>RIP INTEL

File: kaby_baby.png (30KB, 610x459px) Image search: [Google]

30KB, 610x459px

That's nice buddy, but single-threaded performance is still the name of the game.
Kaby Lake is gonna set new records in single-threaded benchmarks, while it matches a 6-core in multi-threaded with only 4 cores.

File: you_asked_for_this.jpg (46KB, 600x515px) Image search: [Google]

46KB, 600x515px

>>57715155
>Kaby Lake is gonna set new records in single-threaded benchmarks

>>57715003
$150 non-used i7 when?

>>57715185
Intel blown the fuck out.

>>57715185
wtf I hate Intel now

>>57715003
Not to be a intel shill, i really want amd to do good this time, i really do but i have little faith they'll deliver. there HAS to be some catch.

Like they marketed their A10 Cpus to be 12 CORES OF PERFORMANCE!

>>57715003
>3.5
Thanks for making me not care AMD.

>>57715185
DELET

>>57715003
Are there other differences between the two SR7 variants or are they really making you pay $150 for 200MHz more?

>>57715185
Comparing them at the same clock speed makes no sense when Kabylake can reach much higher clock speeds at the same power consumption than Skylake.

>>57715185
1% is a new record you retard.

>>57716042
>Are there other differences between the two SR7 variants or are they really making you pay $150 for 200MHz more?
So far, this is all just rumors and possibly fake leaks till confirmed, so grain of fucking salt

the higher cost one has more oc features or is higher binned then the 350$ one

that's all we got to go on, fuck knows what that means, its why if that 500$ one is real, im holding off for a few weeks to see benches and people smarter than me looking into it.

as it stands, I will be getting amd.

>>57716082
But that's what IPC means, genius.

>>57716082
clock equalized, what this shows is that almost all the improvement the chip could have in in overclocking, and seeing as damn near every single 6700k hits 4.6ghz, you get almost fucking nothing in an 'upgrade'

>>57716283
same with 4790k that clocks to 4.8 with ease.

I want zen to come close to the hype if only to ease intels price gouging

But lets be real, anyone with a recent intel quad will be fine for the next decade

AYYMD HOUSEFIRES IS FINISHED & BANKRUPT

>>57716312
Heard alot of people not able to push that passed 4.3, 4.8 is a bit on the high end, the 6700 only 19% manage that speed.

>>57716449
There are a fuck load of people waiting on cheap 8 core, who were waiting on intel for cheap 6 core, how many times has it been thought intel would push the i5 to i3, the i7 to i5 and the i7e to i7 and boost the i7e line?

fuck we thought they would do it when they went to 8 core, then we thought it was going to happen the gen after, and it still has not fuckin happened.

make no mistake that extra load balancing head room 8 cores gives you is VERY enticing, just not for 1000$, but 350$ hard to say no.

>>57715003
Price is way too high. I though AMD was pushing 6 core+ CPUs to the mainstream big time. Gaymen and enthusiasts will probably skip these considering the single core performance will be bad compared to Kaby Lake.

>>57715003
>$500
lmao, no one will buy it.

>>57715003
6 Core variant better clock higher and OC good or imma go with Intel.

>>57715003
>Zen SR7
>Turbo Clock 3.5GHz
>499

HAHAHAHAHAHHAHAHAHAH AMD DEAD

>>57716996
With these clock speeds and inferior single core performance AMD won't be able to do much. I've been waiting for almost 10 years for a 6 core+ mainstream CPU that isn't shit. With the rummored clock speeds of Zen I geuss it'll be another 2-4 years before something happens. Maybe around the time of Zen++.

>>57715185
DELET THIS

>>57715185
Sad... it's like they don't want my money.

>>57716283
The i5-7600K easily hits 6 GHz on air cooling.

>>57715003
>$150 more for 200 mhz more
kek

>>57717106
What are the predictions for single core and what is the bare minimum for AMD avoid bankruptcy?

>>57715003
Can you sauce me up?

>>57717069
>>57717051
Wait and see, no one knows how they will perform.

>>57717241
People are more afraid of how much they'll cost outside the USA.

>>57715185
LMAO
holy fuck

>>57717201
The bear minimum would be damn near Skylake performance if they end up sticking those prices on their CPUs. And even then they need to have high clock speeds and OC very well. Which won't happen unless they surprise everyone.

>>57717069
>i7 6900K
>Turbo 3.7Ghz
>1099.99

>>57717306
hwbot.org/hardware/processor/core_i7_6900k/

Oh wow Intel overclocks like a champion. Who would have known.

>>57717337
Okay? What does this have to do with anything? Do you believe AMD CPU don't overclock? Or do you have data that Zen CPU can't overclock?

Are you by any chance a retard?

>>57717207
Wouldn't be surprised if OP made that pic himself

>>57715185

pls moot delet

>>57715003
>Those tiny, tiny, tiny caches for $499

ayy lmao who the fuck is gonna buy this shit

>>57717369
Not him, but if most rumours say it'll go up to 3.5 then it will overclock like a piece of shit.

>>57717201
To avoid bankruptcy it just has to be better than bulldozer. If it can't compete they will just drop prices like they did in the past and manage to scrape by.

>>57715003

>3ghz with sandy bridge ipc

zen is doa before it even launches lmfao

>>57715185
Eh, I'm still going to buy it. Not falling for the amd meme again.

>>57717425
The 6900K turbo is 3.7Ghz, base is 3.2Ghz

It doesn't correlate to overclock ability.

Base/turbo clock has nothing to do with OCability.

>>57717419
Tiny caches? Up to Intel's 500$ lineup, compared to what?

>>57717469
>6700k is Sandy Bridge
Totally...

Seriously thou, clock-for-clock Zen really does slightly outperform Broadwell E.

>>57717491
Then I hope that applies for Zen as well.

>>57717522
>Seriously thou, clock-for-clock Zen really does slightly outperform Broadwell E.

I assume you have an AMD Zen CPU in your computer then?

>>57717512
Guy is probably looking at the L2 cache and thinking that it's Zen's L3 cache.

For comparison the 6700k has a 20MB L3 cache at $1100 while the $500 and $350 Zen chips have 16MB of L3 cache.

>>57715003
Are the Summit Ridge chips and boards going to expose the 10GbE MACs?

Even if they only have 2x DDR4 channels, these prices could be a steal if the IO and SB are right.

>>57717560
I'm going by when AMD did their clock-for-clock comparison between a Zen chip and the 6700k.

>>57717512
>>57717572
I think he's trolling since Intel does 2.5MB L3/core while Zen is only 2 MB L3/core.

>>57717201
the predictions before we had anything to go on was a 40% uptick from excavator that put it somewhere around sandy/ivy bridge

then amd said they were getting 40%+ but never disclosed how much

they showed off multicore going toe to toe with broadwell e.

Now you have one of two ways to see this, either amd who is using the same type of threading intel is, found a way to do it a fuck load better, OR amd is clock for clock on par with intel in non fpu reliant synthetic workloads.

We wont know till it hits, but they are likely going to be able to undercut intel and unless intel has 8 channel memory (what amds high end server shit is thought to have), amd will likely pick up a good deal of clients, and if the 80% of the workloads they are just as good is true, they may get a good portion of the server space. even a small chunk of this is multi billions in profits.

>>57717168
And ill get laid in my life time because my right hand will become sapient.

>>57717051
you do notice that 350$ one right?

>>57717040
that 1% uptick really makes all the difference, not 2 more cores of headroom when you have all the single core you need.

>>57717053
legitimate question, why? I mean some games may not hit 144 unless you oc, but that would only be one game in 100, most are gpu bottlenecked now even at stock speeds of shitty i5's

The only reason I can think ot overclock would be applications better served to get a more core cpu to begin with.

>>57717534
It should, but the process currently has a HIGH margin for binning, look at polaris, some are 160watt some are 90 watt for the same performance, If they don't bin the shit out of zen for these two versions, you will see some that cant go higher then 3.4 (im guessing this is the lowest clock that a non fucked up chip will reach) and you will see them top out at 4.0

wait... look at intels 8 core, the cpus barely go over 4.3, ill give intel that lead as they have the process advantage.

>>57717522
>Seriously thou, clock-for-clock Zen really does slightly outperform Broadwell E.

says who? the only evidence of that we've seen is from amd's marketing department, who have a track record of fake hype and lying outright about performance. leaks from third parties has shown zen to be on par with sandy/ivy bridge ipc wise. if it really clocks poorly as well then it's not gonna be a competitive product.

>>57715003
dont worry intel is gonna lobby against amd again

File: Comparison.png (17KB, 517x204px) Image search: [Google]

17KB, 517x204px

>>57717522
>>57717587
Such amazing performance using only Blender™

www.anandtech.com/show/10585/unpacking-amds-zen-benchmark-is-zen-actually-2-faster-than-broadwell

>>57717285
you do see 8 core is 350 right? and the quad core is 150$

granted I don't believe shit on this list, but 150 for a 4 core 8 thread cpu...

>>57717621
mock blender all you want, intel recommends it as a benchmark for their cpus, thats why it was used.

>>57717603
My next build will last me 5 years or so, that's why I want Zen to OC well, so when it stagnant in performance imma just OC it.

>>57715185
I don't get it. It would still be a new record.

>>57715185

delet

>>57717621
>Benchmark recommended by Intel themselves
>"Must be a bad benchmark chosen specifically to favour AMD"
genius.jpg

>>57717621
So zen has a higher ipc than skylake?

>>57715185
>Intel gives us better IPC ever year even though they can simply not increase IPC and instead focus on dumb shit like power consumption
>AMD hasn't released a competitive CPU the last 10 years

>>57715003

STOP OVERHYPING THIS FUCKING TRASH YOU FUCKING AMDRONES

>>57717629
Waitfags have been waiting for something other than 4 cores. And that 250$ Zen 6 core will cost alot more outsde the US.

>>57717881
>outside the US
Then who gives a fuck?

>>57717912
Majority of the market. Not that you'd know considering you buy your life at walmarts or something.

>>57717912
Everyone else that isn't a neckbeard living in moma's basement.

>>57716082
Stock speed is irrelevant, what matters is if it can OC higher.

>>57715003
Those stock clocks are pretty low, this shit better OC to 4.5GHz+ with good cooling.

File: 1479969997783.jpg (864KB, 1440x1449px) Image search: [Google]

864KB, 1440x1449px

>>57717936
>>57717950
Enjoy your extremely high prices eurocucks

>Walmart
Damn right I'll keep buying cheap shit because I can

the fact that intel is about to bring their 6 cores to the "consumers" is enough to show that they are shitting their beds...
keller never failed when he desinged a cpu or gpu... every single one of them was a hit

>>57718038
I'll enjoy free healthcare, top of the line hospitals in the case I stub my toe, first-world infrastructure and not being a fatass like 99% americans.

Feels good not being American, err I mean Mexican.

>>57718038
asdfasdfasdf

>>57716312
>4790k that clocks to 4.8 with ease
Nah, not with ease. Maybe "with ease" if you delid it and have a custom water loop to push the voltage, but you have to be lucky as fuck to get a 4790K to 4.8 with <1.3V. Even 1.3V is going to be difficult to keep in check with air cooling and AVX2 load.

>>57718082
>99%
Everything you just said, you pulled out of your ass, you cunt.

>>57717241
>Wait and see, no one knows how they will perform.
It's quite safe to assume that they won't match Intel for IPC, if they can't even match Intel for clock it's going to be another "get this if you render things" CPU.

>>57718038

Fuck you, Tess.

File: serveimage.jpg (60KB, 800x378px) Image search: [Google]

60KB, 800x378px

I wonder if Zen-cpus will be soldered and if they do support ECC. Had enough with jewtel's bubblegum TIM.

>>57718122
Well thats good for me, I render things.

>>57718108
Make sure you treat your burn at the nearest hospital jose. Oh and make sure to bring $500000 to cover the hospital bills.

>>57718038
Son of a bitch

>>57717176
Fucking kek. Worse than intel.

>>57718122
AMD claims that it will have higher ipc than skylake >>57717621

>>57715003
>Zen SR3
>L2 Cache 2MB
>L3 Cache 8MB
Double L2 cache than i5
THANK YOU BASED AMD

File: 1480093420510.png (497KB, 579x640px) Image search: [Google]

497KB, 579x640px

First we take CPU and GPU market, then we take Palestine back. Thank you for your support, Brothers, the intel Jew dies tonight!

File: _20161125_172929.jpg (36KB, 720x1088px) Image search: [Google]

36KB, 720x1088px

>>57718394
Still better than Jewtel.

>>57718394
Allahu akbar.

Its gonna be overhyped trash like everything AMD releases

Cant wait for the tears to start flowing

File: cgjgyjfgyjfyjffh.png (676KB, 945x1005px) Image search: [Google]

676KB, 945x1005px

>>57718442
you make me want to throw up.

Enjoy hell.

>>57718394
I'm glad I support Intel Jews instead of AMD ISIS.

>>57718472

de·ni·al
dəˈnīəl/
noun
noun: denial

the action of declaring something to be untrue.
"she shook her head in denial"
synonyms: contradiction, refutation, rebuttal, repudiation, disclaimer; More
negation, dissent
"the reports met with a denial"
the refusal of something requested or desired.
"the denial of insurance to people with certain medical conditions"
synonyms: refusal, withholding; More
rejection, rebuff, repulse, veto, turndown;
formaldeclination
"the denial of insurance to certain people"
a statement that something is not true.
plural noun: denials
"official denials"

>>57718243
That hardly claims anything that you can actually trust (see the anandtech article) and they've underclocked the Intel CPU. I don't know what to believe by seeing that benchmark.

Does it mean Zen IPC is superior to Broadwell-E across the board? I don't know.
Did they choose 3GHz and underclocked the 6900k because Zen can't really go higher? I don't know.
Did they optimize the Blender build to run better on Zen? I don't know.
Did they choose the Blender workload in such a way that it favors Zen? I don't know.
Why did they show a single 50s benchmark, with a single workload? I don't know.
What sort of memory settings have they used? I don't know.

See? All the benchmark does is tell you that Zen isn't completely garbage, which is nice but doesn't actually tell you how it actually competes with Intel, in the real world. All I can think is that AMD must have had a good reason to show that and exactly that and they must've had a good reason to say as much as they said and nothing more. We've seen that test because Zen did well, I'm sure they wouldn't parade a test in which Intel wins in any case, as such there is very little actual information to take away from that demo.

Don't get me wrong though, I'm on a 4.7GHz 4790K. I would absolutely love to be able to switch to a Zen 8 core CPU which can OC far enough to reach superior (or at least equivalent) single-threaded performance. I hope they can do it, but I have no reason to believe so far.

>>57718557
Thats going to be AMDrones when they find out Kaby Lake utterly BTFOs this poo in CPU

>>57718561
Updooted.

>>57718520
Yeah whatever you Israel cocksucker
keep buying the latest 5% boost in IPC every year, and having to buy a new motherboard with it

>>57718038
Those pants look trashy.

>>57718561
I'd also have to point out that what they've been using for comparisons have been engineering samples, not the final product. This doesn't just apply to the CPU itself, it also applies to the motherboard being used in the benchmark.

So we know it's not going to be shit, but it could improve by a decent margin when everything is in production-ready status.

File: 1477855683432.gif (529KB, 625x626px) Image search: [Google]

529KB, 625x626px

>>57718520
>being cucked by them this much into submision

>>57718561
>4.7GHz
Intel's own 8 cores don't reach those freqs.

File: 1479092644803.png (157KB, 516x440px) Image search: [Google]

157KB, 516x440px

>1 more month 'till CES 2017

File: like a boss.webm (3MB, 1194x640px) Image search: [Google]

3MB, 1194x640px

>>57715003
>no source

>>57718713
But that's haswell, skylake has better ipc and can match 4.7GHz haswell at lower clock speed

>>57718711
>>57718619
I know you two don't have any friends. kys now please. No one will miss you.

>>57718713
No, but the upgrades I'd be hypothetically looking at are either Zen or Skylake-X (or w/e they're going to call it). Skylake at 4.4-4.5 should match Haswell at 4.7 in terms of single-threaded performance even if it won't hit 4.7. I don't need to see the 4.7GHz number on Zen either - if it has superior IPC to Haswell, which we don't actually know yet.

>>57718690
Yeah, they're ES, so things should improve. I imagine they'll run higher than 3GHz, but that's still a pretty low clock freq, I'm worried about Zen being able to reach like ~4.5GHz (OC of course). I imagine IPC will be similar-ish to Haswell at least, maybe fast DDR4 will also help by a few %, but I still find it implausible to think that Zen at say 4GHz could match Haswell at 4.7 in terms of single-threaded performance. All I can do is hope, but if AMD manages to do it I probably won't even mind dumping $500+ on a Zen CPU.

File: rape face.jpg (65KB, 500x500px) Image search: [Google]

65KB, 500x500px

>>57718763
Waiting is all we can do anon...

>>57718851
How chips clock is generally the biggest difference between what an engineering sample and the finished product. The purpose of engineering samples is after all verifying that the chip works correctly, not to see how fast it is or how high you can take it OC-wise.

Seriously, judging clock-ability based on an engineering sample of a CPU is like judging the build quality of a car based on a pre-production car meant to test the mechanicals.

>>57718975
I didn't judge anything, I said it isn't looking promising. For instance if AMD's 8 core ES ran at 4GHz, I'd be very impressed with it, as such it runs at 3GHz and I'm not. Production CPUs could still clock much higher of course, but AMD haven't done much to inspire confidence on their CPU side for a long fucking time now and lately they've been kind of slipping with their GPUs too.

>>57718038
I'm not saying I would, but I totally would.

File: Childish Swarzenigger.jpg (16KB, 300x400px) Image search: [Google]

16KB, 300x400px

>>57718872

File: 1245790--.gif (251KB, 500x343px) Image search: [Google]

251KB, 500x343px

>>57719784

>>57718763
there's gonna be some good shit let me tell ya

File: MediaTek-MT6797-Helio-X20-deca-core-benchmark-scores.jpg (85KB, 700x717px) Image search: [Google]

85KB, 700x717px

>>57715155
>geekbench
Why do people keep using this shit?

Hey guyzz look at this only 4w mobile phone processors that performs half as well as the i5 6600!

>>57715185
kek

what does a ZEN SR3 compare to? $150 4core seems nice for a budget build

>>57720019
nobody knows. You'll have to wait 1/2 months at least.

>>57720019
Just a wild guess, but
>4c
>8t

Probably an i7 3770?
Was going to say 2600, but then I remembered its got ddr4.

>6c 12t for $250
I hope they weren't lying about IPC performance. I also hope it can reach 4.2+GHz.

Zen will not even outperform Sandybridge.
Screencap this!

>>57720191
>screencapping things that won't happen
fuck off

>>57715003
>3.5ghz boost clock max
>3.5
kek
>not even 4.ghz
dead on arrival

>>57715003
If AMD did this they would gone bankrupt.

>>57720219
The i7-2600 will be better than the Zen-SR3 in every benchmark.
How about now ?

>>57720223
Retard

>>57718038
Cute face. I would fuck her any given day.

What are AMDs pricing habits outside if USA? What are thry gona be in Britbongislamistanabad dollars?

>>57720254
Are you an idiot?
Even if it's still only got vishera ipc(which it fucking doesn't) it'll destroy sandy bridge completely.

Zen has ddr4.

>>57715312
You can just overclock it to whatever frequency you want. I'm sure it wouldn't be too hard to hit 4...

>>57720479
Vishera is more than 2 times slower than sandy bridge

File: AMD_Comrade.png (336KB, 640x417px) Image search: [Google]

336KB, 640x417px

>>57718545
I'm note sure isis fits AMD's image...

Benchmarks when? Will some journalists or whatever get samples before release?

>>57718082
>top of the line hospitals
>state managed funding
One of these things is not like the other

>>57720479

ddr4 doesn't mean shit for performance unless you're using massively overpriced ddr4 4000 kits, jedec's ddr4 spec is literally slower than relatively cheap ddr3 1866 and 2133 kits.

>>57720576

isis is authoritarian and socialist as fuck m8

>>57715003
>still can't match a i7

>>57720890
proof?

>>57717864
Higher IPC or better memory coherence between cores.

I hate AMD I hope ZEN isn't good.

>>57718038
Dollar Tree

>>57720926
>not wanting even a small cpu price shakeup

kys yourself to peaces

>>57720926
Me too I can't wait for $500 i3's

>>57718038
>this is considered a top athlete in burgerland

>>57720926
This AMD has no business competing with Intel they should just die off and compete with some poor companies like VIA and Mediatek

File: da3.png (91KB, 800x600px) Image search: [Google]

91KB, 800x600px

>>57720926

>>57715185
it's a new record btw

File: 1457682621063.jpg (37KB, 491x315px) Image search: [Google]

37KB, 491x315px

>>57720991
>>57720926
>begging to be gased

>>57721032
Is it fun to pretend to be a genocider online, huh, freak?

File: rated_b_for_bait.png (14KB, 220x300px) Image search: [Google]

14KB, 220x300px

>>57721049
Yes thanks

File: IMG_7797.jpg (244KB, 1200x1200px) Image search: [Google]

244KB, 1200x1200px

>>57718394
This is baffling, did they get a big Saudi investor or something?

>>57721139
she is ugly, holy shit

>>57721159
10/10 feet tho

>>57721159
She is but for some reason I still want to fuck her mouth.

>500 dollars plus the cost of a new board and possibly new ram

intel will be fine

>>57718038
Just imagine the stress on those heels. The pressure on the ground...

>hula hoops
Top kek

>>57721139
who is the ogre

File: Betty Boop Black and White Boop-Oop-A-Doop.jpg (38KB, 550x1001px) Image search: [Google]

38KB, 550x1001px

>>57721139
boop!

>>57718599
Kaby is going to be faster on Single core IPC you can see it now, it's compared to Broadwell-E that has pretty bad Single core IPC.

I predict Intelmanlets going all around saying how this CPU only gets i5 kabys performance in gaming and will cry that AMD is Finished.

>>57718038
fuck you

>>57720926
i love amd gpus and i hope Zen is great but i dont have any hope.

probably will end up with a i7 8core in the next couple of years

>>57721237
Broadwell-e has the exact same ipc as skylake you retard

>>57721166
You never saw her feet tho

>>57721356
Nope it's much faster in AVX-512 and Kaby Lake is faster as well

>>57721408
Neither of them support avx-512

File: 1474872358316.png (108KB, 357x368px) Image search: [Google]

108KB, 357x368px

>>57720272
Excellent rebuttal. Enjoy your state-of-the-art 1.8Ghz netbook.

>tfw too intelligent to use fast clock speeds

File: coparisn.png (39KB, 650x450px) Image search: [Google]

39KB, 650x450px

>>57721356
Not exactly, here is simple google search result didn't try to cherry pick too much, but it's perfect example how normal everyday CPUs are faster in gaming than on Broadwell-E

Professional use on other hand is way faster, margin of 25% or more.

>>57721518
Arm's killing Intel because fastest clock and single-threaded performance means less each year.

File: cinebench single core.png (44KB, 650x450px) Image search: [Google]

44KB, 650x450px

>>57721519
and here is perfect example where Single threaded operation

You can expect all of this benchmarks with new ZEN saying how much better Kaby is

>>57721551
as long as people are doing PC gaming x86 isnt going anyplace

>>57721386
http://pics.wikifeet.com/Maisie-Williams-Feet-2200405.jpg

>>57721567
Do you not know what ipc means? Here's a hint: it's not the same as single threaded performance.

>>57721589
try as i might i will never understand foot fetish

literally who cares as long as it works lmao

>>57721590
>benchmarks don't matter any more, this number on paper matters more!

When Zen comes out it will be same tired AMD story, good at multi threaded stuff, shit at everything else.

>>57721590
If both CPUs have exactly the same instruction sets, it's pretty much single threaded performance.

>>57721649
What the fuck are you even talking about? You said broadwell-e has bad ipc which is not true since it has the same ipc as skylake.

>>57721655
IPC = instructions / clock
Instructions meaning performance or number of calculations, nothing to do with instruction sets

>>57718038
everytime

>>57721694
>IPC = instructions / clock
So, same problem, render a scene in Cinebench, we compare same clocked CPUs in this case 6900 or 6800 to 6600 and see that 6600 has better IPC.

>>57721812
6900 will be faster since it has more cores, but each one of the 6900's cores will be equal to each one of the 6600's cores

>>57721845
>Kaby is going to be faster on Single core IPC

What part of SINGLE CORE did you miss?

>>57721883
Where did I ever say kabylake won't have better ipc? See >>57721666

>>57721957
This is literally my first post saying that Kaby will have better single core ICP than Zen you fucking moron

How fucking stupid can you be not to be a part of conversation where you jump around screaming at top of your lungs and you don't even fucking know what people are talking about?

>>57718038 clever bitch.

>>57722023
>>Kaby is going to be faster on Single core IPC
Why did you quote this then you dumb nigger.

>What part of SINGLE CORE did you miss?
And why are you arguing about kabylake having better ipc when NO ONE ever fucking said otherwise.

And how is it a fucking conversation if you just said that was your first post? So this retard >>57721237 isn't you? You're just another retard jumping into the conversation bringing up irrelevant facts about kabylake?

N O S A U C E

>>57721590

>implying IPC has been used literally in recent history

it's been doublespeak for single thread perf/clock for ages schlomo

RIP Intel

You will be missed

>babylake will have the same ipc as skylake
>skylake ipc is within a few percent of broadwell
>broadwell has the same ipc as haswell
>ipc is independant from clock speed
>single threaded performance is ipc x clockspeed
>kaby will have a higher stock clock speed

I mean at the base level it's the most simple shit to understand, but the last few threads there have been too many retards. I'd be happier if I was being baited by this right now.

File: 1460926521305.png (368KB, 640x480px) Image search: [Google]

368KB, 640x480px

> tfw no under 100$ cpu for poorfags

>>57722607
is this image real or fake

>>57722613
How can anything we see be real if our eyes aren't real

>>57715185
Hahahahahhaaha

>>57722613

Science has gone too far.

>>57715003
Whats the point of charing $150 more for extra turbo clock? I thought all AMDs were unlocked by default, is there any point to getting the $500 version of Zen SR7?

>>57715003
AMD is finished and backrupt.
>name is in the toilet
>won't severely undercut intel to move products get back in the public eye (for anything other than failure) again
500 buck consumer processes yeah that's going to move fucking millions.

>>57722613
yes

>>57722846
Is the $500 one the consumer model though?

I thought the $150-250 variants were the consumer models.

>compiling software on that sr7
HNNNNGGGGGGGGGG

>>57720518
>>57720859
Why are there so many tech illiterates on /g/ lately?

>>57717864
AMD just knows how to utilise more cores more effectively than Intel.

File: 1480264680678.jpg (138KB, 807x861px) Image search: [Google]

138KB, 807x861px

>>57715003
Sauce fucking where?
Not only do I want to believe in that tasty $350 SR7 as an upgrade to a 3570k in everything plus extra juicy cores for building, I want you guys to stop overhyping this shit. Why does Intel false flag so much?

>>57722839
These are only rumours remember, but a lot of sources have said there's a top end cherry chip, they're just charging more for a better pick in the silicon lottery but also the higher base clock. Then again, the $350 sample might be the chips which only scrape their way into the bin.

>>57722846
>8c/16t high end model
>consumer model

fucking retard

>>57717603
>then amd said they were getting 40%+ but never disclosed how much
Stop regurgitating this.
They said that it was over a 40% uplift vs Piledriver when talking about their datacenter parts. They have never claimed more than a 40% uplift in IPC over Excavator.

>>57717864
1.) IPC is obviously not a static figure, it depends on the given instructions being processed, every arch performs better in somethings
2.) This bench is not directly showing IPC at all. In fact IPC is utterly invisible here. The relative per core performance at normalized clocks could be much lower for Zen, but their implementation of SMT could be significantly better than intel's which would give them larger multicore performance gains.

>>57718131
All consumer CPUs and enterprise Opterons are build using the same Zeppelin die, so they all have ECC support. It comes down whether or not mobo manufacturers will support it.

>>57723630
>I live in the past

>>57718561
>and they've underclocked the Intel CPU
They are comparing IPC you humongous retard

>>57718010
intel cant manage to get a 4.5 off their 8 core, why would you assume amd can?

>>57718038

get pills against my orders, get moving

all check this >>57724864

>>57718095
good air cooling is about on part with watercooling at this point.

>>57715185
D E L E T E T H I S

>>57721237
>Broadwell-E that has pretty bad Single core IPC
I don't think you know what bad ipc actually is.

>>57723090
he isn't wrong though, at least in single core performance,

>>57721519
sub 1 fps difference, awsome.
>>57721567
largely dependant on core clock speed,

you got a real world example of what the difference means next to a synthetic one.

The synthetics will show minute differences to be FAR bigger than they actually are, see ssds where a 2000mb/s and a 600mb/s have VASTLY different speeds, but real world they are within margin of error at worst, exactly the same at best.

Single core has not been an issue, or at least a real issue since sandybridge, and zen is that at minimum so all that's left is its multicore support in question as one blender bench doesn't show what workloads it handles well.

>>57715003
Another year, another 2nd place for AMD

>>57715185
fuk

>>57725772
No one should have expected anything from Kaby Lake to begin with. Intel's own slides on it told the full story, and we've seen them for the better part of a year now.
Improved media decoders, and process refinement. Thats Kaby. Expecting a huge IPC increases is purely a pipe dream.

>>57720926

If you hate AMD you should be hoping Zen is amazing, because then it means all the Intel CPUs you buy might actually need to come down to reasonable prices from competition.

File: AMDvsIntelretailPC.gif (30KB, 896x553px) Image search: [Google]

30KB, 896x553px

>>57726033
Intel has never lowered their prices in response to competition.
When AMD was competing against them, and winning, intel illegally acted to destroy AMD's market share.
They didn't make their own product more appealing in price.

Get ready for the top end mainstream intel processor to have a $450 MSRP with Coffee Lake.

>>57718038
godammit mom, I love you

File: Pissed Ira.jpg (89KB, 617x531px) Image search: [Google]

89KB, 617x531px

>>57718038
I love this image.

>>57718038
JUST

>>57715284
>there HAS to be some catch
>Like they marketed their A10 Cpus to be 12 CORES OF PERFORMANCE!
There's a catch, you're retarded >>57716042
Possibly SMT vs non-SMT, other leaks point to segmentation i this way
>>57717621
Fuck off Intel shill

>>57726641
All but a couple SKUs have SMT enabled.
The top SKU is the best clock binned chip, thats it.

>>57718054
>the fact that intel is about to bring their 6 cores to the "consumers" is enough to show that they are shitting their beds...
You should remember that they are releasing a 28 core Xeon to face off 32c Naples
It's pretty obvious the competition is on in the entire market
>>57721139
When they spun off their fabs they sold them to Saudi princes wanting to burn money
iirc they also got some cash injections from them around 2010 too
>>57718786
>skylake has better ipc
Between 0% and 5% better IPC in the best case scenarios
>>57720691
>Benchmarks when? Will some journalists or whatever get samples before release?
iirc some rumors pointed to limited availability in December, which pretty much meant only journos would get it
But I believe that was a rumor from wccftech so it literally is fabricated

>>57726802
>iirc some rumors pointed to limited availability in December, which pretty much meant only journos would get it
No, limited availability means preferred OEMs only. Not journalists.
The chips that AMD would send out for review would be a tiny handful, basically a single tray. Review samples are an inconsequential write off.

File: g6oNO.gif (872KB, 256x192px) Image search: [Google]

872KB, 256x192px

>>57715003
Meanwhile in amd hq
>>hey dude nvidia rapes us in the gpu market,what do?
>well it has nothing to do with our gpus being shit,they are perfect
>>i know let's try to enter cpu

>>57717621
>The version of Blender they used is internal and the drivers aren't ready for regular use.

You know its true. AMD has a track record, I don't even have to back it up.

>>57727181
Blender was compiled from source for each system. AMD made their builds public, they were heavily scrutinized because some suspected that AMD made the Zen system avoid using certain FPU ops. Nothing panned out. There was nothing suspect with their demo.
They even released the little demo scene so everyone could test it themselves.

Welcome to half a year ago, shitposter.

>>57718545
Isis not israeli secret intelligence service
good goy

>>57715155
>single threaded performance

>>>/2003/

>>57727436
Serial performance always matters.
Strong serial performance directly relates to multicore performance.

>i7 6900k is 140w at 3.2 ghz
>Zen SR7 is 95w at 3.2ghz

That's bullshit but I want to believe it.
If AMD pulls it off I'll buy it.

>>57728255
Zen cores are not nearly as big as intel's Core i cores.
Less die area, less transistors, less power per clock up to a point.

>>57728289
but the SR7 is supposedly on par with the 6900k
for once AMD is not the housefire

>>57728304
Zen isn't going to match Broadwell-E across the board in all metrics, but it'll come close in most common workloads.
Intel's cores have much, much larger FPUs. Zen still has 128bit data paths, intel's are twice as large, they can handle native 256bit ops.
Those additional transistors contribute a lot to why intel's i7E line have a higher TDP.

>>57718038
>4
waeadwaeaw

>>57721627
it's like tits
it's a hard-wired biological thing
if you don't like them you won't like them ever

>>57715003
Zen will be a failure just like Bulldozer.

>>57721139
Holy fuck she's cute.

>tfw the new i3 is overclockable and comes with a meaty clock speed out of the box too

Intel has literally won. Gaymur fags will buy that budget i3 up in the masses.

>>57728688
Forgot to mention the standard i3 hyperthreading too

>>57728688
>>57728699
>dual core

>>57718038
nani

>>57728614
She's got such a big fat face and a neanderthal brow, though.

>>57718038
Fuck

>>57728688
marginally better than a 3258.

File: 1470555159859.jpg (188KB, 500x860px) Image search: [Google]

188KB, 500x860px

>>57718038
Not today

>>57715312
why do retards care about base clock frequency? its not that hard to overclock

>>57728976
good dog

>>57728980
Because they don't know any better.
It's like the dorks that state their specs when they have a problem, like i5 3.4GHz, 8GB RAM, 500HDD etc.
They don't know what the different models, architecture or stepping is. Ignore them and move on.

Generally, though, it's (probably the shittiest, admittedly) a base on which to start rules of thumb for overall headroom when it comes to budgeting.

>>57728980
Because most people who buy PC's or PC parts will never overclock. Further overclocking is always a silicone lottery. All you're ever guaranteed is that boost clock.

>>57728902
Leaked benches show the to have better single threaded performance than an i5 6400

File: 1456549371955.jpg (87KB, 640x633px) Image search: [Google]

87KB, 640x633px

>>57715003
>8 core zen
>more than the price of the top i7 unlocked

top kek, AMD kills themselve again with bad marketing and prices

>>57729088
and the 3258 overclocked to have better single threaded performance than an i5 4690.

None of that mattered because modern games demanded good multicore performance too which is physically not possible with only 2 physical cores.

File: cover radrayliovesu.jpg (207KB, 1600x1600px) Image search: [Google]

207KB, 1600x1600px

Instagram @radray_la_grace
#radraylovesyou

>>57728688
imipressive to emulators and pretty much no one else.

>>57728980
Overclocking is pointless in 2016.

>>57729476
The only two emulators where it really matters are PCSX2 and Dolphin.
Maybe one day in the far distant future we'll have stable 360 and PS3 emulators that demand shitloads of CPU performance, but thats not any time soon.
I'm not too sure about the WiiU one thats being worked on right now.

>>57715003
now if only mad start to cooperate with coreboot again

>>57729132
wtf are you talking about? The SR7 is supposed to compete with the 6800k, 6850k, 6900, and the 6950x.

>>57729998
Ignore blatant troll posts

$350 for 8c/16 threads. Not bad.

If they're overclockable, support ECC and have amd-v support, I'm gonna buy a few.

>>57727436
>>57727545
single core only matters to a certain extent.
Really the only thing that pushes single core hard is videogames, dx12 and vulcan both solve that issue and once you hit sandybridge, almost anything programed well hit 120-144 on cpu, gpu notwithstanding.

>>57730639
Multicore performance is serial performance x multicore scaling factor x number of cores.
2 strong cores are always better than 2 weak cores.

>>57728347
>>57728304
>>57728289

From what I can tell, damn near any task you do amd and intel will be par clock for clock, and no one can really tell me where the bigger fpu will come into play, god knows I asked enough, so its likely you will not run into it through casual or even semi pro use of the cpu.

>>57730656
Intel's FPUs can handle larger vector ops faster. Having larger datapaths means they can process more smaller ops simultaneously as well, so long as the decoders and scheduler can keep up.
Tons of content creation, scientific, and big data workloads heavily make use of them.

>>57730651
no shit, but here is the thing, if you can handle a single thread/core application well, that's all the speed you need, from then on more cores is the cheaper way to get more performance.

I mean look at intel, what is going to do more for you an extra 2-10% faster single core, or 2 more physical cores?

The main issue with this approach is you run out of cpu area at some point, or at least reasonably priced cpu area.

If we assume that intel is up against a brick wall single core wise, what if they were waiting for amd to catch up to put out the more core versions of their shit too?

File: stock.png (63KB, 1643x531px) Image search: [Google]

63KB, 1643x531px

I believe in change
/end thread

>>57715185
>Intel is so good that their CPU has to be compared with another Intel one in order to make a fair comparison

Really make you think.

>>57728699
>>57728688
>>57728826
gamers are already getting fucked with a game demands 4 threads on i3's

>>57729088
until a game demands you have more than just a dual core, i3's can handle a 4 thread workload but they sure as hell don't like it.

>>57729034
base clocks matter because its what you are guaranteed. This is the bare minimum performance.

Personally, I only overclock when i stop giving one fuck about my computer because it can not handle the simplest tasks I give it. Thankfully my 955 black, while its showing its age, does everything I ask.

File: obama-1024-6.jpg (68KB, 1024x768px) Image search: [Google]

68KB, 1024x768px

>>57730714
best president ever says
stop making intel vs amd threads

>>57715003
>If sr5 is has well tier Intel is ded

>>57730797
>ooga booga

>>57728304
the house fire score mostly depend on how old their process is. Since 10nm is delayed AMD can catch up.

>>57728826
If multicore performance mattered in games then AMD would have won years ago.

>>57718038
GAS YOURSELF

>>57731232
>If multicore performance mattered in games then AMD would have won years ago.

Plenty of games have engines well threaded enough for an FX 8350 to fair very well. Of course those same games have Haswell-E and Broadwell-E at the top of benches too, but it shows how well they scale with threads.

>>57715185
remember fermi? first we must make CPUs efficient again. (mATX/ITX/similar form factor builds already benefited a lot from skylake because of this).
amd will never be competition to intel again when it comes to processors.

>>57731974
Most people don't care about small form factor pcs

>>57715185
Delete this.

File: WP_20161121_006.jpg (1MB, 2592x1456px) Image search: [Google]

1MB, 2592x1456px

>>57733071

>>57731232
single core is important for legacy games, most games now are coming out with 4 thread minimum requirements, and in new apis it amds old shit plays very well... shocked to see the phenom II 955 i have push doom over 100fps at 1080p

>>57733133
>Posts 1 picture
>300MM people in the U.S.

Good %

File: Screenshot_2016-07-21-10-04-42.png (498KB, 1920x1080px) Image search: [Google]

498KB, 1920x1080px

>>57733324
On a 1060, yeah. This just shows how efficient nvidia hardware is when paired with the low end.

>>57717337
AMD Centurion would like to have a word with you.

File: 98acf8ad7d94d71170cb558c6aea4922.gif (861KB, 500x281px) Image search: [Google]

861KB, 500x281px

AMD is in a much shittier position than 10 years ago and they didn't deliver then and won't deliver now.

>>57718038
fuck off

>>57727370
That isn't what I meant at all and you fucking know it. AMD has been pushing their heterogenous computing compiler bullshit for years and nobody uses it years later, you won't get those numbers in blender without using special builds.

>>57734000
>posting the anime where one guy forgot to have an additional empty chair and everyone is now dead

>>57718038

lost it at shower curtains

>>57718038
Joke's on you, my mother's already dead. :'^)

>>57718038
:)

>>57735032
It's a CPU only test you massive retard, have you finished embarasing yourself yet? Are you going to claim AMD is new in the CPU market like this retard >>57727084
I don't know how you can still trip after embarrassing yourself so much
>>57728688
>unlocked i3's
Intel is already sweating bullets
And no, unless it's HT as well gaimu kiddies won't buy it, there's quite a lot of games that won't run with less than 4 threads
>>57729528
>Maybe one day in the far distant future we'll have stable 360 and PS3 emulators that demand shitloads of CPU performance
Both of those were parallel architectures, they will be much easier to parallelize
>>57730219
>If they're overclockable, support ECC and have amd-v support
It's obvious they will given AMD's track record
>>57731974
Zen is focused on energy efficiency m8

>>57718038
fug :DDDDDD

File: hmmm.png (11KB, 470x454px) Image search: [Google]

11KB, 470x454px

>>57737277

>It's obvious they will given AMD's track record

I do wonder - especially for the HEDT focused zen chips (i.e the one we've seen) how much ECC support actually matters. Now i'm totally ignorant if Intel's current x99 chips support the stuff but even if it does - is it something that has a reasonable practical application? Naturally there is always someone out there with a need for it but that someone does not a market make.

>>57738040
>especially for the HEDT focused zen chips (i.e the one we've seen) how much ECC support actually matters
You shouldn't care unless you're nigger rigging a critical mission server or cheaping out in a workstation

>>57715003
6 and 4 core versions will harvested from 8 core version. So all version should have same L3 cache.

>>57739481
Not if there are defects in the L3 cache and they disable the defect part.

>>57739554
all fx cpus had intact L3

I don't get why neither intel nor amd can focus on a cpu that actually uses the entire die for cpu cores/performance, rather than going "ok 1/3 is for the cpu cores and the like, the remaining 2/3 is for integrated gpu nobody who buys the damn thing will use".

>>57739633
>nobody
get your head out of your ass

>>57718038
oh,shit got me

>>57739653
It'd be the kind of thing only businesses and cheap ass laptops use, anybody who bought a high end desktop chip isn't going to use the gpu it contains.

>>57739633
Both Intel and AMD makes those you retard

>>57739633
they usually design one or two CPUs per generation, the variety of models you see is just binning. It would be way too expensive to design one CPU with an IGP and one without. Also, there's plenty of professional customers and universities who buy PCs and profti from the cheaper price because they can skip the GPU. The hobbyist desktop market is tiny compared to the professional desktop market.

>>57739633

>how to spot an IBM customer

Then again VIA's latest chip is pretty swanky for what it is.

>>57739481
I'm pretty sure the 4C chips will not be die harvested but will be APUs.

>>57738150
But muh ZFS

>>57739633
Intel has like over half the fucking market share in pure GPU sales faggot.

>>57718038

You sneaky bastard

>>57718038
I don't get the joke, are shower curtains actually a type of clothing?

>>57740930
For her yes. So would a 6 man tent.