ITS HAPPENING - Ryzen Benchmarks - Faster Per Core Than Intel’s

>>58935966
wccftech are making up numbers, that's videocardz's chart.

The 6950X is a $1600 CPU with 10C/20T, its all core turbo should be around 200-300MHz

Pretty impressive.
If it's true.

File: 9.jpg (28KB, 356x356px) Image search: [Google]

28KB, 356x356px

>>58935966
>1700x
Where is the 1800x on this?

>>58935966
If the 1700/1700x/1800x are all 8/16 core threads is there any reason why anybody wouldnt just get the 319$ cpu over the 499 one and just overclock the hell out of it since all Zen cpus are unlocked from the start?

Fake, pure fake

>>58936325
Because not even 5% of computer lego builders overclock, that number is even smaller for people who don't build their own PC.
It's cheaper to sell these chips unlocked then waste money locking the overclocking, a very small amount of people overclock, those that do usually do influence the normalfags though, so its good marketing.

>>58936355
are they going to be binned chips or something? all of the ryzen cpus are unlocked

File: 1486135360585.jpg (45KB, 567x567px) Image search: [Google]

45KB, 567x567px

>>58936258
it's off the charts

>>58936355
So what you're saying is just get the 1700 for 319$ and you can just overclock it to 1800x level or even above that considering they are the same chips just running at different mhz?

>>58936376
Don't use that word when you don't understand it fully.
Every chip is technically binned, if you wanna talk about it be more specific like telling me a binning target.

>>58936399
Yes.
Though the 1800X will have a higher percentage of better silicon so it'll overclock higher if I'm reading this like I should.
Then again you can fumble silicon lottery and get a 1700 with better quality than a 1800 and overclock it higher.

>>58936430
Alright, I'm looking to retire my old overclocked 2500k and it looks like Zen is gonna replace it. Now I'm just waiting for gaming benchmarks. I have my doubts about that considering most games dont even utilize 4 cores fully let alone 8.

>>58936430
don't you need at least the 1700X because the non X has 30W lower TDP?

>>58936506
TDP is ignored when overclocking, it might as well not exist to the chip at that point.

>>58936541
thought there was a physical reason for the TDP declaration

I really want integer benches now, if the FP side is this close, then the integer one should be a sight to see.

>>58936577
TDP is just a guideline for the OEMs for the cooling solution, it just means the cooling must dissipate a AT LEAST certain amount (ex 95W) of heat reliably, the chip might consume more or less depending on the workload.

>>58936111
>triples of truth

Intel is literally on suicide watch. A 500$ chip performing near the same as a 1,600$ one? Fuck

>>58936667
Uhh, I got some news for you..
The 1700X is a $390 chip.

File: 1371871249114.jpg (73KB, 1068x600px) Image search: [Google]

73KB, 1068x600px

>>58935966
So the 4c/8t 1400x seems like the sweet spot for gaymen. The 6c/12t is going to consume more power and not clock as high. In this case, it seems like anything above the 4c/8t is going to be detrimental to performance.

>>58936711
It's more than likely that the 4c/8t parts won't clock as good as the 6 or even 8 core parts since they're literally rejects of rejects, chips that failed a bunch of clock and power targets multiple times, meaning they're just trashbin silicon.

>>58936705
LUL I thought that was the 1800X.

Even more d a m n

>>58936705
Oh yeah that makes Intel look much better.

File: 0aa9d1aa-4dea-42f4-c229-4cd07a235319.png (411KB, 760x766px) Image search: [Google]

411KB, 760x766px

>>58936742

>Intel getting shafted by AMD's trashbin silicon

>>58936759
>>58936705
>>58936767
It's a 1700x OC'd to 1800x level though. Probably to estimate 1800x's performance.

>>58935966
>synthetic benchmarks

>>58936794
But no one is stopping you from getting the 389$ chip over the 499$ one if you know how to OC.

>>58936794
The 6950X can overclock too, though I'd be scared shittless overclocking a $1700 CPU because I'd kill myself if it went boom.
If I did I wouldn't touch the voltage, even then that's a lot of wattage and voltage sensors controllers running a clockspeed the chip isn't made to sustain, that's why turbo exists.

>>58936742
Is there a source for that? The 1400x base clock is 3.5GHz at 65W while the 1600x base clock is 3.3GHz at 95W.

>>58936325
Not everyone OCs and perhaps the 1800X will be special silicon handled only by the purest maidens so it can (on average) OC more than the cheaper variants. That's not very likely though, the price tag is most likely just the "premium" from being the top of the line product.

>>58936328

HAHAHAHAHAHA INTEL FAG IN SUICIDE WATCH!

>>58936867
No source, just experience.
Those 4 cores come from an 8 core die, for those 4 cores to exist the chip has to fail a LOT of targets, and if it fails a lot of targets it's more than likely trash silicon.

>>58936873
Most likely the 1800x is just there for retarded alienware/prebuild builds or retards who cant OC. AMD can get more profit from them.

>>58936911
So you are saying they might not OC as well? If the 6c/12t is clocking lower, shouldn't that mean that the 4c/8t is actually the better of the batch? Especially since the 4c/8t gets there at a lower TDP?

4/4 CORES BTFO!

https://twitter.com/CPCHardware/status/830851176853823488

>>58936949
They'll OC, just not to what you'll see on higher quality chips.
No I don't mean this one will OC to 4.8(example) and this one will OC to 3.9(example), there will be some 300-400MHz difference, nothing too enormous.

I'm genuinely excited. I have a 4670k at 4.4GHz, but I want more cores.
>tfw getting the 1700, maybe 1700x if it OCs better
Can't wait, lads.

>>58936841
CPUs don't go boom from overclocking, not unless you're absolutely fully retarded about the entire thing.

>>58936949
The 4C CPU may have a lower TDP but it also has fewer cores and as such it also has greater power consumption per core in order to hit its higher clocks. It may end up clocking better in practice though due to various practical reasons beyond its quality, such as requiring less cooling/running cooler with the same HSF. You can push a higher voltage into a 4C CPU than you can into a 6/8C CPU while maintaining the same overall power consumption, so you may get higher clocks in practice, depending on how good the cooling is and how big the silicon quality difference is.

>>58936989
I feel you bro. I'm on a 2500k and I was temped to get a new cpu around christmas. I'm glad I waited.

>>58936951
It will still render Pentium and Core i3 useless.

File: 1480435026045.jpg (63KB, 250x323px) Image search: [Google]

63KB, 250x323px

>mfw people bought Kabylake for premium prices and didnt wait for Zen

>>58937018
>https://twitter.com/CPCHardware/status/830851176853823488
Same. I was eyeing the 6700k/7700k a few months ago. I'm so glad I decided to wait.
I'm a yuropoor, though. I'll probably have to wait until the end of 2017 for msrp prices.

I keep reading contradictory shit about this, will X370 boards support 2 full PCIe 3.0 x16 slots, for full-bandwidth SLI/CF? Some articles say they do, other say that X370 only provides 24 lanes for PCIe expansion, so the best you could do is x16 + x8.

>15% performance for 19% clockspeed on Zen

Not bad scaling, at all.

>>58936328
like your mom boobs

File: amadaexcite.png (38KB, 422x537px) Image search: [Google]

38KB, 422x537px

>>58935966
HABBENING

File: 3708.png (246KB, 391x327px) Image search: [Google]

246KB, 391x327px

The only thing I'm bummed about is that Zen will only be supported on W10. I know what ''supported'' means and that Kaby ran just fine on W7 but that was because it's the same fucking architecture as Skylake which does run on W7. Zen however is a entirely new architecture which probably wont get recognized by W7.
Well I guess I have to get on the W10 wagon.

TEST THE ALUs TEST THE FUCKING ALUs I DON'T CARE ABOUT THE FPU JESUS FUCK WHY ARE YOU HIDING THE MOST IMPORTANT STUFF

>>58937082
Zen is just a x86-64 CPU like any other, it uses the same ISA, W7 will run on it. Where you're potentially fucked is drivers for all the new shit on the board.

>>58937082
Every device, new and old, that I've tested runs better on Windows 10. The under-the-hood changes are really amazing, even if all the UWP crap is dogshit (It's very easy to avoid and disable/neuter).

It even made a shitty A6-4455M Bulldozer fake dual core reject CPU run well. It's a goddamn miracle.

File: AMD-Ryzen-Chart-2.png (13KB, 1215x744px) Image search: [Google]

13KB, 1215x744px

>>58937088
Here you go.
This isn't even the 1800X and if I remember correctly it was tested with grandma-tier 2133 DDR4 with 18+ CAS latency and 2T command rate.

File: AMD-Ryzen-Chart-8.png (13KB, 1217x748px) Image search: [Google]

13KB, 1217x748px

>>58937117
Some more heavy integer stuff.

>>58937117
it was 2400 17-17-17-24 2T iirc

>>58937146
Man, where do you even get such shitty memory anyway? It's like the dude was waiting in front of a micron factory begging for a few DIMM sticks and whatever the trashman had on him gave him outta pitty.

>>58937171
probably grabbed whatever was on hand and would post with the board+cpu

>>58936911
I highly doubt the 4 cores are rejected 8 core units simply based on the 4c/module design. But hey, it's possible I suppose. I'm really digging the 16MB l3 cache 6/12 cpus.

>>58936258
the overclocked 1700x should be where the 1800x is, the 1800x should be 4 ghz

>>58937334
The modules need to be symmetrical, you can't disable a single module(CCX), you need to disable 2 cores in each CCX to get a 4 core.

>>5893735
Why wouldn't they be able to simply fab single module units?

>>58937365
Because that's a whole different die and one module units are for Raven Ridge a few months from now.

>>58936611
>TDP is just a guideline for the OEMs for the cooling solution
But what about PSU's? Can't I use the TDP shown on the charts as a reference when choosing my power supply?

citing WCCFTECH!

What the fuck OP

>>58937425
No.

>Why
Because TDP isn't power consumption.

>>58937470
AMD has typically ran their power consumption within the TDP, though. But once overclocking, that goes out the window as well.

>>58937146
17-17-17-39 2N, you can still find it in the Passmark database.

>>58937171
I had to double and triple check this. When DDR4 was first released there were 2400mhz kits with CAS14 and 15 timings. Not 17. These timings were intentionally set.
This likely means whoever has the system was measuring the performance impact of memory latency on several runs, or they wanted to generate as much hype as possible with a salacious bench result.

Intel is fucking SHOOK. They're rushing Coffee Lake to market way early because Babbylake is going to get fucking rocked by Ryzen in price/performance matchups

https://arstechnica.com/gadgets/2017/02/intel-coffee-lake-14nm-release-date/

>>58937569
That explains a few things, still even with the slowking-tier memory the results are pretty damn good.

>>58937582
>They're rushing Coffee Lake to market way early

It's just another skylake rebrand so really they can release it whenever they want. it's only early from a marketing perspective.

File: 1466086752488.gif (3MB, 264x240px) Image search: [Google]

3MB, 264x240px

>>58937582
>shilltell going Tick-tock-tock-tock now

>>58937628
So who the fuck is going to line up to buy an overpriced i7-8735K or whatever?

>>58937582
This is the best evidence for genuine happening. Intel don't break its policies unless situation is dire af

Aren't these stock benchmarks?

Weren't there rumors that Ryzen could OC to 5.0 Ghz?

File: 1437357585240.jpg (51KB, 573x609px) Image search: [Google]

51KB, 573x609px

>>58937582
This what you get for asking premium prices for 3-4% lazy improvements every year intel. I hope u fucking burn.

>>58937678
An intel engineer admitted that they seen Zen as competitive.

>>58937679
On a single core.
Don't expect to OC 8 cores to 5.0 though, that's too much.
If they can hit 4.4 I'm more than satisfied.

>>58937582
b-b-but muh 15%* performance over skylake

>>58935966
No fucking way. What a bunch of bullshit.

>>58937582
"The first 10nm chips consumers will get their hands on won't come from Intel but from Samsung and Qualcomm. The upcoming Snapdragon 835 SoC will be built on Samsung's 10nm process, enabling a 30 percent smaller die size and a claimed 40 percent reduction in power consumption."

Intel rested on their laurels for far too long. AMD and Qualcomm are going to tagteam them. Even Microsoft is switching sides.

Reminder: Some anon said he'd hang himself if Ryzen could even hit 3ghz.
So we're owed a livestreamed NEET suicide any day now.

File: 133653811.jpg (198KB, 1100x900px) Image search: [Google]

198KB, 1100x900px

>>58937582
>https://arstechnica.com/gadgets/2017/02/intel-coffee-lake-14nm-release-date/

TOCK TOCK TOCK

This Intel panic only increases my hype for Ryzen.

File: 1458792100174.jpg (31KB, 365x609px) Image search: [Google]

31KB, 365x609px

>>58935966
How can this happen?
It's not supposed to be that way.
These benchmarks are fake.

>>58937765
You can't really compare process nodes anymore, it's way too complicated now.

>>58937582
Intel(tm) is on my system, but I want them to wake the fuck up. I'm definitely getting Ryzen this year so I can finally retire the 2500k and move on.

>AMD doing this well with FP

I can only imagine my shit-eating grin when the full set of tests come in rolling.
Gaming should be a nice surprise as well, the large caches should compensate somewhat for the lower clocks and IPC compared to Kaby.

>>58937825
The point is that Intel lost the nm race. Something that would have been unthinkable even 10 years ago.

What mobos accommodate Ryzen?

>>58937901
The point is it's clickbait trash that ignores reality. Intel could rename their nodes, and still be in spec.

>>58936383
underrated post

>>58937950
AM4 socket, x370, B350, A300, and x300 chipsets.I think I'm missing one.

>>58937956
We'll see who's pushing clickbait trash when Samsung and MS release a low-cost laptop using a 10nm Snapdragon CPU with x86 compatibility layer that shits all over anything Intel can bring to the table.

>tfw finally becoming a wagecuck in a month
>tfw can afford buying Jim "Pussy Destroyer" Keller's second masterpiece

>>58937971
i think it was just 4

>>58937971
A320

>>58938008
>>58937971
>all these fucking chipsets
>all these retarded number and letter combinations
why the fuck,... what does it all mean?!?!

>>58936402
>Studied computer science
>Unemployed
>Lives in basement
>Bullies people over the Internet

AMD confirmed for finally slaying the intel jew.

>>58937425
You're confusing 2 different watt measurements.

TDP measured in watts when talking about a CPU is heat it puts out (Thermal Design Power).

When discussing power, like the actual electricity consumed, that is a different watt, though closely related in that they tend to correlate. As in the more watts your cpu sucks down the more watts of heat it puts out.

File: ryzen_chipsets.jpg (38KB, 620x349px) Image search: [Google]

38KB, 620x349px

>>58938018
a300 and x300 are mini-itx

File: Screenshot_2017-02-13_21-57-02.png (33KB, 533x316px) Image search: [Google]

33KB, 533x316px

>>58937992
It's denying reality to compare different process nodes naming schemes to each other. They have nothing in common anymore.

>>58938053
why do they need special chipset for that? or is it not really different, just renamed?

makes no fucking sense

File: [HorribleSubs] Fushigi na Somera-chan - 06 [720p].mkv.[00:01:23.000].0001.png (770KB, 1280x720px) Image search: [Google]

770KB, 1280x720px

Will / Is Zen and all of it's chipsets be supported in Linux day 1 / already? Or will I have to be careful about which mobo I get?

>>58938018
x370 is the overclocker & sli/crossfire chipset, b350 is the overclocking only chipset, A320 is the mini-itx chipset, x300 is the mini-itx overclocker chipset

>>58938082
What is the "that" to which you are referring?

>>58938163
>missing the point
i mean how do they decide on the chipset "name" ? what does 370 350 etc stand for?

>>58938177
different chipset name just because it's ITX

>>58937971
None support DDR4 RAM?

>>58938112
Nothing is ever supported on day one in Linux, friend.

>>58938163
a320 is full desktop sized motherboard, a300 is mini-itx

>>58938112
well a320 and b350 boards have been on the market for like a year or something but i don't know for sure

>>58938213
Larger number == more features and lareger size
Better model if you want to overclock
eXtreme for gaymen

>>58938213
They are basically naming it based on intel chipsets.

Also most likely they have different product managers for the different chipsets to sell.

>tfw already purchased 7700k

Did I fuck myself anons?

File: 1462847110935.png (234KB, 800x612px) Image search: [Google]

234KB, 800x612px

Holy shit. If this is true...
>intel btfo
Fuck this is the best timeline. I mean I didn't think winning would continue into 2016 but 2017 is looking like another landslide.

>>58938249
listen... that's fucking obvious, im talking about how the choose those exact fucking numbers...

why A for essential chipset
why B for mainstream?
why 3xx ?

do you understand what im talking about or are you just goin to miss the fucking point again?


>>58938267
makes some sense, but what about the letter choice? wouldnt it be more logical to use M for the "mainstream" chipset?
and if you're using different letter prefix, why do you even need different number behind that?

>>58938220
Everything made by Intel and Nvidia is.

>>58938008
Yeah, that's right.

>>58938018
>>58938213
Not actually sure where the 300 comes from (Third generation AM socket, maybe?), A is baseline, B is mid, and X is enthusiast. The x300 chipset is a super cut down x370 that's the size of a pinky nail. Still enables 28 PCIE lanes for cramming all kinds of I/O onto the package. The smaller chipset and mostly bare pcb enables plenty of room for extra ports / headers.

>>58938215
DDR4 2400 is standard maximum supported memory speed. The B350 and X chipsets will obviously enable higher clock speeds.

>>58938280
Maybe.

File: 1467766208151.png (973KB, 801x1500px) Image search: [Google]

973KB, 801x1500px

>>58936788
Intel house fire confirmed. The poo has landed in the loo.

>>58938280
yes.

>>58938292
see >>58938053

>>58938312
>Third generation AM socket
nope, the socket is confirmed AM4.

>A is baseline, B is mid, and X is enthusiast.
all letters except X means nothing in relationg to "mainstream" and "essential" like they wrote in pic here: >>58938053
A and B makes no sense for that.

>the x300 chipset is a super cut down x370 that's the size of a pinky nail
alright makes some sense then

>>58938330
learn to read you retarded wankstain

>>58938309
Go look up the HDMI bug or the case open bug on skylake. Shit is buggy as hell for months, faggot.

>>58938385
Better than the not detected bug and the hardware doesn't function at all bug on AMD hardware.

*rolls eyes*

File: 1463859525029.png (593KB, 1000x920px) Image search: [Google]

593KB, 1000x920px

>Intel
>Got blueprints from IBM
>didn't want to deviate at all from them
>AMD found more optimal way of doing things right away
>Cyrix backwards engineered a better x86 CPU than what Intel ever could
>Only way Intel could survive was shady business tactics
Intel was always 10 years behind competition. This is a large reason people are excited for Amd's new CPU even if they aren't buying one, it's a new architecture, something Intel lately hasn't done for... since the Pentium 3... so Intel has been floating along on the back of Pentium 3 nearly 20 years, and will be over 20 by the time they make something new, while AMD has had 3 different cores in that time frame with 2 of them being good, one of them being good for specific applications for a short time, would likely still be good enough today if they released full desktop variants but fuck it, when gamers are concerned its hard to argue in favor of construction and cat cores. Throughout the entirety of the bulldozer base, look at what AMD did gen over gen... and now look at Ryzen, and imagine it to be Bulldozer levels of optimization left to happen. now realize Ryzen currently on engineering samples that throw errors like a mother fucker, is within/over 5% IPC of Intel's current CPU. I cant wait to see if either there is a performance stand still, you can't get much faster than Intel is, or if we have been rammed up the ass by Intel for so long we forgot they were even there and its only now that we finally see real vagina we remember a cock is still lodged up our asses.

>>58938361
samefag again

Why didnt they go with something like:
>Enthusiast: X350
>Mainstream: M350
>Essential: E350
>Enthusiast SFF: X300
>Essential SFF: E300

I get the X3xx for enthusiast since all companies like to use X for
>EGGSDREME !!!!!!!!!!!
etc

but the rest is just random nonsense

>>58938415
Intel has way more fuckups than Bulldozer, Bulldozer is AMD's only spectacular fuckup in the CPU market in 40 years, the only other minor fuckup was Phenom which was not even a architecture fuckup since it was fixed in B3 stepping, the other fuckup was K5. But these are fucking nothing compared to Presshot, Itanium, Larabee, iAPX 432, and their own fuckup with P5 FDIV that wasn't nicely fixed with a stepping, but a fucking recall. What about their complete failure with the Atom? What an abortion, mobile market? 14nm and lower lithography woes where they allowed their competitors that were lagging 4 years to close the gap to less than a year? These magnificent failures would destroy 5 companies over, it's a good thing Intel has more money than sense to live through it all.

>>58938401
I typically like your posts, but it's ignorant as hell to suggest shit just werks on Linux on day 1.

>>58938361
oh you mean why did they brand them with those codes? they just did. there you go, calm your autism.

>>58938361
Yes, but the AM socket family started with AM2/2+, then AM3/3+, now AM4. AM1 came post FM2

does it still use am3+?

File: hurrimretarded.png (10KB, 300x300px) Image search: [Google]

10KB, 300x300px

>>58938435

>>58938458
That's too bad pal.

>>58935966
Intel seriously underestimated AMD this run.
They've been spending stupid amounts of money in acquisitions that hardly even gave them revenue. They're laying off employees by the hundreds and even thousands. They released Kaby Lake which is essentially a DRM Locked, MARGINALLY better, and 5-10% more expensive. Intel has been literally fucking themselves by thinking AMD would never recover.

Intel is in complete panic and damage control mode. This isn't even bait. Sure their revenues and profits have increased but man they're suffering, relative to themselves at least.

We must hope these benchmarks are accurate.

>>58938057
They have the same things they've always had in common, you just never understood what they were. Choosing to upload that picture doubly proves it.
A foundry doesn't name a process node for any back end scaling metric. They name it for minimum feature size. The smallest possible line they can etch. That is what they're named for. Really the only people on the entire planet who have a problem with this are know nothing autists who like playing the loud mouthed expert on internet forums.

Some points on that chart are horrendously wrong any way.

What about power consumption and heat? How is ryzen vs intel in this regard?

>>58938481
That's on top of the already shrinking desktop market.

File: lvl1okhand.jpg (16KB, 209x230px) Image search: [Google]

16KB, 209x230px

>>58938480
>clueless
>tries to be clever
you're not fooling anyone but yourself

>>58936402
That ain't true though. The way /g/-tards always use it is wrong. All dies on a single wafer are made with the same target, and generally only a small fraction are perfect examples. The term binning stems directly from the word "bin" as in trashcan, as much of the silicon would be tossed out if the architecture weren't designed with having only a portion of it functioning properly in mind.

There are many ways imperfections can affect the function of the silicon, and many variables to consider in the architecture that determine what can and cannot be salvaged, though it generally comes down to maximum/minimum voltage and clockspeed stability and number of functional cores.

However, testing each chip fully adds significantly to production time and overall cost per die. I would bet a part of the reason "automatic-overclocking" was implemented was to more easily determine relative die quality, and the key difference is how high it could clock and at what voltage. It doesn't mean a lower binned chip can't reach clockspeeds above its rated turbo, but it dows suggest it's less likely than the pricier ones and would likely require much higher voltage.

Tl;dr the more expensive ones cost more because they're proven to reach higher clockspeeds than the lower ones.

>>58938506
Ryzen is more power efficient in spades compared Intel. Intel is literally a house fire in comparison.

>>58938506
>>58938450
>>>/google/

>>58938416
I agree, it's total bullshit, the chipset, and cpu names. What bothers me is the "A" chipset seems to be the budget choice, while the "B" is mid? "A" is better than "B" why is it not first? Intel has H and B, for Home, and Business. I still hate the 3, 5, 7 scheme soo fucking much. It werks with the Xeon range, but not for consumer gear. Give me 4, 6, 8 for the cores, plz.

>>58938512
i don't understand why my posting the feature list of each chipset like you asked has triggered you so badly but i'm guessing actually living with that kind of autism is less pleasant than dealing with it. god bless.

>>58938535
>"A" is better than "B" why is it not first?
If we used it by grade order then A would also be better than X. AMD is clearly not doing that.

>>58938509
Intel's investments have totally fucked it.

File: 1481682217288.jpg (34KB, 600x461px) Image search: [Google]

34KB, 600x461px

>>58938535
this, so fucking much this

>>58938568
because that's not what i asked you fucking retard. learn to read.

>>58938492
Just getting to a new process node is expensive af. For 7nm, it's about 270 million right now. For a 10nm chip, it takes $120 million for the design cost, plus 60% for the embedded software. There is a reason an iphone is $750, the consumer pays for advancing the tech. 14nm is going to be a very long lived node, and the player with the better density will always return the best profit margins. Most customers will wait until 7nm hits anyway, 10nm is very much an in-between node anyway. Scaling is dead anyway, but you already know that, let me guess.

>>58938524
The canard PC benches showed the Ryzen ES at 91w power draw versus the 6900k's 96w. The extra ~45w in the intel chips is for their AVX module that only ramps up under specific loads.

>>58938535
Look at it as alphabetical order, A is the beginning, or baseline. But I'm not sure there is really any rhyme or reason for any of the naming schemes. Seems like they let the egnineers do all the naming, and they either just picked shit arbitrarily, or it's some inside shit.

File: hacc.png (699KB, 506x734px) Image search: [Google]

699KB, 506x734px

>building PC this week
>last part is the CPU
>was going for i5
>see this

wat do

I'm impatient as fuck - is there an exact release date yet

>>58938509
The desktop is shrinking because there is no reason to buy a desktop anymore. We have advanced far enough a tablet or laptop is sufficient. IPC is dead, so why even spend money for a new desktop anyway? Unless you can afford a $3k rig, there is nothing to really get excited over. Maybe a D series Xeon, but fuck me $850? C'mon, Intel. Fuck.

>>58938535
Because things keep getting better from the worse offering. Otherwise they'd end up looking like a seller's review on ebay. AAAAAA+++++++++

>>58938614
just wait 3 more weeks

>>58938614
Info I've received through the grapevine is the 1st (no confirmation if this is a reveal or full launch) and an assurance there will be plenty of supply to meet demand.

Would the Ryzen 3 1100 be good enough for gaming? I don't really do anything else like 3D rendering or anything. What do you think? I just wanna play on high settings.

>>58938614
just wait until Intel reacts to amd's new lineup and buy whatever they offer. There aren't any good games anyways so you might as well wait

>>58938292
>makes some sense, but what about the letter choice? wouldnt it be more logical to use M for the "mainstream" chipset?
>and if you're using different letter prefix, why do you even need different number behind that?


Ask the vp of marketing that pulled it out of their ass.

>>58938616
>IPC is dead
nah, AMD just revived it.

and "sufficient" is not enough for a lot of people, and we still dont have desktop performance in notebook size, so desktop market will live on quite some time.

>>58938570
Everyone knows X is XXXtreme gaymur tier anyway, I left it out of the discussion for obvious reasons.

>>58938642
obvious shill is obvious

B350 + 1400x/1600x probably enough for me.1500 looks tempting though, as i dont OC crazily, with extra 2 cores and 65w

>>58938614

just get the i5

if you're feeling it's under-performing then just replace it later in the year

>>58938587
>and the player with the better density will always return the best profit margins

This is not even remotely a single scrape of truth to this. High area scaling can give you more potential candidates per wafer, but this has nothing to do with how well the process yields. Intel's dense BEOL on their 14nm process didn't do Broadwell CoreM, or desktop Broadwell any favors.
Candidates per wafer is only one single cost metric. You're totally ignoring the fact that different processes with different gate topology will use different numbers of masks, so right off the bat there can be a huge difference in cost to the customer.

>Scaling is dead anyway
Again, not even remotely true.
Area scaling will continue to progress, at every foundry, to 5nm and beyond.

Loud mouthed know nothing.

>>58938653
I dunno man, it looks like they caught up, and maybe some on the efficiency side also. The next trick it to advance over Intel on the next node, then we talk about IPC increases.

>>58938614
I have a 6600k. It's nice and all, but it's not 8 cores. I would wait.

>>58938651
sorry pal, i thought 'what does it all mean' meant what do all those numbers and letters mean. i explained it, posted a very handy and easily read table to explain it all for a normal consumer without crippling aspergers. you now want some sort of rationale for arbitrary branding i guess? there isn't one, it's arbitrary. ask your minder to explain that word to you. again im very sorry that you're so angry about something so unimportant, must be awful. take care friend.

>>58935966
should i invest in ayymd stocks

>>58938614

AMD sucks
Will always suck
They have to pay people to misguide users and pretend they have the upper hand
Meanwhile Intel has a pre installed backdoor trojan on every processor
We are fucked with whoever

File: 45687.jpg (45KB, 1500x1500px) Image search: [Google]

45KB, 1500x1500px

>>58936383
top kek

>>58938416
We'll get enthusiast-level mITX boards eventually. The only reason we won't on launch is manufacturers will mostly be following a reference design while they design and proof their own more custom variants.
It's like with GPU's, reference comes first (unless the reference is only on paper like the 970 was), then semi-custom which may be nothing more than a different cooler and/or better passive components, then full custom with unique PCB and VRM design.
Just because AMD only proofed a mITX board with poverty-tier chipset and VRM doesn't mean someone like ASUS won't design their own crazy high end version with the top-tier chipset.

>>58938721
>We'll get enthusiast-level mITX boards eventually
but we're already getting it from start? that's what X300 is...

>>58938614

I'm assuming you've bought the mobo, in which case go with the i5.

>>58938685
This is going to be exactly the same thing as 20nm. 20nm was a transition node. It only served a few customers. 10nm will be the same thing as 20nm. Customers who used 20nm before will most likely use 10nm, Customers that skipped 20nm, and went directly to 16nm, will most likely do the same thing. They will skip 10nm and go to 7nm.
Compared to 16nm/14nm, 7nm provides a 35% speed improvement, 65% less power, and a 3.3X density improvement. 10nm has a 22% speed improvement over 16nm/14nm. 7nm won't even be EUV in the beginning. They've moved from .7 to more like .64 scaling anyway.

>>58938710
P S P
S
P

>>58938810
Intel isn't doing Tick-Tock any more, it's Tick-tock-tock-tock... people wont stick to what you're saying if the time between shrinks is too long and the efficiency/IPC improvements are too small (lmao skylake -> kabylake)

File: wait.webm (2MB, 1280x720px) Image search: [Google]

2MB, 1280x720px

>>58938614
wait the test jesus...

>>58938432
Samefag harder. Maybe then you will have more replies other than (You) and me.

>>58938280
If you had anything over Ivy Bridge, then you should end yourself.

>>58938846
Once at the end of downsizing, all left are optimizations.
INTC10 to TSM7, nearly same density based on my gathering of open source info.
SAM10 to TSM7, about 40% more dense for TSM based on my gathering of open source info.
SAM seems thinking they can still have competitive PPC metrics with somewhat less dense process. With some optimizations, 30% more power saving may be doable without scaling and that will just match to TSM7's metrics.
TSM10 seems mainly iPhone node, a stop-gap node, and also mobile only. For most customers it would be just waste of resources. TSM7 is just one year behind to TSM10.
GF10 and GF7, what they are doing?
Given vast majority of all SOC are for mobile, even a TSMC10 "stop gap" node is a must for chip makers.

The main issue not being said here is that for Samsung is that their 7nm requires full EUV which makes the process prohibitively expensive. That is why they are trying to milk 10nm node as long as possible.

On the other hand TSMC manged to figure out a way to use multi-patterning at 7nm (with only partial EUV), making their 7nm node much less expensive, which gives them a technical and price advantage over competitors.

>>58938810
20nm was a short lived node because it was the absolute end of what they could get out of a planar transistor at that feature size. The short channel effect is a real bitch.
Samsung and TSMC only offered these nodes for cellphone SoCs to decrease the cost of their upcoming FinFET process nodes. TSMC and Samsung will both keep their respective 10nm nodes online for years to serve the commodity ARM market while 7nm leapfrogs it for high performance ARM SoCs. Companies like Rockchip and MediaTek will put out 14nm and 10nm parts for years to come. The majority of foundry customers don't tape out designs on brand new bleeding edge process nodes. They wait for costs to fall and yields to rise as the process matures.

Intel will stay on their 10nm node for several years until their 7nm facility is finished and brought online.
Global Foundries is the only high end fab outright skippping a 10nm process.

>>58938710
Intel should pay you better pajeet, you could be making more by collecting cow shit and piss for human consumption.

>>58936867

This is exactly why we used to had Semprons with 1 core that unlocked to 2, 3 and even 4 cores.

My old Phenom II 720 unlcked his 4th core.

I remember a friend with a Sempron 160 or so that unlocked to an Athlon X4 without L3 cache.

File: 1434954448084.jpg (29KB, 310x263px) Image search: [Google]

29KB, 310x263px

>synthetic benchmarks

Oh yeah, so excited. I mean, it's not like the pieces of trash known as bulldozer and vishera did well at those as well or anything.

>bawww buy x cuz its more ethical
when will this meme die?
im not going to do pity purchases for billion dollar companies

>>58938614
>waiting for AMD anything

They literally - LITERALLY - never deliver. And even if they did, you realize just how powerful a modern i5 is? People are using the 2500k to this day and age and there is still no reason to upgrade whatsoever if your target is 60 fps gaymen. No i5 is gonna bottleneck you and ryzen won't be any better than them. In fact, they were aiming for about 90% of the IPC of haswell or something along those lines. Any i5 you buy nowadays is guaranteed better than that. No use waiting for amd shit that is all empty hype on top of smoke and mirrors.

>>58938633
2

>>58939102
>>58939113
>0.2 shekels have been deposited into your account

>>58939113

>People are using the 2500k to this day and age
>you realize just how powerful a modern i5 is?
just how retarded are you

>>58939141
I wouldn't be surprised if they got payed in USD to avoid the conversion fee. Thus allowing them to in effect pay them less.

>>58937019
nah, the i3 won't be useless, but appealing.

>>58939102
Fuck you Nvidia.

>>58939141
No seriously. Why should I do charity with billion dollar companies?

>>58939113
7th gen (more like 6.5 gen) i5 kaby lake is 25% faster than 2500k

>>58939196
intel*

>>58939197
>supporting shitty companies
how about you explain why we should do that

>>58939113
>What is Athlon
>What is Athlon 64
>What is Athlon 64 x2
>What is Athlon II / Phenom II
>Who is Jim Keller
>Never heard of them

>>58935966

Okay, so if these benchmarks are to be believed, AMD is finally competitive with Intel again.

But I honestly don't really give a shit about that. What I want to know is - what does this mean for the future? Intel has been delivering 3-4% gains year over year since forever - has AMD somehow found a secret sauce that will allow them to surpass them by leaps and bounds over the next few years?

I ask because I have this funny feeling that Intel's meager gains over the years isn't an accident, that they're actually running up against the limits of how much blood they can wring from the stone that is moore's law, and that there aren't any significant speed gains to be had without going back to the drawing boards.

I'd like to be wrong though...

>>58938428
Larrabee wasn't a failure, it was a complete success. It just wasn't used as a GPU.

>>58939325
Look at their roadmap, they're fucked if Zen delivers.
More so when APUs start to ship in with how efficient Polaris actually is. They might tear Intel a new asshole.
It doesn't even stop there, assuming everything goes well, we could be looking at 7nm APUs with Vega tier GPUs on 2019-2020. There is also Zen+ as an optimization and possibly shrinking it up in 2019. Intel isn't moving away from 14nm anytime soon.

File: 1486478419568.jpg (153KB, 600x339px) Image search: [Google]

153KB, 600x339px

>>58935966
Thank you based Shitwrecker.

>>58939353
it was a salvageable failure, but massive failure seeing as they wanted in the gpu space.

>>58939113
>2017
>60fps

>>58936325
Because of binning.

>>58939362
>processor with HBM iGPU/Cache

Exactly how fucked is everything else if this actually happens?

>>58935966

Maybe now Intel will half their outrageous prices already?

I'm still gonna buy AMD Ryzen middle segment though. Fuck Intel and their overpriced shit

>>58937582

Didn't they just fucking released their rebadge? Caby Lake?

Fucking kek, what a knee jerk panic reaction

>>58938415
>not liking cock in your ass
terrible analogy

>>58938707
Dude you're supposed to buy them at 1.90.

>>58939621
Very. A whole new market where both Intel and NVIDIA have no tools to compete.

>>58938707
Not graduating in three years really fucked me up. I could have invested almost all of that entire year's wage cuck salary and gone out with almost a million dollarinos.

I honestly don't care much about buying an AMD processor myself, but I really wish for all the hype for Ryzen to end up being real.

The market needs adequate competition to evolve - we're stalling last 7 years or so.

File: 1442101843495.gif (864KB, 160x270px) Image search: [Google]

864KB, 160x270px

>>58939648
yes, Kaby Lake. Zero IPC improvement. they just increased the stock clock and made it able to reach slightly higher boost clock and calls it "next gen"

now they're rushing to get "8th gen" out the door, complete damage control

>>58936328

Hahahahahahahahhaha

>>58937082

Blackbird is an essential download.

>>58940047
ye but it cant prevent kernel from sending telemetry...

>>58939803
>Very. A whole new market where both Intel and NVIDIA have no tools to compete.
They need to just shove 8 ryzen cores + vega on an interposer with 32gb of HBM. With SSDs you could make an entire PC not much larger than a video card.

>>58940005
Kaby Lake has some hardware bugs fixed from Skylake, performance per clock is pretty much exactly the same. Some benches show a drifting 1%~ change either way, but this is margin of error.
The chip doesn't really exist for the desktop market, it was just brought there because why not. The biggest selling points were the new video decoder, and higher frequency. There is a huge caveat on that second point. Kaby Lake was designed for the mobile market. The process improvements intel touted are not at clock speeds of 3.5ghz or higher, they're at lower frequencies. At 5w per core Kaby Lake will clock higher than Skylake or Broadwell, and that makes it a significant enough uplift for a laptop part.

Its been amply tested and shown now that at desktop frequencies Kaby Lake is pulling more power than Skylake. Overclock a Kaby and Skylake i7 to 4.5ghz and the Kaby Lake chip will be pulling more power.
This is the direct result of the process being tuned to favor lower frequencies. The sweet spot in clocks/volt is lower on the curve than it was before.

At 4.2ghz a single Kaby Lake core is nominally pulling 21.5w under load, and power consumption jumps radically with each 100mhz higher you push it.
At 3.6ghz they're only at 12.5w.

Kaby Lake is a laptop processor brought to the desktop market as a gap filler.

>>58940005
Not only that, but they cheaped out on the internal thermal paste shit. People are actually delidding their processors again.

File: 1445388012430.gif (2MB, 389x279px) Image search: [Google]

2MB, 389x279px

>>58940110
Again? Fucking kek

>>58940110
>tfw Ivy Bridge

File: Intel 1.png (186KB, 1920x974px) Image search: [Google]

186KB, 1920x974px

>a-amd can't c-c-ompete

>>58940158
Fuck, even Intel's engineer admitted Zen is clearly competitive.

I thought u learned from Poolaris to never ever link wccftech tier articles again.

>>58937702
decreasing power draw in a mainstream market that hasn't needed more than they offer. amd's solution is more power draw and more cores for everything. too bad this doesn't add up to anything for 90% of computer users.

>>58940247
>Ryzen
>more power draw
Wrong

>>58939325
It's more a question of 'what exactly is holding intel back'. Consider what their main focus has been over the years since the introduction of the core i series. They've been most focused on breaking into the mobile market (which has been a hilarious failure and cost them tens of millions at least, probably hundreds) and pushing for increasingly smaller nodes on from their foundries. Both of which have proven significantly more troublesome than anticipated.

Meanwhile focus on the architecture side of things has been more about getting something functional on their new nodes than actually making massive functional improvements. Chances are they've cut down on engineers working on desktop architecture massively to make room for those working on mobile and production.

Meanwhile AMD has had 6+ years to work on Zen with their only other real focus being GPU's, which they've streamlined massively with GCN, and smaller scale stuff like APU's. They don't have a CEO promising industy leading node shrinks and mobile performance in unrealistic time spans. In this case it's a matter of misguided goals set by a CEO confident relative performance wasn't as important.

The other consideration is maybe we're nearing the limitation of x86 as an instuction set. Keep in mind x86 was defined decades ago and the most significant change in recent history was the adaptation to 64bit, which was the product of AMD.

tl;dr intel being lazy and stupid doesn't mean someone else can't do it better.

>>58940291
What happens if we do hit a limit on x86? Anything before that is a bit before my time.

>>58940291
Intel spent over $3 billion on their contrarevenue program competing against ARM parts in the smartphone market.

>>58940306
>What happens if we do hit a limit on x86?
Newer ISA and emulation.

File: 1483872520027.jpg (51KB, 900x900px) Image search: [Google]

51KB, 900x900px

>>58940321
>emulation
What about it?

>>58940363
The idea is that a new ISA that replaced X86 would have hardware emulation of X86 ops so it couldn't grossly sacrifice performance or totally break compatibility with legacy software.

>>58940306
Something that takes better advantage of computing as it is today. Consider the following:
ARM is significantly more suited to mobile than x86 could ever hope to be, or at least than intel could ever hope to make it. Despite their undeniable presence in the design and production of microprocessors and ages of corporate experience, they just can't match Qualcomm's SOC's.

Also similar although for technically different reasons, we've seen a shift in the GPU market towards low level API's.
API's are an abstraction layer well above an instruction set, but it's the fact that how we render 3d models and animations has changed so much since DX and openGL were introduced. We've reached the point where GPU's are held back by how they're told to things and the best way to overcome it is to rethink what they do and how they do it. The results of which are not JUST mantel, DX12, and Vulkan, but CUDA as well, yet another thorn in the side of intel.

So yes, if x86 proves to be the bottleneck, we'll have to return to the drawingboard in many ways, but in no way does it mean we've reached the limitations of what we can achieve with the silicon transistor.

>>58939621
I thought Raven Ridge flagship was supposed to be 4/8 zen and vega hbm? igpu

>>58940437
No confirmation of it featuring HBM yet, but it would need it.

>>58938616
>3k rig
That's only the case because of the monopoly Intel and Nvidia have on the market.
The desktop market will flourish once more the moment AMD actually competes in CPUs again, and somehow doesn't fuck up with Vega.

Also, the only thing that's stagnating in the market are prebuilds, because more and more people are building their PCs by themselves or asking shops to build it for them.
The only prebuilds that are still selling like hotcake are HTPC sized prebuilds.

>>58940639
Raja better get his shit together, I want red dudes to be great at high-end again.

File: 68708461.jpg (15KB, 251x206px) Image search: [Google]

15KB, 251x206px

Am i the only one who literally has no need for new PC but wants to build a new one with Ryzen?

>>58940738
Yes. KYS.

>>58935966
Uhh.

I'm not intelligent.
Someone tell me if OP is taking a piss

>>58940738
I know your feel, currently have 4790K and Fury X.
I'm changing my GPU with whatever Vega shits out, if it's worth it.
CPU on the other hand, have no complains about it...

Uh oh.

Is this for real this time, please someone tell me.
AMD has way too often done these "we gunna beat them gunny" and failed

>>58936705
So are you telling me

a 1800 USD Intel CPU
is getting beaten/nearly beaten by a 400 AYMD processor?

>go to look up the 6950X
>intel have changed their ARK site

I don't give a shit about AMD but intel changing the ARK site layout is a fucking outrage.

>>58940787
>tfw man with shovel came out and all the buttblasted idiots that were posting hype threads for months before launch

bide your time.

>>58940838
New ARK is literally unredable.

>>58935966
[x] doubt

Just wait until after it's out and look up an average user benchmarking it.

>>58936383
Toplel

>>58940416
This is very insightful.

Originally computers were custom built for application, as manufacture and such was just too costly to sell to everyone. As tech developed so did the idea of general purpose computing. While it was actually less effective then older versions, the economies of scale and cross market support meant it was vastly cheaper to just buy more then one computer to make up the difference. With this new market model the power of computers grew in nearly every way, but most of that was new manufacturing technologies, we all know how important shrinking the die was and given the small number of dies effort to shrink were focused.

But now we have hit a number of nasty limitations in the hardware, many of which I don't think can be directly solved. So we see more and more high level software solutions as short cuts to get what we want given the years of glut processing we have wasted.

But functionally such a market has become a stagnate commodity model.

If we want things to change, we need to change the market. Dropping x86 for better alternatives would unlock huge potential gains, as our understanding have grown a lot since we started, but the cost to redevelop the market would be mind boggling expensive, in a market that already costs far more then people are willing to pay.

How do you get people pay for the next computer revolution, when the cost is so high and few see direct gains in their lifetime? It is like justifying the space race before the end of WWII, retrospectively everyone loves it but nobody wants to pay to build it.

>>58936328
Intel shills on suicide watch

>>58935966
dlet this

>>58938309
I'm really glad you use a tag, I know which positions I should oppose since you're a fucking serial shill

>>58938614
>i5
why.png

>>58940824

a 1800 USD Intel CPU
is getting beaten/nearly beaten by a 400 AYMD processor?

Yeah, processors aren't Intels priority anymore

when are they releasing

>>58937146
what memory do we get for it?

>>58941558
Only the best
http://www.gskill.com/en/product/f4-3600c15d-16gtz

>>58938415
Pentium 3 is based on the Pentium Pro
Intel did an arch from scratch with Itanium and Pentium 4 which were massive failures
So their successful arch is an iteration of a 1992 CPU
>>58939353
Larrabee was squarely aimed to become a GPU, being ridiculed in SIGGRAPH and never becoming competitive led Intel to be able to salvage it into a coprocessor/GPGPU
>>58938428
This ignores the i740 and the entire GMA line and their graphic efforts until HD3000

>>58938509
But a blooming server market thanks to le cloud
Which is squarely what Zen aims for

File: anzujaamu.jpg (70KB, 500x744px) Image search: [Google]

70KB, 500x744px

>>58940052
>windows 10 spies on me but windows 7 doesnt even though they're made by the same company
if you care about privacy, go GNU/Linux.

certified
shit
wrecker

File: 1346451836483.jpg (175KB, 462x435px) Image search: [Google]

175KB, 462x435px

>>58935966
>per-core performance on par with Broadwell/Skylake
>more cores
>way cheaper
>lower power usage
>all unlocked
>cheap B350 motherboards for overclocking

>>58935966
1700x raping 6900k for 1/3 of the price.
What a glorious day to live is today.

>>58942419
Ain't it good when the competition is back?

>>58935966
I remember seeing these kinds of graphs with bulldozer destroying intel cpus. Ill believe it when release comes around.

>>58942446
There never were any graphs where a 4c/4t bulldozer destroyed a 4c/4t Intel. Never fucking ever.
The only time BD "destroyed" Intel on graphs was in highly threaded workloads and 8c/8t vs 4c/4t and maybe 4c/8t.

These graphs however show AMD performing nicely at the same core and thread count.

File: KdbkN7Q6GH2igMvrUjauFU-650-80.png (67KB, 601x801px) Image search: [Google]

67KB, 601x801px

http://www.pcgamer.com/new-amd-ryzen-details-and-pricing-leaks/#
Writer for PC Gamer compiled a Passmark chart using results from his own systems. He did test systems at stock clocks, then pinned everything to 3.4ghz, fixed frequency for comparison.
Physics, Prime Numbers, and the CPU Mark scores are impacted by the memory used in the system, everything else looks stellar.

Integer performance is ungodly high.
Even float point performance is ridiculously high. I'd never have expected AMD to design a core with such high FPU throughput.
Single threaded performance is competitive per clock with intel offerings.
Encryption, Sorting, and Compression metrics show very good for Zen as well.

File: csw.jpg (27KB, 570x340px) Image search: [Google]

27KB, 570x340px

>>58942519
>Integer performance is ungodly high.
>Even float point performance is ridiculously high. I'd never have expected AMD to design a core with such high FPU throughput.
>Single threaded performance is competitive per clock with intel offerings.
>Encryption, Sorting, and Compression metrics show very good for Zen as well.
You did what, you doubted Keller? What blasphemy.

File: 1481670688249.png (418KB, 627x627px) Image search: [Google]

418KB, 627x627px

Jim motherfucking 'certified shitwrecker' Keller

File: xwSbTJUX3L4rmH2QQanu3U-650-80.png (70KB, 601x801px) Image search: [Google]

70KB, 601x801px

>>58942519
int perf

>>58942538
Hes basically a certified hero at this point. He doesn't fail. Ridiculous.

>>58942558
>int perf
What the ever living fuck.

File: R9eKJn9rVmxVEyuBsj2i4U-650-80.png (69KB, 601x801px) Image search: [Google]

69KB, 601x801px

>>58942558
fpu perf

File: e563XbpqhcSY5W2HLgEg3U-650-80.png (69KB, 601x801px) Image search: [Google]

69KB, 601x801px

>>58942566
CPU Mark score

>>58939362
I'm waiting really patiently for ryzen+vega APU laptops

File: UorgMK6TqvTEx5i7kGuJBU-650-80.png (69KB, 601x801px) Image search: [Google]

69KB, 601x801px

>>58942587
encryption

File: iMAEGeZbFKHiCyv8JR9w5U-650-80.png (69KB, 601x801px) Image search: [Google]

69KB, 601x801px

>>58942599
compression

>>58942558
Holy shieeeeeett. Intel's best offering is being blown the fuck out of water every second! This is incredible.

File: ctwiTKEyNqWQXfQs4BnZFU-650-80.png (68KB, 601x801px) Image search: [Google]

68KB, 601x801px

>>58942605
sorting

>>58942608
Broadwell-E would be somewhat higher, but these test results are all coming from one guy so they're a solid baseline vs random submissions.

File: JXNqzW7QXXQMyqKnKBziBU-650-80.png (66KB, 601x801px) Image search: [Google]

66KB, 601x801px

>>58942614
physics

>>58942614
I'd suggest you compile them all into one image and make a new post. This is brand new information to me and a lot of people. Intelfags need to know this. There's still some time and hope for them.

>>58942599
>>58942587
>>58942566
>>58942558
>>58942519
sir delete this please sir sir delete sir this is test material not for public sir please delete

File: 9MGsVSuStcGjbbWDradh5U-650-80.png (66KB, 601x801px) Image search: [Google]

66KB, 601x801px

>>58942630
and lastly prime numbers

>>58942631
Maybe I'll do it at some point later. I don't want to contribute to the mass spamming of threads all on the same topic.

>>58942558
>>58942566
>>58942587
>>58942599
>>58942605
>>58942614
>>58942630
>>58942645
Y-YAMERO YOU'RE ATTEMPTING HOLOCAUST AGAIN

>>58942645
how come this one is so bad compared to the others

>>58942668
It's using leper-tier RAM and it impacts 2 tests pretty badly.

File: 9eff5f0750abd8d6a8b6220b9ef92e47.png (305KB, 799x634px) Image search: [Google]

305KB, 799x634px

>>58942519
Jim "Putting Intel to the bin" Keller.

>>58942645
>>58942630
>>58942587
>>58942668
These three test metrics are sensitive to memory bandwidth, and latency particularly. A bunch of forum users across S|A, Anandtech, and OCN have amply confirmed it. The Physics and Prime Numbers scores more so, but the aggregate CPU Mark score varies to a substantial degree as well based on memory performance.

The Ryzen test system submitted to Passmark had a 2400mhz DDR4 kit with 17-17-17-39 2N timings. Passmark reported memory latency at 76ns if memory serves.
Individual
Those CAS timings work out to be 14.161ns-14.161ns-14.161ns-32.487ns. Average timings would yield latency for the first word value around 10ns, really fast RAM is closer to 8ns. Thats a pretty big latency hit for anything memory intensive.

>>58942706
so does this just mean intel computers had better ram or what? why would ryzen systems not use faster ram?

This "3.4ghz Ryzen" seems to be top notch CPU.

>>58942727
Whoever was testing the system was likely trying to do a worst case scenario to gauge performance. They were also using a low end OEM motherboard.
For a 2400mhz kit to have timings that high they were probably set there intentionally.

>>58942727
No, it means they used shitty DIMMs for benches.

>>58940975
The push won't stem from the consumer market, not at all. It will come from enterprise, server, machine learning and the like.

Long, long before superior semiconductor alternatives like graphine become a realistic option we're bound to see a desire by certain markets for more processing power. At this moment, x86 is still more or less the best option for a lot of high level computing, regardless of its general purpose origin.

It's very possible the consumer market won't grow much beyond x86 any time soon, but it's also possible the demand on the ultra high level will result in these changes trickling down and eventually proving to be the better option.

Or not. Fuck man who knows. All I know is AMD has outclassed intel in terms of design on every level in the past and if Zen+ delivers, they may just do it again.

File: AAAAAA1385779913511[1].gif (3MB, 359x346px) Image search: [Google]

3MB, 359x346px

>>58937582
>releasing another skylake-ish cpu with barely any improvement.

Oy vey the goy will buy it anyway

File: JIM.jpg (36KB, 400x533px) Image search: [Google]

36KB, 400x533px

>>58942558
>>58942566
>>58942587
>>58942599
>>58942605
>>58942614
>>58942630
>>58942645
SAY MY NAME

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

13KB, 381x286px

>>58942789
CERTIFIED

File: 1486363295363.jpg (169KB, 1500x1000px) Image search: [Google]

169KB, 1500x1000px

>tfw don't know jack shit about CPUs and just here for the drama

So, who's winning?

>>58942833
The consumer

>>58942833
the jews
the jews always win

>>58942833
Probably you. You'd like cheapo high performance CPUs?

>>58938280
>buying unnecessary upgrades
good goyim, now you see how Intel have fucked you

>>58942845
No I'd much rather overpay so long as the CPU is American.

>>58938280
You flushed a couple hundred down the drain.

>>58942519
>faster per clock than i7-7700K
Only question is whether it'll clock as high...

>>58942909
>retard agaist free market
>>58942914
No, it won't, it's an 8core part.

>>58942914
It can hit 4ghz on all 8 cores at least. 5ghz on 8 cores is probably not happening, but its possible, just with insane power consumption.
Top binned chips are hitting 3.6ghz at 10w per core, which is really good. There are no other high clocked 14nm LPP parts out there to see how it scales though.

Anyway, Broadwell-E would be a good example for the range of power consumption, and heat output you'd be looking at for heavily overclocked 8 core Ryzen.

File: 1422894169116.png (7KB, 801x577px) Image search: [Google]

7KB, 801x577px

>all those benchmarks noone will be using this for
Just give us the gayman benches already

File: 1395656178807.gif (3MB, 300x225px) Image search: [Google]

3MB, 300x225px

>mfw waiting for Navi and Zen+ and building a 4k 120fps gaming machine that's the size of a hardcover book

>>58935966
Is the era of intel jewing finally over?

File: 1437233573075.png (138KB, 457x645px) Image search: [Google]

138KB, 457x645px

>>58942961
If all you care about is MUH GAYMES, CPUs haven't been relevant outside of 1 or 2 games for years. Now fuck off back to >>>>>>>/v/, fgt.

>>58942909
Intel currently has no fabs in the US.
AMD sources parts from Global Foundries, which operate 3 fabs in the US.

>>58942630
>>58942645
Both of these tests supposedly suffered due to crappy RAM being used (2400 MHz with very loose timings). Hopefully official benchmarks will show results not quite as bad as this.

All of the other benchmarks should have Intel quaking in their boots though.

>>58938614
02/03/2017

Any release date for 6 cores?
Or is it same date with 8 cores?

>>58939325
>what does this mean for the future?
next 2 years? tech wise nothing, 2019 is the year when real competition begins if one of companies will not go down in flames in between

>>58943111
4/6/8 core parts should all be available from the start. Complete unveiling should be in 2 weeks, with full release a week later.

>>58943073
>No fabs in the US
Intel has 6 fabs in the US, 75% of all their fabrication is done in the US.

>>58943073
CPU parts come from NewYork fab.

Don't intel has most of their fabs in US due to nature of "anti trust laws or whatever" there?

>>58943367
They do.

File: Untitled.png (19KB, 581x260px) Image search: [Google]

19KB, 581x260px

Nobody posted these yet?

>>58943414
That's been out for 3 days

File: 1471396486391.png (182KB, 578x691px) Image search: [Google]

182KB, 578x691px

>>58940738
Xeon 2670v3 ES here
Grabbing top binned Ryzen for single thread perf.

The only thing I want to know is if Intel CPU prices will go down as a result.

>>58943438
No.

>>58943438
No, they never do.

>>58943429
>AMD
>single thread perf

lel

>>58943480
read the thread.
too much NOT conflicting,unlike 480, information coming on this
it's pretty much confirmed at this point it is skylake equal +-3%

>>58943158
Hope more MATX board announce when it release. Especially MSI.

INTEL IS FINISHED

>>58935966
>http://www.3dmark.com/fs/11633541
>i5 getting fucking stomped by the 6 core
there goes intels profit margins i guess

With Jim "PD" Keller gone, will there still be a zen+?
get bored and decide to go make something awesome for tesla? was he tired of Lisa's Poo stinking up the Loo? Next at 9

>>58935966
>i7 6950X @ 3.0Ghz
>@ 3.0Ghz
Nice FUD

>>58937114
>The under-the-hood changes are really amazing
They didn't change a single thing "under-the-hood"

>>58939172
>same price
>same features
>AMD has cheaper overclocking boards
>AMD has 4 actual cores
Why would you ever want an i3 after this point?

>>58939362
The APUs won't ship with Polaris attached, they'll be shipped with Vega, which is expected to be much more efficient.

>>58944620
Because the i3 can overclock much higher

>>58944620
meant less appealing, But i'm assuming that that intel is higher ipc and clocks higher.

also, you can get fairly cheap intel boards, cheap enough that the difference between amd and intel is kind of out the window at those price points.

>>58944672
And yet, still won't reach the multicore performance of a true-quad, and might not hit that much above in single core.
You would need a 30%+ clock speed difference with the Ryzen chip to match it, and we're already seeing them hit 4Ghz. The 4-core also has a lot better power delivery in the core and dramatically more dead silicon to dissipate heat over.

Hyperthreading isn't a magic bullet for multicore. It's at best 30% as long as the cores aren't already being hammered to their fullest.

File: 19305564.jpg (183KB, 1300x1497px) Image search: [Google]

183KB, 1300x1497px

>>58943429
Great! I feel like selling my full rig and getting Zen + Vega!

>>58940738
Sort of. I wish I could encode faster, run multiple VMs, and have all virtualization technology on my chip. Oh wait, Ryzen is what I want.

>>58944672
The only overclockable i3 is ridiculously expensive. Even a locked i5 is better value.

Face it, i3 is dead.

File: Ho-Holy SHIT.png (213KB, 592x589px) Image search: [Google]

213KB, 592x589px

>>58942558
WHAT THE FUUUUUUUUUUUCK

File: 1403978922713.gif (1MB, 300x254px) Image search: [Google]

1MB, 300x254px

>>58944906
2017 is the year of AMD. Intel's monopoly and price gouging is finally over.

>>58944899
The i5 line as been worse value over time, with it's gimped clock speeds for the lower cost ones and the k version being way to expensive regardless.

Then the new pentium with HT has made the entire i3 lineup useless and Ryzen is coming along to undermine everything i5 and above.

>>58944980
This is true
I built a Skylake i5 machine this summer. I really wish I held out just another 8 months