Quantcast
[ 3 / biz / cgl / ck / diy / fa / g / ic / jp / lit / sci / tg / vr / vt ] [ index / top / reports / report a bug ] [ 4plebs / archived.moe / rbt ]

Due to resource constraints, /g/ and /tg/ will no longer be archived or available. Other archivers continue to archive these boards.Become a Patron!

/g/ - Technology


View post   

[ Toggle deleted replies ]
File: 1.22 MB, 1514x1278, zen.png [View same] [iqdb] [saucenao] [google] [report]
57618009 No.57618009 [Reply] [Original] [archived.moe] [rbt]

Are you ready for The Great Uncucking?

>> No.57618063

You mean like the last 20 "uncuckings"?

>> No.57618125

>>57618009
POO

>> No.57618131

>>57618125
IN

>> No.57618136

>uncucking
>not exceeding performance

>> No.57618141

>>57618131
GPU

>> No.57618177

>>57618136
No-one expects AMD to come up with a CPU that has better IPC than a company with like 100x the R&D budget these days. However, to nearly match IPC and offer twice as many cores for the same price would be epic and that would indeed be "exceeding performance" (at least in heavily threaded applications).

However, I'll believe it when I see it.

>> No.57618209

>>57618136
>simultaneous multithreading on all cpus
>4 cores at entry level
>probably less than half the price of all equivalent intel cpus

>> No.57618221

>>57618009
Let's go guys let's all overhype amd it's not like it's has ever been bad right?

>> No.57618234

>>57618009
Wait and see. Just wait and see.

Holding off till it gets released

>> No.57618243

From the looks of it there will be 8 slow cores again. They will just be faster slower cores.

For most games and applications the 6700k will trump with its 4k fast cores

>> No.57618266

>>57618221
Everything AMD has stated about Zen is a solid improvement, unless AMD is straight up lying about what they've added/improved on the thing can't be slower than Carrizo and Carrizo is definitely fast perf/watt it's just clocked low. If Zen is faster than Carrizo and it scales up then Zen should be a rather large improvement overall.

>> No.57618289

>>57618177
they did that last time, 8350 did have about as much power as 1st generation i5 but devided across 8 cores, to this day there are hardly any uses for that kind of power in every day casual computing

>> No.57618296

wccftech?

That just means this is bull-fucking-shit

but fuck it! lets keep building up that hype!

>> No.57618332

>>57618243
If you're not playing games, the AMD octacores are going to sweep the market if they are indeed priced at $300 with the Intel hexacores priced at ~$400 and requiring the expensive 2011 platform.

>> No.57618356

>>57618332
Not entirely sure about that
intel had better multithreading with LESSER cores because they were faster cores

>> No.57618372

>Hassan Mujtaba

>> No.57618387

>>57618009
Prepare for a massive disappointment.

It's AMD we're talking about here.

>> No.57618394

>>57618009
$500 for a binned chip is going to break poorfags heart. I don't see the big deal. They are comparing a 8 core to a 6 core intel. Why just not buy a 5820k?

>> No.57618408

>>57618063
Fpbp

>> No.57618472

I'm pretty sure that AMD's 'gamer' tier CPUs are the best marketing strategy that Noctua has come up with this far.

>> No.57618494

if AMD shills say it matches haswell then expect it to be 10-20% slower than the hyped performance. sandy bridge equivalent 8 cores for ~$300 even though you can buy sandy bridge 6 cores (3960x) for like $150 these days.

>> No.57618513
File: 105 KB, 255x345, CERTIFIED SHIT WRECKER.png [View same] [iqdb] [saucenao] [google] [report]
57618513

you better not fail us jim

>> No.57618564

>>57618009
P R I C E
E
R
F V
O S
M
A
N
C
E

>> No.57618566

>>57618494
It will almost certainly match Haswell in terms of IPC. The question is if it will clock high enough.
>though you can buy sandy bridge 6 cores (3960x) for like $150 these days.
Not everyone is comfortable buying used tech on eBay anon.

>> No.57618595
File: 54 KB, 960x540, 3086066-5455250892-polar.jpg [View same] [iqdb] [saucenao] [google] [report]
57618595

>>57618566
>It will almost certainly match Haswell in terms of IPC.

it almost certainly won't, lol. AMD has a long history of overstating performance in marketing material (pic related, poolaris was only a ~1.5x perf/watt improvement, not 2.8x).

>> No.57618599

>>57618009
If it takes 8 Zen cores to beat 6 Broadlel cores, it's not actually looking that great. That's really disappointing to me if true, I was hoping Zen could legitimately compete purely on a performance basis, but this would put them quite a bit behind either in IPC or in achievable clock and right in the same "only good for the value" category they're in now, although still much closer to Intel.

I've got a 4.7GHz 4790K, I'd love to be able to switch to a platform which can run PCI-E x16/x16 without losing single-threaded performance.

One thing's for sure though, if you're looking to buy a CPU now you should at least wait and see what Zen can bring to the table.

>> No.57618616

>>57618009
I am skeptical. The FX-6300 is a slow CPU that is popular because it's cheap, and often compared to modern day i3's even though that's just not true.

>> No.57618617

>>57618566
>Not everyone is comfortable buying used tech on eBay anon.

nothing wrong with buying a used CPU, they take decades to fail even if they're OCed 24/7.

>> No.57618650

>>57618599
You're forgetting that that 6-core broadwell cpu costs more than 2x what this will cost.

>> No.57618664

>>57618617
With CPUs I think the main concern would be buying an engineering sample or a cheap ass intel celery instead of what you want.

>> No.57618667

>>57618650
>You're forgetting that that 6-core broadwell cpu costs more than 2x what this will cost.

if by '2x' you mean '$50 more', then yes.

>> No.57618673

>>57618009
No infact I don't even think there is a reason to buy Zen.
Anything made by AMD is pure garbage nowadays. I miss Phenom II era, where the CPU and motherboard were fairly priced and delivered performance.

>> No.57618679

>>57618664

you can just file an item not as described claim and get all your money back then. ebay is cucked and sides with the buyer 100% of the time with stuff like that.

>> No.57618687

>>57618650
I am not, which is why I said
>this would put them [...] right in the same "only good for the value" category

>>57618667
A 6850K isn't $350 anon, more like $600 MSRP

>> No.57618691

>>57618599
If 8x3GHz Zen matches a 6x3.6GHz Broadwell-E, the IPC deficit is trivially covered by pricing adjustments.

AMD's prior problem is that Bulldozer through Excavator were *much* more than 10% behind.

>> No.57618715

>>57618650
>>57618667
>>57618687
a 6800k is 6*3.4GHz for ~$430.
a 6850k (which clickbait article said is the closest match) is 6*3.6GHz for ~$620.

>> No.57618719

>>57618687
>A 6850K isn't $350 anon, more like $600 MSRP

are you retarded? 5820k and 6800k are both 6 core CPUs that are $350.

>> No.57618730

>>57618595
This was factually correct, you are just a moron who think "Durr Hurr 2.8x means across everything"

When even most retarded marketer would tell you that you only provide best case scenario in marketing.

This is why you faggots go all over new Intel CPU that tells you it's "25% faster than previous generation" when in reality it's not even 5%

>> No.57618734

>>57618691
>pricing adjustments
So AMD is the "value option" once again, as I've said in the post you're quoting. Better for the money, but not better performing.

>>57618719
Are you retarded? That article is clearly comparing the top Zen CPU to a 6850k, not a 6800. Did you even read the title in the OP?

>> No.57618743

>>57618009

Can't wait until it comes out, doesn't deliver on any of the hype and promises and then we re-re-re-re-re-rewatch AMD marketers telling everybody how "n-n-nobody e-e-e-expected anything b-b-better! I a-a-actually t-think it's v-very good f-for the price!" just like every single AMD product since bulldozer.

>> No.57618746

>>57618719
> that are $350
> Recommended Customer Price $434.00 - $441.00
> Recommended Customer Price $617.00 - $628.00

>> No.57618753

>>57618719
>It's on sale in Newegg that means 350 is it's price

>> No.57618763

AYYMD HOUSEFIRES

>> No.57618774
File: 33 KB, 274x409, 1166026772871.gif [View same] [iqdb] [saucenao] [google] [report]
57618774

>no graphene
It's shit.

Let me put it this way OP:
Unless your CPU announcement threads contain an announcement of a CPU implementing graphene and giant frequency leaps, instead of core count jerking and energy buzzficiency,
take it and stick it up your ass and don't stain /g/ with it.

Unless anyone between Intel and AMD releases a graphene CPU, take all the CPU articles and stuff them in the toilet, or your pooper, and fuck off.

>> No.57618775

>>57618664
the main concern is usually having to put that shit into an older motherboard without the features you want.

>> No.57618776

>>57618734
> Better for the money, but not better performing.
There's no such thing as just "better performing". Of course, top of the line server processors are Xeons, but if we have a limited budget ($100, $200, $300, doesn't matter), then we have to look at the best CPU for our money.

>> No.57618779

>>57618730
>This was factually correct,

[citation needed]

there is no use case where poolaris gpus are 2.8x higher perf/watt over amd's previous generation(s), even if you compare it to the disaster hawaii

>When even most retarded marketer would tell you that you only provide best case scenario in marketing.

which is exactly why should view AMD's claim that they can '''match''' haslel as dubious as fuck

>> No.57618790

>>57618009
I'm ready to buy a new cpu and mobo in february, however I'm waiting for solid reviews before buying anything.
And let's not kid ourselves, if AMD's main advantage is pricing and not performance, I'll go with a Kaby Lake i7.

>> No.57618800

>>57618743
Don't forget that they'll only make 4 of the things and so the prices spike to double because there aren't any available, making them horrible buys but they sold all 4 anyway.

>> No.57618806

>January 2017
>Donald Trump takes office
>Immediately makes H1B visas void
>millions of pajeets have to go back
>3/4 of AMD's employees gone in a week
>2 months later AMD's market cap falls by 80%
>AMD gets bought by Nvidia
>Nvidia uses AMD IP and their own IP to make super efficient cheap low-end cards

THANK YOU PRESIDENT TRUMP

>> No.57618830

The honest truth about Zen is that it won't save AMD unless it sweeps up a lot of the OEM market, especially in laptops.

January 2017 will be meaningless for AMD's bottom line unless they capture very significant server buys that I just don't see happening because of conservativism.

As for the OEM market, Intel pays good money to keep them balls deep in Pentium/i3 boipussy, so any change there would have to constitute a major 3rd party revolt against Intel hegemony. If you can come up with a good reason why they'd do that, then perhaps there is serious hope in the immediate future. If not, then this is just the beginning of a very, very long fight where AMD will have to consistently improve on their design until they make some big wins.

>> No.57618856

>>57618595
See the little 2 on that statement? That means there are conditions to that statement. Find out what those are and I'll guarantee you that their figures are correct.

>> No.57618868

>>57618776
>There's no such thing as just "better performing"
What? If CPU A is faster then CPU B in every type of workload, then CPU A is better performing than CPU B. I'd say this also applies if CPU A is faster than CPU B in certain workloads, but has the same performance in others.

If Zen can match a 6850K in heavily threaded workloads but falls short in single/lightly-threaded tasks, then a 6850K is the better performing CPU.

>> No.57618870

>>57618806
>>Nvidia uses AMD IP and their own IP to make super efficient cheap low-end cards
That's not how monopolies work.

>> No.57618886

>>57618243
>tfw you will never see a CPU with 4096 cores
Why even live.

>> No.57618908

Laptopfag here. The zens seem interesting on paper but since AYYMD is incapable of making any mobile CPU worth while since the turions I think I'll pass...

>> No.57618951

>>57618830
Zen is more likely to do better in 2S rackmount servers where the on-chip 10(/40?)GbE and copious DDR4 and PCIe bandwidth will serve some applications better than having just the highest tier of x86 IPC.

These things will be the packmules hosting racks full of NVMe drives for content farms.

>> No.57618958

It will underperform just like Bulldozer and Polaris underperformed. CPU business needs a new competitor sooner than later to break this quasi-monopoly that Intel has.

>> No.57618987

>>57618243
All programs can use all available cores these days thanks to Windows 10, so even the 8350 is a damn good CPU

>> No.57618995

>>57618886
A CPU probably. That many cores would probably be made to handle a specific type of workload and at that point you might as well build specialized cores to deal with those workloads rather than general purpose cores.

>> No.57619006
File: 14 KB, 568x164, Untitled.png [View same] [iqdb] [saucenao] [google] [report]
57619006

Best we can hope for is a much needed kick in the ass for Intel. Monopolistic tech companies re branding the same products needs to stop being a viable business strategy.

>> No.57619036
File: 119 KB, 956x512, mobile.jpg [View same] [iqdb] [saucenao] [google] [report]
57619036

>>57618908
Apparently Zen scales down to a TDP of 5W.

>> No.57619044

>>57618987
>All programs can use all available cores these days thanks to Windows 10
Not that guy but what the fuck have you been smoking? Programmes that are fundamentally limited to a certain number of threads are still limited to that number of threads.

>> No.57619052
File: 52 KB, 1280x720, maxresdefault (3).jpg [View same] [iqdb] [saucenao] [google] [report]
57619052

>A few months ago
>Zen kills the Intel 6900k guys! Intel is finished!

>Turns out its barely as good as a 6 core Intel
>Which means its single core performance is going to be absolute shit...again
>post yfw we're in for another Bulldozer
>post yfw people fell for the Another Massive Dissappointment meme

>> No.57619060

Zen took too long. I already bought a fx 8370. If AMD wanted to sell me a Zen chip they should be released it.

It doesn't matter if it's great if it cannot be bought.

>> No.57619067

>>57619044
Okay stick with your i3 then buddy.

>> No.57619078

>>57619060

>buying bulldozer in 2016

(you)

>> No.57619093
File: 25 KB, 650x425, Jim-Keller-AMD.jpg [View same] [iqdb] [saucenao] [google] [report]
57619093

>>57618513
>doubting the SHIT WRECKER

>> No.57619098
File: 567 KB, 300x456, 2b5144e5270e5d9bb6187e4284b5f442.gif [View same] [iqdb] [saucenao] [google] [report]
57619098

>8 core AMD barely on par with a 6 core Intel which already struggles to compete with an i7 6700k in single threaded applications

>AMD literally put MOAR SHITTY CORES instead of good ones again and AMDrones ate their vomit up like candy

Its amazing how much people claim Nvidia/Intel shilling but AMDrones go through these amazing lengths of bullshit

>> No.57619113

just like bulldozer was supposed to be the second coming of Christ, surely zen will meet all the expectations as well :^)

>> No.57619128

>>57619098
>2 more cores
>MOAR COARS

I understand this about Sandy and FX back in the day, but this doesn't make any sense.

>> No.57619135

>>57618806
how would microsoft ever recover? would they just move their HQ from redmond to bombay?

>> No.57619152

>>57619078
Sorry but I had zero intention of buying from intel, what with the anti white people hiring practices and h1b visa abuse on a massive scale.

But that's not the real point which is Zen simply hasn't materialized. Fuck waiting. Fuck delays. And fuck Zen. Shit should've dropped months ago.

>> No.57619156

>>57618009
why is it so expensive?

>> No.57619165

>>57619128
It means after 5 years AMD will be barely on par with Sandy Bridge in single core

Broadwell E has shitty IPC as it is, and they couldnt even beat that

>> No.57619166

>>57619036
We'll have to wait and see. Most people that I know had amd powered laptops had overheating issues (it's not just amd but when that bottom plate deforms you know you're in some deep shit)

>> No.57619168

>>57618332
>not playing games
lmfao who does that fucker

>> No.57619178

>>57619152
Zen was promised at the end of 2016, instead of waiting for a fucking month, you bought a piece of shit heater that can also do simple math.

>> No.57619196

>>57619152
>h1b visa abuse

you realize AMD abuses the H1-B visa much worse than Intel right? they fired half their engineers and replaced them with pajeets

>> No.57619203

I'm buying Zen on day 0.

>> No.57619218

>Polaris is gonna beat a GTX 980!
>This just in, its actually just a shittier GTX970

>Zen is better than a 6900k!
>This just in, it barely matches a 6 core 6850k

>> No.57619223

>>57618009
>Performance as 6850K

Let me guess. Their 8-core will be on par with Intel 6850 6-core in video encoding...

>> No.57619254

>>57619223
just imagine how pathetic the IPC is gonna be

AMDrones claimed it was gonna be different this time, too...

>> No.57619269
File: 172 KB, 736x558, OG90avx.jpg [View same] [iqdb] [saucenao] [google] [report]
57619269

>>57618296
Not baiting, but i want to know what wccftech has been wrong on, do you know?

I keep hearing it, and it seems plausible, but I can never think of a time that wccftech has been provably wrong.

also pic related.

>> No.57619270

>>57619165
I don't know what benches you have seem, but 6850k does outperform sandy in single core performance.

>> No.57619277

>>57619196
By sheer number intel actually has a bigger h1b abuse...also publicly anti white people is fucked up.

Intel will never be allowed in my home again.

>> No.57619283

>>57619269
wccftech was first one to jump on bandwagon that rx480 would be same performance as gtx1070 for way smaller price.

>> No.57619293

>>57619067
I've got an a10 laptop and an fx 8320 desktop, both running windows 10. But keep thinking that windows 10 magically makes software more multithreaded than it is.

You may want to look up Amdahl's law though.

>> No.57619301

>>57619270
But realize this is a 8 core AMD competing with it

>> No.57619311

>outperform
*** in proprietary video games, the most important application on Earth

>> No.57619331

>>57619166
That probably has more to do with poor budget laptop design that AMD running hotter. I've got an a6-4455m in a HP laptop. Conventional wisdom will tell any sensationalist anon that this thing should catch fire on first boot. It runs fanless most of the time, while still keeping the APU cooled to ~40-45 celsius. Speccy still shows the temp as 123 celsius because fuck speccy.

>> No.57619348
File: 93 KB, 500x405, 1479402493665.png [View same] [iqdb] [saucenao] [google] [report]
57619348

>>57618806
>wanting monoplies
I WANT ANCAPS TO LEAVE

>> No.57619370

>>57619301
What you need to understand is, there was price leak and some small spec leak, the whole "on Par Core i7-6850k" is complete click bait.

>> No.57619475

>>57619293
Nobody gives a shit about amdahl's law because nobody has a purely sequential problem that needs significantly more than even one second of CPU time. Especially if SIMD is used. CPUs nowadays have enough single core performance that most inherently sequential workloads finish quick enough already.

For example. Depth first search is inherently sequential if you want to run it on the entire tree. In practice that doesn't matter. You can split a binary tree into two trees and then run the DFS algorithm on two cores.

>> No.57619515

>>57618394
isn't it like 700?

>> No.57619523

>>57618009
if what they say is true, i can't wait for intel prices to crash. imagine a 6700k or 700k for around 200 bucks.

>> No.57619547

>>57619523
7700k. fuck.

>> No.57619564

>>57619523
7700k is already somewhat confirmed $200 more than 6700k

>> No.57619632

>>57618009
next year is going to be fantastic. either insane CPU +GPU combos for mid-tier prices or everything fucks up again and prices rise.

>> No.57619713

>>57618009
Do they sell the 8 core CPU for $300? Or does the article refer to the 4 core CPU? $300 for 8 real cores sounds pretty good to me even if the performance is lower than an equivalent intel processor.

>> No.57619737

>>57619713
you can already get 8 real cores for $70 on ebay

>> No.57619745

>>57619564
it's going to be 350 bucks at launch as skylake was. and it is the ""mrsp""". probably it will be 400-450 dollars

>> No.57619772
File: 67 KB, 523x440, Battlefield 1 1080p cpu.png [View same] [iqdb] [saucenao] [google] [report]
57619772

>>57618289

Holds up pretty well in modern vidya.

>> No.57619802

>>57619737
The CPUs are cheap on ebay because the mainboards are expensive.

>> No.57619809
File: 136 KB, 773x418, WSSbRJM[1].png [View same] [iqdb] [saucenao] [google] [report]
57619809

Well there's certainly a market for MOAR COARS even if they're inefficient, it has to be cheap though.

>saw pic related a few weeks ago
>price is back to 200, which must eman that people are willing to pay

Either that or they're afraid of a cold winter.

>> No.57619812

>>57618779
this 480 runs at about 90 watts and it's not the only one. Whether that's due to better asic quality, improved drivers, or both... i don't know.

https://www.youtube.com/watch?v=zWASNajSdpg

>> No.57619838

>>57619812

The XFX GTR 480 is the best binned polaris chip you can buy. Much like the nano is the best binned (for low power draw) of the full fiji dies.

Hell even on the cpu front the 8370e is a 4ghz 8 core chip with a TDP of 95w - intel doesn't have an 8 core chip with that low TDP (mostly because hyperthreading but eh, its still a technically true statement).

>> No.57619862

>technology has slowed to such a crawl that /g/ is excited for AMD catching up to Sandy Bridge after 5 fucking years

>> No.57619863

>>57618177
You have to remember that AMD is repicking the lowest hanging fruits of performance optimizations for their CPU. Bulldozer was a step backwards in single-threaded performance, even compared to the older K7 that preceded it, while Intel had already swiped all the lowest-hanging fruits with Nehalem/Westmere/Sandy Bridge. One of the reasons why Intel's gotten stagnant on the IPC end is because there is no more room for them to grow, they can either up the clock rate of their CPUs through a more efficient architecture or reduce CPU usage for idle or background operations like encryption or encoding. AMD has gone a step back by creating a compute bottleneck with shared resources, so all the gains they made with Zen were effectively free since the know-how for the process existed with K7 and continued on with the Cat cores.

And even then AMD is still stuck in the Nehalem/Westmere era seeing as how gimped the core multipliers are (less than 4.0 GHz for all SKUs). Zen+ is probably going to be their version of Sandy Vagina.

>> No.57619894

>>57619862
Competition is healthy, anon.

>> No.57619911

>>57618806
>Nvidia uses AMD IP and their own IP to make super efficient cheap low-end cards
If by "cheap" you mean at least double the price of current offerings, sure.

>> No.57619930

>>57619894
>AMD
>"competing"

being years behind both of your competitors who have already been struggling to push past their current limits says a lot about how retarded they are

>> No.57619936

>>57619713
why does their reality matter to you?

>> No.57619949
File: 35 KB, 650x337, 74847.png [View same] [iqdb] [saucenao] [google] [report]
57619949

>>57619930

Why can't intel make an igpu worth shit?

>> No.57619959

>People think AMD counts as competition

literally a token competitor so the gubbmint doesnt go after Intel or Nvidia

nothing they make is actually good

>> No.57619968

I dont understand all the amd hate. I have a $62 athlon 860k and it eats everything i throw at it.

>> No.57619969

>>57618830
The only thing that could make Zen viable for OEMs is if they made a 4 core sub-35W TDP mobile processor that uses less power overall than Intel's Skylake-U, better sustained frequency over Skylake-U, or made all mobile Zen CPUs parts scalable like Intel has done since the Core 2 era. If they somehow made an 8-core sub 35W-TDP mobile CPU with more available PCIe lanes than Skylake, then Zen could have the Apple MacBook Pro market on lockdown for the forseeable future.

>> No.57619970

>>57619165
>Broadwell E has shitty IPC as it is, and they couldnt even beat that
The IPC is literally identical to skylake, except in a few select scenarios where skylake beats it by 5-10%.

>> No.57619989

>>57619772
>that jump from the i5-4670K to the FX-8350
>tfw the 6600 somehow over the 8350.

Was there actually a huge perf increase from haswell to skylake, or is something else going on here?

Anyway, good to see Frostbite engine once again BTFO the i5 crowd.

>> No.57619990

>>57619036
>Styx and Basilisk
Fake and Gay

>> No.57620002

>>57619936
Because I don't want a processor where two cores depend on a shared resource. If I want that I could get a processor with SMT.

>> No.57620029

>>57619949
They can but only if you compare then to their $300 CPUs which are going to be paired with a discrete GPU anyway.

>> No.57620043

>>57619036
Stop reposting these fakes, you fucking retard.

>>57619196
AMD has more American employees than intel does.
Intel just fired 12,000 US born employees and replaced them in Indians.

Intel has more H1B street shitters than AMD has employees period.

>> No.57620044

>>57619968
Because Arma 3 doesn't run as well.

>> No.57620063

>>57619989

>tfw the 6600 somehow over the 8350.

That one is easy to explain - despite twice the threadcount it has less than half the IPC of the 6600.

>Anyway, good to see Frostbite engine once again BTFO the i5 crowd.

The writing has been on the wall since mid last year - the days of the i5 aren umbered for a high power gaming machine. Christ the 9590 in that pic is only as fast (per core) as ivy bridge due to the insane clockspeed and it hangs with chips many, many generations newer.

>which are going to be paired with a discrete GPU anyway.

Kind of defeats the point no?

>>57620044

ARMA 3 doesn't run well on anything - its almost broken by design.

>> No.57620077

>>57618774
>Thinking graphene will come any time soon
Shiiieett, do you think that cars will fly in a decade too?
I'm still waiting for hydrogen batteries too.

>> No.57620078

>>57619970
the 6700k kills the 6850k in single core applications
now imagine only 25% of the power of the 6850k since AMD needs 2 more cores just to get on par with it

>> No.57620079

>>57619969
friendo, they already have an 8 core at that TDP, it's in the playstation and it's a Jaguar core.

>> No.57620090

>>57620002
t. doesn't understand how HT actually affects single thread performance or how you can turn it off>>57620063
.

>> No.57620098

>>57619949

>comparing intel's lowest end igpu (gt2) to amd's highest end

really makes you think, doesn't it...

>> No.57620113

>>57620063
what I don't understand about that claim concerning the 6600 is that it appears so different from the haswell chip under the 8350, not the difference between the 8350 and the 6600.

>> No.57620116

what if Zen is actually god tier but the Pajeets are underhyping it because overhyping Polaris didnt work?

>> No.57620125

>>57620078
The 6700k is about 10% faster than 6850k in single-threaded tasks... mostly because it's clocked 10% higher. The IPC is practically identical.

>> No.57620134

>>57620079
Gee, I wonder why no OEM ever picked up Bobcat or Jaguar in large numbers. When was the last time you saw a laptop with an AMD quad-core APU? There is a reason for all things and "muh Intel monopoly" is not one of them.

>> No.57620150

>>57620113

AVX2 perhaps? I honestly don't know.

>>57620116

Polaris is everything AMD said it would be. It was the retards on the internet that decided polaris would be a 1080 killer and ran with it.

>>57620134

> I wonder why no OEM ever picked up Bobcat or Jaguar in large numbers

Probably due to the colossal discount Intel gives OEMs.

>> No.57620178

>>57620098
>No! Don't compare the $140 AMD chip to a $210 Intel chip!
>Compare it to a $350 Intel chip, that's much more fair!
If this is you thinking, I'd hate to see what you're normally like.

>> No.57620200

>>57620150
Nope, it has to do with AMD being backwards as fuck with their mobile processor design and requiring OEMs to design completely different PCBs and layouts for their Jaguar/Puma cores and Steamroller/Excavator cores. And even with Excavator, there were different memory requirements that forced all OEMs to stick with single channel memory out of cost considerations, which made Carrizo-L slow as fuck and sell poorly. Why the fuck would OEMs waste their time with such a disorganized company when Intel's been producing low-cost and significantly higher performing solutions for years?

AMDrone BTFO

>> No.57620211

>>57620077
>thinking graphene wouldn't come sooner if retards like you didn't waste money on 3% performance increments, and made slowpoking technology a profitable venture

If these retarded and irrelevant CPU updates suffered sharp interest and projected profit losses from the 30th percentile and upwards,
graphene would be here within 4 years, and that's a guarantee.

But stupid fucks like you and OP can't stop marketing slowpoke shit for free and giving relevance to it, so of course graphene will not even come out in 100 years if stupid fucks like you keep doing what you do.

>> No.57620224

>>57620134
Are you retarded?
Rhetorical question, of course you are.

Jaguar based parts are some of AMD's best selling ever. They were in hundreds of crapbooks.

>> No.57620231

>>57620178
>moving the goalposts from performance of the igpu to price after being called out

try harder pajeet

>> No.57620233

>>57620134
>When was the last time you saw a laptop with an AMD quad-core APU

I have one, with A4-5000 chip, my father in law has one with A6-6310. There are a few of them floating around, just not as many as Intel based laptops.

>> No.57620277

>>57620224
>They were in hundreds of crapbooks.
Compared to the thousands of Intel Airmont and Silvermont in those same crapbooks which sold significantly better than their AMD counterparts because they weren't battery-draining pieces of shit.
If I'm the retard, you must be a vegetable.

>> No.57620313

>>57618009
I've been waiting since Athlon

>> No.57620334

>>57620090
I don't see how HT affects single thread performance at all unless you are fully utilizing the integer units. The only problem in that case is that the cache is split between the hyperthreads which is probably a solved problem already via cache partitioning.

You use HT if you have a branchy or non continguous workload with cache misses or when you want to utilize both the integer units and the FPU at the same time.

If I want more than 4 physical cores because I need to fully utilize the integer units then having 8 logical cores doesn't help me at all but having 8 additional SMT threads on a processor with 8 cores doesn't hurt me that much (assuming that AMD also has cache partitioning).

>> No.57620378

>>57620200
>requiring OEMs to design completely different PCBs and layouts for their Jaguar/Puma cores and Steamroller/Excavator cores
Yeah, because the compatibility between the Atom series and the core m series is legendary.</sarcasm>

Also, up until recently the kitty cat cores performed better than the Atoms while using less power. The Atom series took fucking ages to be anything other than painful. Remember the power guzzling ion chipset that was mandatory for three generations? Those were great! And totally compatible with the core i series!

>> No.57620379

>>57619218
>Polaris is gonna beat a GTX 980!

Nobody said that ever. From the first leaked firemark scores, we knew that it'll just be a 390x with half the memory bandwidth. And that's exactly what it ended up being.

Besides it does beat the 980 in DX12.

>> No.57620396

>>57620231
>Can't see the glaring problem with comparing a budget chip to an enthusiast chip
Even you can't be this stupid Jamal.

>> No.57620407

>>57620379

>Besides it does beat the 980 in DX12.

To be fair so does hawaii. Hawaii all in all is aging really well.

>> No.57620408

Funny how they paired it against the more expensive 6850k when the 6800k is exactly the same, except for the extra 12 lanes that the 6850k offers.
6800k is like 40% cheaper.

I do hope that they manage to pull through though.
Intel needs some real competition.

>> No.57620418

>>57618009
wouldn't it be uncuckening?

>> No.57620421

>>57620408

Imagine if full fat HEDT Zen had 40 lanes for mainstream i7 prices - that would be a serious game changer for a lot of people (no, not gamers).

>> No.57620422

>>57620407
>To be fair so does hawaii.

That's what I said - Polaris is a 390x with less memory bandwidth.

Which is technically impressive cause it also has half the ROPs compared to Hawaii.

>> No.57620445

>>57620231
AMD doesn't produce APUs in that price range. Perhaps Intel has found a niche that they are objectively supperior at? Maybe there are some high end iGPU users that needed that additional 10fps boost but weren't served by AMD's products. Good for them.

>> No.57620452

>currytech

AMD false hype strikes again.

Zen POS confirmed.

>> No.57620459

>>57618958
We need some ARM based platform with SATA and possibly pci-e. The rate ARM is improving compared to x86-64 is in a whole other class.

>> No.57620482

>>57620211
>>57620211
>But stupid fucks like you

My CPU is from 2010...

>> No.57620485

How much is the 4C/8T chip slated to cost?

>> No.57620490

>>57620378
>Yeah, because the compatibility between the Atom series and the core m series is legendary
But there were no market overlaps, unlike Jaguar and Steamroller.
I shit you not, the dual-core A6 and Jaguar A4 were priced almost the same to the OEM and featured near identical hardware requirements, but using the Jaguar CPUs needed a completely different PCB because it was an SOC. It made the A6 obsolete, but AMD kept pushing it and the equally gimped A8 APUs because of costs. The A8 and A10 could have been better mobile APUs had they had dual channel memory as standard, but some retarded as fuck Pajeet decided to make the lower end APUs only use a single channel memory layout "to save costs". This ended up driving up the cost of producing dual-channel board variants, which is why almost all APUs of that era only came with single-channel memory and ruined the A8 and A10 APU's chance at competing with i3s and i5s.
Intel played it smart by differentiating their mobile CPUs not only by physical content, but by TDP and use case. The Core m are specifically designed for fanless or ultra-thin laptops, the Core i U are for more powerful laptops, and the Pentium/Celeron Apollo Lakes are for Chromebooks and tablets. But all of them are based on similar architectures with similar hardware compatibility and minimal upfront manufacturing costs between designs. This is why Intel is so far ahead of AMD in the OEM market. Even AMD admitted it after Carrizo failed.

>> No.57620496

>>57620459
Xeon-d is widely ignored by ARM enthusiasts when this topic comes up for some strange reason.

>> No.57620513

>>57620459
just becuase it's improving doesn't mean it isn't going to trail off and start settle at a shitty low power point incapable of anything worthwhile like it always has been

>> No.57620525

>>57619713
The prices start at $200, and Summit Ridge starts at 4 cores/8 threads. $300 for the 8 core sounds unrealistic - I'd imagine it to be higher seeing as it would have much better value than the entry level CPU at $300.

>> No.57620527

>>57620211
>thinking money can fix basic physics
mass producing graphene is still far away, if impossible.

>> No.57620536

>>57620002
Zen has SMT

>> No.57620581

>>57620445
>AMD doesn't produce APUs in that price range. Perhaps Intel has found a niche that they are objectively supperior at? Maybe there are some high end iGPU users that needed that additional 10fps boost but weren't served by AMD's products. Good for them.

intel owns 80% of the gpu market (and nvidia owns 85% of the remaining 20%), so if there is any niche brand it would be AMD's, which offers worse perf/w for less money.

>> No.57620595

>>57620459
Right now you're better off with a J1900 based intel mini itx mainboard. Sure it's more expensive than a Raspberry PI but you get SATA 6gb/s ports, PCIe for expansion, a better software ecosystem and it has a TDP of 6W. It's hassle free. With ARM you have to make sure that your board is either supported by the manufacturer or that the community fixes whatever bullshit the manufacturer did and the IO is crippled because they rely on old tablet SoCs that only have USB 2.0 because that is what is used to charge them.

>>57620496
ARM's big advantage in the datacenter were the integrated 10GBE NICs. Now that Xeon D's have it there is basically no advantage. The ARM CPUs don't even have application specific accelerators beyond encryption. Even simple gzip hardware accelleration would be enough to justify themselves as loadbalancers or NAS but they don't have any of that even though that is the primary advantage of ARM.

>> No.57620603

>>57620527
>thinking money can fix basic physics
>physics
Physics was resolved long ago,
the problem currently is production,
which is engineering,
and engineering problems can be fixed with money.
Just ask WW1 and WW2.

>> No.57620606

>>57620536

The blemnder demo leads to some uncomfortable possibilities when it comes to zen and SMT. Either AMD's Single thread is superior to broadwell-e (with worse SMT or the single thread is worse but their SMT scaling is superior to Intel's.

>> No.57620610

>>57619093
Alex Icke

>> No.57620623

>>57620595
> it has a TDP of 6W.
I'm not sure where I got that from the TDP is 10w.

>> No.57620658

>>57620623
I bought a 5350 for some strange reason. It seemed like a better deal than a celeryon. 35w tdp is also nice i suppose.

>> No.57620690

>>57620595
>TDP of 6W
No it doesn't. That the J3150 you're thinking about and it's shit. It spends most of its time turboing up to its max frequency and ends up being less power efficient than the J1900 that came before it. It uses at least 60W on its own, not 6W. I actually own the Pentium N3700 which shares the same architecture and it draws no less than 70W from the wall.

>> No.57620694

>>57620490
>But all of them are based on similar architectures with similar hardware compatibility and minimal upfront manufacturing costs between designs.
Yup, that totally explains the 15 watt celeron and pentium chips that are based on Atoms and occupy roughly the same space as the core i3 u chips. You're also greatly exaggerating the overlap between the kitty cat cores and the construction cores and ignoring the vast difference in IGP performance between the two.

The whole shared motherboard thing came along with Carrizo, which was the first construction core SoC, allowing it to be unceremoniously plonked into the same motherboards as the kitty cat cores use. That's what caused the shitty OEM boards with single channel RAM and it was probably also responsible for the wider availability of Carrizo than of Kaveri.

>> No.57620733

>>57618779
The use case they offered was the r9 270x at 180w tdp going up against the rx 470 at 110w.

>> No.57620771

>>57620694
>occupy roughly the same space as the core i3 u chips
But it doesn't. The Pentium/Celerons are meant for low-end tablets and Chromebooks and the i3s are entry-level notebooks with actual heatsink+fans. The TDP requirements are entirely different due to their market differentiation. You can not put a Core i3 U in the same Chromebook as a Pentium/Celeron, it would exceed the designed TDP of the i3. The Core m is designed for the higher-end ultra-thin laptops. The Core i Us are for regular laptops.

AMD made no such distinctions between Steamroller and Jaguar, but tried to make those distinctions with Puma and Excavator. It failed because both the Puma and Excavator had similar TDP requirements and could not be used in tablets due to their higher battery drain and heat output.

>> No.57620791

>>57618806
>AMD
>American
Literally all their business is outside the US
Trump will not get rid of visas

>> No.57620812

>>57620771
Some chromebooks do have a i3.

>> No.57620833

>>57620812
But they're thicker than your average Chromebook because they have a heatsink+fan. They're garbage versions of the ultra-thins using the Core m3.

>> No.57620911

>>57618009

they'll fuck it up

>> No.57620939

>>57618063
word.

>> No.57620944

>>57620771
>The TDP requirements are entirely different
Yes, you're right, I'm sorry, I forgot that there is an absolutely MASSIVE difference between 8 watts and 10 watts. </sarcasm>
>AMD made no such distinctions between Steamroller and Jaguar
Again, you're absolutely right, there's no difference between something occupying the 5-15 watt group and something occupying the 17-35 watt group.
>could not be used in tablets due to their higher battery drain and heat output.
You do realise that there have been kitty cat cores with ~2.5 watt SDPs for a number of years now, right? I also hope you realise that there have been Atom based tablets since long before Atom became competitive with kitty cat in terms of performance/watt.

>> No.57621009

>>57620911
Sounds like this competition is going to be like breaking into servers of a company. It doesn't matter how good you are. It matters how bad your target is at security. If intel can't improve performance with the generation after kabylake then AMD might catch up.

>> No.57621052

>>57620658
25w TDP son.

>> No.57621276

>>57620944
>I forgot that there is an absolutely MASSIVE difference between 8 watts and 10 watts
There actually is a difference between 6 watts and 10 watts. 6 watts is manageable for ultra-thins that use the aluminum case as a heat spreader. 10 watts and over needs other passive cooling solution.
> no difference between something occupying the 5-15 watt group and something occupying the 17-35 watt group
But that's not how they were built and sold by OEMs. Jaguar CPUs were shipped with 15W TDPs and the A6s were also shipped with 17W TDPs. There was no market differentiation between the two.
>You do realise that there have been kitty cat cores with ~2.5 watt SDPs for a number of years now, right
And how well did they sell? Exactly. Equivalent Atom tablets outsold them because they were faster and had four proper cores rather than two shitty cores while being in the same thermal and power requirement.

>> No.57621374

>>57621276
>6 watts
8 is 6 now? Wow! Didn't know those were interchangeable!
>There was no market differentiation between the two
You know, aside from the noticeable performance difference CPU side and MASSIVE performance difference on the GPU side. And the price difference. But hey, core m chips overlap with Atoms, don't they?
>because they were faster and had four proper cores
They weren't faster, the kitty cat chips do have four cores and the SDP is equivalent. What difference is there? Branding and price, along with several special deals. Intel was selling their Atom chips at break even prices until very recently, something AMD can't afford to do.

>> No.57621413

>>57618715
and mobo which costs 2 times more than normal one

>> No.57621452

>>57619838
>Nano
>Binned

Fury is binned Fury X, nano has exactly the same chipset as Fury X.

>> No.57621456

>>57620211
I'm sure 1 person (me) out of 100000000000 (everyone else) that buys an Intel CPU will make Graphene come faster.

>> No.57621477

>>57621452
That's not what binned means, idiot. Fury is a CUT-DOWN version of Fury X. Nano is a BINNED version of Fury X. Two entirely different things.

>> No.57621503

>cinebench
>rumours from wwctech

>> No.57621534
File: 13 KB, 460x276, panic460[1].jpg [View same] [iqdb] [saucenao] [google] [report]
57621534

>>57618009
>performance on par with core i7-6850k
>8 coarz

>> No.57621537

>>57621477
Nano has no drawback, it's exactly the same fucking chip, only difference is air and water cooling.

>> No.57621560

>>57618177
Why not?

I'm betting that most of that R&D funds goes into things like manycore accelerators and xpoint these days

>> No.57621579

>>57621560
also fabs

>> No.57621587

>>57619772
>i7 2600k better than the i5 6600
what

>> No.57621595

>>57621560
Because they have no money to fuck around and to test, they need to create one thing and perfect it.

>> No.57621616

>>57621587
Vidya

>> No.57621663

>>57621456
Graphene will only come faster when companies hit a wall

>> No.57621751

>>57621374
It was always 6 watts, retard
http://ark.intel.com/products/87261/Intel-Pentium-Processor-N3700-2M-Cache-up-to-2_40-GHz

>> No.57621768

>>57621537
Nano is better binned than Fury X. That's why it has such a low power draw.

>> No.57621817

>>57621768
or maybe it's because it throttles to no end and never can actually hold a stable boost clock.

>> No.57621836

>>57621374
>You know, aside from the noticeable performance difference CPU side and MASSIVE performance difference on the GPU side
Literally doesn't exist. In fact, the quad core Pumas were faster than the dual-core Carrizos. The GPU were well within parity too. The Carrizo's GPU was gimped by the lower frequency and single-channel memory.
>They weren't faster
They were, retard. Tablet Jaguars peaked at 1.0GHz and Pumas only hit 1.2GHz. Wow.
Meanwhile 2013 Atoms were hitting 2.41GHz with the same number of cores with similarly powerful cores and lower SDPs.

I'd love to see you refute this.

>> No.57621876

>>57621836
>They were, retard. Tablet Jaguars peaked at 1.0GHz and Pumas only hit 1.2GHz. Wow.
Meanwhile 2013 Atoms were hitting 2.41GHz with the same number of cores with similarly powerful cores and lower SDPs.

What does Ghz have to do with anything?

>> No.57621957

>>57621876
They had similar IPCs at the same frequency. An Atom Z3735 quad-core at 1.33GHz with Turbo disabled had the same single-threaded and multi-threaded scores on Passmark as the tablet A6-1450 with turbo enabled to reach 1.4GHz.
Going from 1.33GHz to 2.41GHz boost going to be almost twice as fast as a shitty little AMD quad core that can't even hit 1.4GHz on all four cores while using more power.

>> No.57621963

>>57621836
>I'd love to see you refute this.
There is no refuting this, there is only headdesk. Facepalm does not cover this.

You have convinced me with your arguments of gigglehurtz and claims that sixes are eights.
>>57621751
Yes, there is only one atom based pentium. You have seen through my clever ruse of referring to a higher model.

8/10 troll, I legitimately thought I was speaking to someone holding an uninformed, idiotic opinion.

>> No.57622034

>>57621963
>comparing core frequencies are irrelevant when both architectures perform within a margin of error with each other at near similar clockspeeds.
>and one CPU line uses the same amount of power while hitting almost twice the peak frequency than the other
Who's the retard again?

>Yes, there is only one atom based pentium.
And all the Airmont CPUs had similar or less design TDP. Only the desktop variants went up to 6.5W
You're a fucking braindead AMDrone who can't even into maths and is probably still in his first year of a shitty CS degree from a diploma-mill.
Nice try, but GG senpai. You're done.

>> No.57622074

>>57618009
Did you mean The Great Unification,
thanks to HSA and HBM?

>> No.57622075

>>57619989
Windows 10 Anniversary Update lets programs use all available cores, so the 8350 has actually made a pretty huge comeback. I imagine thread count will be more important than single core performance for gaming now that the 4 thread limit is done with

inb4 that one autist starts spouting shit. One of my friends literally worked on the update part time, so he knows the code. Went from 55fps average on Ass Creed Unity with his 8350 and 980Ti Classified SLI to over 120fps average. The OS was the only difference. Also runs Witcher 3 4k/ultra/60fps with around 30% max usage spread across all 8 cores. It's insane how much of a difference it's made to gaming performance for the 8350

>> No.57622150

>>57622034
>Who's the retard again?
That'd be the guy who referred to pissmark. That's the thing that made it blatantly obvious that you're trolling.
>And all the Airmont CPUs had similar or less design TDP. Only the desktop variants went up to 6.5W
I'm not even going to bother addressing this. There's no need.
>You're a fucking braindead AMDrone
The funny thing is that I'm typing this on an Intel machine and I don't actually own any AMD hardware. I'm very bad at this whole AMDroning thing but keep coming with the insults, it definitely doesn't give away the fact that you're trolling.
>GG senpai
You do realise that it's usually the losers who type GG before leaving, right? So called "aggressive" GGs are extremely rare.

>> No.57622154

>>57621957
Why are you comparing Z3735 to A6-1450?

Z3735 is 6months older, uses 4watts less and performance half as bad as A6-1450

>> No.57622165

>>57622150
You're done. GG no re

>> No.57622244

>>57622075
>inb4 that one autist starts spouting shit.
So that'd be you making that same claim earlier. Can you post a link where some website gives details on this part of the update? Many some benchmarks from a reputable website? This would be pretty MASSIVE fucking news if it were the case.
>One of my friends literally worked on the update
Yeah, well my uncle works for Nintendo and he says you're full of shit.

>> No.57622283

>>57622075
Wot did they change?

>> No.57622321

>>57622283
Use_All_Available_Cores = True;

>> No.57622375

>>57622075
>so the 8350 has actually made a pretty huge comeback
lol no
There is not a single tech outlet reporting anything close to what you're claiming. The Anniversary Update did jack shit to AMD's older architecture. In fact, a quick check on Google shows that Windows 10 Anniversary update released with major bugs for AMD's platforms.
>https://community.amd.com/thread/203859
>https://support.amd.com/en-us/kb-articles/Pages/Windows-10-Anniversary-Update-Crimson-Install-Issue.aspx
What a troll.

>> No.57622481

>>57620043
its okay those fired guys will just go work for AMD, r-r-r-right?

>> No.57623476

>>57621817
Holy shit just google "chip binning" and stop talking about things you have no clue of. No one's saying you're right or wrong, you're just having a separate argument.
>the Euro lost stability because of Brexit
>no, you're wrong because peaches are a better overall nutrient source than apples

>> No.57623563

>>57620791
This statement is false.

>> No.57623640

>>57619052
What does "matching a 6 core Intel" have to do with single-threaded performance? In fact how is the first part even bad? You didn't even mention a price point. Your entire post was embaracing.

>> No.57624582

>>57618332
>if you're not playing games

The whole point of amd is getting a similar/superior experience on games for half the price, if not what's the fucking point then? Any upgraded thinkpad makes for a good shitposting machine.

>> No.57624920

>>57622244
>>57622375
>open game
>run Windowed
>open task manager
>watch all 8+ threads load

That easy

>> No.57624972

>>57619254

Yeah, I hope AMD crashes and burns so we can have a stagnate technology advancement and high prices! HAHA!

>> No.57625084

>>57618221
single core is the only area that amd suffered from on the cpu side, and a 40% uptick from excavator would put it somewhere ballpark 2000, Intel, due to their severe problem of dick dragging is not too far ahead of amd if that's true, lying to investors is grounds to get your shit pushed in fucking hard so i doubt they lied.

Intel will still out single core amd, but not by enough to matter, but amd is looking to out multicore intel up to their 600$ cpu range for half price.

Now, here is speculation, they showed amd is clock for clock able to keep up with intel's 8 core, so if amd can overclock they are just underselling all the cpus to account for wild variations in the process (we have seen this in the past with the 7000 gpus) there is no reason these cpus can perform higher than intels 6 core but below a 8 core on average with silicon lottery playing a roll in if you get one that beats the stock 8 core.

>> No.57625095

>>57618234
Im holding off till a bit after release just so I know whats up with the skus.

>> No.57625170

>>57619564
the intel gen after zen will be the one they have to adjust prices, otherwise it looks FAR to much like they were fucking everyone in the ass and it would be horrible pr

>> No.57625297

>>57620606
Here is my conclusion.

1) amd is on par with intel ipc, this is the harder one to believe, but I will fully believe that the process amd is using can't clock up well, or its a true silicon lottery cpu, so the lower clock rate makes it only compareable to the 6 cores, however if you can oc it, it can match the 8 in certain loads

2) amd is well above the 6 core in some loads, but in its worse loads its ONLY comparable to the 6 core, another distinct possibility, mixed in with lower clocks, so its 'on par' statement is really just that, this is what you should expect.

3) amd has an ipc disadvantage in single core, but due to the cpus themselves being leaner, there is less to go through so the ht is better then intel's implementation. I still fully expect better ipc then amds best single core, but it will be a wider gap between amd and intel but because of intels dickdragging sense sandybridge, the gap is not wide enough to matter this time.

>> No.57625321

>>57618009

I already jumped ship from my 8350 in August to a 6700k. Will end myself if it proves to be a pretty good improvement.

>> No.57625337

>>57618289
first gen i5 were dog shit that even the 8350 beat out on single core, you are thinking the second gen with thoroughly beat the 8350 on single core but it was until the haswell series that amd had a multicore advantage on i7.

granted that advantage meant shit if you ever needed single core for anything.

>> No.57625366

>>57618063
fpbp

>> No.57625385

>>57621587
>>57621616
not vidya, a program that makes proper use of threads makes a cpu with threads kick the shit out of those without.

>> No.57625412

>>57619863
um... even amds current lineup beat out nehalems and westmeres single core, and this is with the shit ipc

>> No.57625425

>>57618209
4 cores and threaded if i remember right.

>> No.57625433

Give me 10bit hevc 60fps decoding and hdmi 2.0a and i'll buy the zen meme.

>> No.57625457

>>57618356
doesn't matter, if an application can use 8 threads or 16 threads who cares if one cpu has 12 and one has 16 and perform equally?

The only place this problem with crop up is with video games, and given what amd can currently do, and what their uptick should be, they are going to hit gpu bottlenecks across the board before cpu ones are even seen, especially with nvidia moving away from sli

>> No.57625464

>>57618886
we call those gpus.

>> No.57625503

>>57619283
site source, and if you are talking about the one person with a golden silicon, yea, its possible, however the process has such a high variance that a 480 could draw 160 watts, or it could draw less than 90 at the same clock rate. the amount of variance is stupidly high and gamers are the lowest binned parts, embeded get first run.

>> No.57625516

>>57618009
The Great Disappointment? Yes, I'm ready.

>> No.57625525

>>57618566
>>57618494
intel does not lose value, able to show me a 6 core sandy non xeon?

>>57618775
god knows most motherboards are not made to last as long as the cpus either.

>> No.57625581

I hope AMD goes bankrupt
they literally only exist so Intel can get away with its monopoly without looking like one

with no actual competition aside from a loose band of Pajeets who are perpetually 5 years behind, Intel pretty much does whatever the fuck they want

>> No.57625603

>>57625581
enjoy your 1.5ghz housefire dual cores for $1999 then

>> No.57625648

>>57625603
We're already experiencing what thats like because AMD contributes absolutely fucking nothing to stop it

Do you think 20% marketshare both cpu and gpu is anyone but retarded poorfags who dont know any better and their massive army of fanboys?

>> No.57625671

>>57625648
What do you think Zen is hoping to achieve?

>> No.57625716

>>57625671
Sandy Bridge IPC 5 years after it was released

>> No.57625747

>>57625716
>muh IPC
Anyone with half a brain would rather have an 8 core sandy bridge than a 4 core skylake chip

>> No.57625849

I'm still excited for Zen, but is an 8320e even a worthwhile upgrade from a 1090t? It's 80 and change with a mobo.

>> No.57625898

>>57625084
I don't quite follow. Are you trying to say that they're going to sell the CPUs for cheap because they may be shit but might be good?

>> No.57626166

>>57619036
amd ever confirm hbm for apus? If they do this it would fuck intels day up, along with nvidia if they still make any lower end chipsets.

>> No.57626184

>>57623640
>>57619052

clock for clock, amd is equivilant of intels 8 core, however due to new process, its massive silicon lottery if your cpu can match intels cpu, so matching 6 core is the minimum you can expect...

at least thats how i see it till we get more info.

>> No.57626186
File: 7 KB, 800x618, ffmpeg.png [View same] [iqdb] [saucenao] [google] [report]
57626186

>>57625747
>Anyone with half a brain would rather have an 8 core sandy bridge than a 4 core skylake chip
Yes, yes - buy into the cores meme. I'm sure ahmdahl's law is nothing to worry about; as long as you add more cores, nothing can do wrong

>> No.57626205

> P R I C I N G O N P A R

>N O T

>P E R F O R M A N C E ON P A R

F U C K E D

A M D
I S
F U C K E D

A B S O L U T E L Y
F U C K E D

>> No.57626231

>>57619970
skylake only beats it due to clock rate. most of intels ipc sense sandybridge is all in clock rate.

>> No.57626246

>>57619203
Ill likely wait a month to wrap my head around what the skus mean, if there are 2 8 core skus that are 350 and 500, I want to be damn sure I know the difference as the 500 is the most i was willing to pay for zen.

>> No.57626258

>>57626231
Higher clock has literally no affect whatsoever on IPC

>> No.57626270

>>57626231
Increasing clock rate doesn't increase ipc since ipc stands for instructions PER CLOCK. Skylake has the same ipc as broadwell-e.

>> No.57626272

>>57626258
*effect

>> No.57626281

>>57620379
here is what happened
we got a bench result from china that showed stock.
we than got a bench that showed 1500~ clock, which we know is possible on binned gpus.

we then had retards flood the threads, most likely nvidiots posting retarded rumors and when it failed to match sli titanx, it was deemed shit.

All in all, fantastic gpu, at 200$ and 240$ its hard to go wrong, its within spitting distance of 1060 and when vulcan or dx12 is used above it.

>> No.57626294
File: 118 KB, 436x594, Ben-Affleck-Beard-Gone.jpg [View same] [iqdb] [saucenao] [google] [report]
57626294

>>57626258
>affect
<<<

also i dont think that's what he was saying at all

>> No.57626305

>>57626294
Are you retarded?

>> No.57626308

>>57626294
>most of intels ipc sense sandybridge is all in clock rate

>> No.57626318

>>57620581
>(and nvidia owns 85% of the remaining 20%)
they are down to 70% or less now, amd is on the uptick for market share.

>> No.57626343

>>57626305
yes I have weapons grade class 3 assburgers and i'm struggling to live with it on a daily basis, thanks for asking :^)

>> No.57626400

>>57619152
This is also why their CPUs haven't had any real improvements since sandybridge. Honestly the only companies that hire quality engineers are Nvidia which is like 90% asians (not the brown ones) and apple. Pretty sure apple will be the next big CPU company.

>> No.57626426

>>57619196
AMD has been hiring a ton of pajeets lately. Not sure wtf is going on so things aren't looking good for the future.

>> No.57626437

>>57620525
the problem is why would you buy amd over intel if they make it to high. Imagine everything below 8 cores as 'the silicon fucked up but it useable so put it out' cpus, but these cpus are still great values so they don't keep the price TOO low.

>> No.57626565

>>57625898
What im saying is they are selling the cpus based on lowest common denominator.

lets say this,

you have an architecture that can go up to 4.8ghz, but the average worst chip that's functional only manages to get 3.5 reliably so they bin those lower ones to 3.5 bit a bin higher ones at 3.6 and make a sku that is 4.0, these are the guaranteed clock rates.

take a look at https://siliconlottery.com/

you can see what their numbers are on the clock rates they are able to hit. an example

Intel 6700K 95% of them are able to hit 4.6 ghz,
Intel 6700K 3% of them can hit 4.9 ghz
Intel 6700K 19% can hit 4.8 ghz

but the stock value of this is 4.0 ghz

they make more money this way than individually testing.
considering the non k version is 3.4 there is a damn good chance these chips are identical but the non k version just can not clock up

>> No.57626581

>>57626186
Not sure about you, But I do more than one thing on my computer at a time, and would love the head room more cores provides for partitioning resources out.

>> No.57626594

>>57626565
Thanks for the explanation! Interesting stuff. Not even being a smartass

>> No.57626613
File: 128 KB, 480x352, freshprince.png [View same] [iqdb] [saucenao] [google] [report]
57626613

>>57626294
See >>57626272
And he's saying Intel's newer microarchitectures have higher IPC mainly due to higher clock speeds, which is of course wrong.
Also
>i
>dont

>> No.57626724

>>57626594
Thanks, I mean we call intel jews all the time, Just knowing that the non oc chips may be non oc for a reason is intresting.

It may even be a good thing

Imagine if intel put out only one sku, the 6700 and clocked it at 3.4, you get it because some CAN clock up to 4.9, but yours only clocks to 3.6

Personally, If this is the case, im 100% ok with paying, what 35$ more for one that is almost guaranteed to hit 4.5-4.6

hell, look at the boost rates
4.0 GHz 3.9 GHz 3.7 GHz
single dual quad
one core on each chip seems to guarantee 4ghz, and 2 are not that far down, but 4, that's .3 that .3 is likely the maximum clock that every working 6700 could sustain.

Now, you have another aspect, the power. A very important factor for some people is the power draw, and amd said 95, the clocks are likely just to keep it in that envelope while everything is going 100% as we see with the 480, there are 160 watt cards, and there are 90 at the same clock, so the normal card they put out was the assumed 150 watt draw, amd fucked up as some drew more watts, so a voltage lowering fixed it and increased clocks overall.

When you start looking at specs like they were based entirely on rng, the whole cpu gpu market gets a bit more interesting.

>> No.57627055
File: 598 KB, 480x360, tmp_13738-1477815415105158804560.webm [View same] [iqdb] [saucenao] [google] [report]
57627055

>build up hype
>spread rumors of mind blowing performance
>expectations are unmet in launch day
>massive consumer backlash against AMD

Gee, I wonder who's behind this scheme.

>> No.57627171

>>57627055
fanboys opposed to amd mostly, some fanboys on the inside too, but largely they have 1 shown benchmark and rumors to base shit on.

>> No.57627232

>>57626724
That is really interesting when you think about it like that. I always figured they locked the chips to get more money, but the idea that they lock them because they're borderline rejects is incredibly interesting. Just when I think I've learned everything I want or need to about hardware, something like this comes out of left field and blows my mind. I think I might actually enjoy studying for my A+ cert

>> No.57627339

>>57627055

fake hype has been AMD's go-to marketing scheme for literally a decade now pajeet, no need to act russian about it

>> No.57627400

>>57627232
Oh, make no mistake, the higher cost is just to make more money, they could call the worse i7's a lower number and this comparison would almost be impossible without detailed schematics or microscope slides but they don't, the point made is at least to me 35$ is reasonable to ask if they are separate singing the rejects from the good ones in house rather then a lump sum of all 6700's and have a pr shit storm with a number of people saying the 6700 is a crap shoot rather than everyone telling you how much higher they can clock a 6700k.

At least this is the way I see it and would handle the situation if I was intel.

Always keep in mine,
1) they want to maximize profits
2) they want to minimize shit storm
3) if they have no true competitor, they will fuck you hard because you have no choice, see their pricing of 8 cores and 10 cores as an example.

If amd is at their 6 core lineup, that price is going to go down next cpu lineup
if they hare halfway between a 6 core and 8 core, the 8 core will come down a bit,
if they can touch an 8 core, you are looking at intel's new 500$ buy in cpu

As it stands, the only thing an i3 has over a 4 core 8 thread amd will likely be the single core performance due to clock and an igpu, something amd's apus will kill, This may be the last time i3's are dual core, or intel could sell them cheaper and undercut amds price, I cant personally see i3's as a major market for intel, at least on the desktop, I see it as more of a 'we have a low end sku too'

We may see a shift for the i5 and i7 line too, possibly with i5s getting threading and i7's starting at 6 core, or i5 getting 6 cores and no threading and i7's starting at 6 and threading.

Regardless of what zen does, its going to be a hell of a shake up.

>> No.57627406

>>57626318

no they aren't. steam hardware survey shows that NVIDIA has been consistently growing it's market share for ~18 months straight.

>> No.57627436

>>57627406
gaming = only use for gpu

Nvidia burnt a fuck ton of bridges and those decisions are coming home when people in charge remember how much of a pain in the ass they were to work with.

As far as gaming goes, if amd keeps their driver updates going like they currently are, the next gpu revision will see a sizeable chunk of the gamer section going to amd too.

>> No.57627658

>>57619475
tfw I wait 4+ hours for a single-threaded formal verification model build

>> No.57627829

>>57618595

>up to 2.8x
>up to
>2

retard actually took marketing materials seriously

>> No.57627878

>>57618806

>Nvidia uses AMD IP and their own IP to make super efficient cheap low-end cards

>Nvidia GTX 1280 Founders Edition
>Nvidia GTX 1280 Preorder Edition
>Nvidia GTX 1280 Early Bird Edition
>Nvidia GTX 1280 Harder Daddy Edition

>> No.57628030

>>57619968
Compile chromium and let me know how many days it takes

>> No.57628034

>>57618141
kek

>> No.57628223

>>57627400
Interesting stuff. I thought intel was trying to delay Cannonlake further? I guess we'll have to see what tricks lies in zen. Hopefully it won't be a disappointment.

>> No.57628780

>>57626581
>But I do more than one thing on my computer at a time
Yes, you need an 8 core machine to have a word processor and a facebook tab open at the same time.

>> No.57628816

>>57628780
rendering a video, transcoding, may have a movie or something play in the background for noise on top of playing a game while i'm waiting for shit to be done.

then for hobby shit, render 3d

an 8 core allows me to partition half the cpu to do that shit, and not effect my 'down time' activities.

>> No.57629004

>>57628816
>muh video transcoding
I thought you were going to present me with something actually realtime/intensive that required all of these cores simultaneously. Video transcoding isn't an argument for multicore processors unless you actually need it to be realtime. Just let it run in the background, where's the problem?

>> No.57629007

>>57628816
>an 8 core allows me to partition half the cpu to do that shit, and not effect my 'down time' activities.
As if you can't do that on a dual core too

>> No.57629049

Is it gonna have 125W TDP as usual AMD poocessors?
Nah thanks, I will wait for Cannonlake masterrace.

>> No.57629062

>>57619269

They seemingly post any half true rumor as if they have some ultimate insider source. They get it wrong often. The best was when they fell for a lie and posted faked pictures of upcoming high end AM4 motherboards. They took it down but other shit tech sites took wccf's story and ran with it:

http://www.tweaktown.com/news/52796/amds-next-gen-am4-socket-pictured-asus-rog-crosshair-vi-impact/index.html

>> No.57629072

>>57629007
not nearly as well. hell, you can see benchtable differences between 4 6 and 8 core, at least in the context of videogames, because on a 6 core, lets say a process starts up in the background, you feel that a hell of alot more on even a 6 core then you do 8 because the load is far more balanced.

>>57629004
you ignored 3d rendering, not sure if you know, but render times of well over 90 hours are not uncommon, and I'm not making money so i could never justify paying a render farm or buying a renderer license.

>> No.57629396

>>57629072
>not nearly as well. hell, you can see benchtable differences between 4 6 and 8 core, at least in the context of videogames, because on a 6 core, lets say a process starts up in the background, you feel that a hell of alot more on even a 6 core then you do 8 because the load is far more balanced.
Sounds like your scheduler is shit m8. What OS are you on?

I can load my CPU 100% with background jobs without affecting the responsiveness of interactive applications at all. That's why you give the background jobs low priority, so they don't steal CPU time from your foreground tasks.

l2nice

>you ignored 3d rendering, not sure if you know, but render times of well over 90 hours are not uncommon, and I'm not making money so i could never justify paying a render farm or buying a renderer license.
If you have 90 hour render times and you need to cut them down, you're not going to want to get a meme 6-core desktop CPU. At that point you want to get a dedicated rendering server or something.

>> No.57629447

>>57629396
win 7 and soon a dual boot of 7 and 10

true, a render server would be good, but that also costs significantly more money, and I don't really have enough to dedicate to just that. One big computer upgrade, 1 big computer upgrade that should function for work and leisure.

If I get good enough to make my hobbies my job, I could justify the extra hardware expense.

>> No.57629557

>>57629447
>win 7 and soon a dual boot of 7 and 10
Oh, no wonder. The windows scaler is atrocious. Fine, if you need to bruteforce your way through shitty software, be my guest. Your money.

>true, a render server would be good, but that also costs significantly more money, and I don't really have enough to dedicate to just that. One big computer upgrade, 1 big computer upgrade that should function for work and leisure.
Nah, you can pick up a used 32-thread xeon server for less than the price of a new 6-core intel CPU.

>> No.57629605

>on par with high end i7 skylake
Why do these people keep hyping this shit to the moon? This is even stupider than when people were claiming the rx 480 would be on par with the 1070.
The best we can assume with what we currently now is performance on par with sandy bridge. Which is still respectable but you guys need to stop fucking fooling yourselves.
And if they had something that competed with the latest and greatest from intel buy with MOAR COARS they sure as shit would price it more accordingly.

>> No.57629611

>>57629557
got a link to something? Been looking stuff like that up for a while now as an alternative to an iseries or zen if zen turned out bad, but never seen an intel board with more then 2 sockets and the cpus were old sandy bridge xeons, together they were 24 threads and less powerful then one current 6 core and the only 8 cores i know if that could do that would at best be on par, perhaps slightly ahead, anything that goes to 3ghz in the xeon line costs ballpark i7 equivalent.

>> No.57629624

>>57629605
because clock for clock amd showed it matched the intel 8 core. One bench sure, but it showed something at the very least.

However the current thought is amd, due to process, can't clock it up like intel can.

>> No.57629731

>>57618009

I'll believe it when I see it.

I want to support the underdog. I want them to be competitive so intel and Nvidia will stop the Jewing.

I just don't believe it will happen is all.

>> No.57629884

>>57629611
>got a link to something? Been looking stuff like that up for a while now as an alternative to an iseries or zen if zen turned out bad, but never seen an intel board with more then 2 sockets and the cpus were old sandy bridge xeons, together they were 24 threads and less powerful then one current 6 core and the only 8 cores i know if that could do that would at best be on par, perhaps slightly ahead, anything that goes to 3ghz in the xeon line costs ballpark i7 equivalent.
e.g. http://www.natex.us/Intel-S2600CP2J-Dual-E5-2670-128Gb-Kit-p/s2600cp-cpu-128gb-12800.htm

16 cores of SB-E, 128 GB of ECC RAM and a server-grade mainboard to boot for under $500. Really can't beat that.

>> No.57630052

>>57629884
Thank you, I have an option now, still going to wait and see zen first, I'm guessing that it won't be as good as this result, but if it can match intel's 8 core, it doesn't put it far behind if at all.

While I do trust the cpus and ram to work, I have had very little luck when it came to longevity with anything that had capacitors on them.

This will be my option of zen fails though, seeing as my other option is around 600$ for just cpu and motherboard.

>> No.57630094

>>57630052
>While I do trust the cpus and ram to work, I have had very little luck when it came to longevity with anything that had capacitors on them.
I wouldn't count on that offer still being up by then, mind. Seems like the E5-2670 price is already going back up as the supply of old used servers is coming to a halt.

>> No.57630115

>>57626186
>Anyone with half a brain would rather have an 8 core sandy bridge than a 4 core skylake chip
>>57629072
>you ignored 3d rendering, not sure if you know, but render times of well over 90 hours are not uncommon,

You make it sound like 90+ hour render times are the norm for “anybody with half a brain”

>> No.57630185

>>57630094
even at the normal price that's a good deal, would suck to pay more, but i have a finite amount of money and a capacitors dieing is beyond my ability to fix even if they did no damage. I'm just imagining buying that, the motherboard dieing, then needing to replace it with either something used that could die again, or new old stock that would cost a fuck load.

I may come into a bit of money soon, so a secondary computer for rendering may not be out of the question if they are this cheap... Have to ask, you know anything about controlling computers remotely in windows on the same network or video rendering applications that would offload the work form one computer to another?

If shit goes my way I may have money if its sub 1000$

>>57630115
not the same person, i was only the second quote. However, I thought it was common knowledge that render farms are used in any kind of 3d work due to the retarded long time it takes to render single frames, and if normal people didn't know renderfarm by name, at least that 'super computers' were used. I mean, one movie quality frame takes something like 2-3 weeks to render off a normal workstation.

>> No.57630264
File: 24 KB, 496x499, 1479447058381.jpg [View same] [iqdb] [saucenao] [google] [report]
57630264

>>57618009
>Are you ready for The Great Uncucking?
I think you mean
>are you ready to be cucked again?

>> No.57630272

>>57618266
Did you forget what happened with Bulldozer?

>> No.57630304

>>57619168
>he plays games on his computer
fucking underage
>>>/v/

>> No.57630325

>>57619348
>that pic

kek

>> No.57631595

>>57618009
sizable investor here

zen is not hype
am excited

we're getting sued at the moment but it's no biggy.

>> No.57631816

>>57630272
I don't think AMD is screwing around this time. Because frankly, If Zen is shit then AMD is going to lose the CPU market to Intel. Which could literally kill the company.

>> No.57631836

>>57618063
absolutely btfo

>> No.57631912

>>57630272
Bulldozer was a more ambitious design and we didn't really know as much about what it brings as we do with Zen. The modular core design was uncharted waters while Zen appears to be more similar to K8/10 but with general improvements, new instruction set extensions, and SMT

>> No.57632093

>>57618599
If they give you 8 for the price of 6 it's not a bad deal.

>> No.57632631

>>57631912
Isn't zen taking the modular design to the extreme now. Each module is now a 4core block, which is laid out similarly to the outgoing cat cores, two of these modules on a SR chip, one on the upcoming APUs paired with a large gpu block. There are rumours of those monster apus. The 32core with 4xSR, the upcoming 48core which is either 6x8core or a 4x new12core blocks.(I know these are not single die).

Any way the point is the modular design is coming in full force, it's the basis of AMD's semi-modular business which has been successful in winning the xbox/ps contracts. It's also the reason I think those RTG sell off rumours are hogwash.

>> No.57632881

>>57627436
nintendo was the one they didn't work with before
haha, I guess intercorporation reputation doesn't exist
switch is dead on arrival

>> No.57632910

>>57629049
I have news for you.

Cannonlake going to be laptops only, next big desktop is cofeelake in 2018, finally 6 cores.

>> No.57632954

>>57632631
What I meant was that the way the "cores" shared resources was different. It seems like Bulldozer was a more space-conscious design, a "module" was like two cores in around the same space as a single core. They could in theory perform like 2 cores on integer math but floating point math they'd perform like 1. With Zen each "core" is independent aside from the caches where separating them likely wouldn't do any good. In that way Zen is more like K8/K10 at least superficially.

>> No.57634047

>>57630185
>Have to ask, you know anything about controlling computers remotely in windows on the same network or video rendering applications that would offload the work form one computer to another?
All I know is that this is usually done with program-specific clients. Adobe and Blender have their own shit here, and I'm sure Autodesk does too.

>>
Name (leave empty)
Comment (leave empty)
Name
E-mail
Subject
Comment
Password [?]Password used for file deletion.
Captcha
Action