol Ridge is probably to fulfill contractual obligations with OEM's
>>56335113
Bulldozer was a mess, Piledriver was acceptable

>>56335228
I said half-assed smt, learn to read retard. No its not smt, but it takes the concept of sharing resources per thread and implements it in a retarded way. And yes, Cmt is why bulldozer sucks you stupid twat.

If they made phenom iis with smt using same process, they'd still be in the competition.

When can we expect budget Zen parts? AFAIK they're launching purely as server and HEDT parts?

>>56333622
it consumes as much power as two hexa-core westmere xeons while losing in multithreaded workloads

>>56335328
Sometime in the first half of 2017.

>>56331282
top kek

>>56335305
>I said half-assed smt, learn to read retard. No its not smt, but it takes the concept of sharing resources per thread and implements it in a retarded way
It's a clever cheap way of doing it, Alpha did the same with success
>And yes, Cmt is why bulldozer sucks you stupid twat.
It's not, CMT was by far the least problematic thing with Bulldozer, AMD bet on high clocks with a hueg pipeline, the leaky as fuck process completely destroyed that, the shared frontend was badly executed too
>If they made phenom iis with smt using same process
Piledriver was competitive at launch, specially on the server market
>they'd still be in the competition.
They're still competing, acquiring ATi saved them, the APU's are sold on a lot of AIO and laptops

>>56334633
>it all in 6 months at the absolute latest..
Raven Ridge is more like Q3-4/2017

>anybody to buy any of these fucking Bristol Ridge APUs?
Last desktop I had was over a decade ago, so yes, a DDR4 AM4 APU system does make quite a lot sense for someone like me to get started. The upgrade path to Zen/Zen+ and a dgpu in year or two is a quite big selling point.

>>56331991
at 3ghz. Why didn't they test it higher?

>>56336067
They were demoing IPC performance