wccftech.com
wccftech.com
>4.8GHz is achievable on all cores
>4.4GHz PPC equal to a 5GHz 9900K
>All-core 5GHz is doable, but will come later in a form of "special edition" hand-picked golden sample offering (this is NOT same as a 16 core Zen 2 mainstream variant, but a separate thing in on itself)
>1.35V for all-core 4.5GHz is guaranteed even on B tier 4xx motherboards (nothing was said about 3xx)
Other urls found in this thread:
amazon.com
amazon.ca
amazon.co.uk
amazon.com.au
amazon.de
youtube.com
youtube.com
youtube.com
hothardware.com
twitter.com
intel is fucking finished (and that's a good thing)
curry tech news has been wrong about most zen 2 related stuff. so this probably a lie as well. but like all good lies, a little bit of truth. the only truth in curry tech is most likely the 4.4ghz being the same as a 5ghz intel. and thats evident enough by amd's own press release at comp. so it doesn't take a leak to figure that out.
greetings Jow Forumsentoomen. its time to preform some armchair level of performance theorycrafting with zen 2. to understand zen 2 performance we must first look at zen 1. AMD stated an average 15% increase in IPC over zen 1 and this is crucially important to figure out zen 2's overall performance. so first, lets compare zen 1 vs zen 1.5 refresh and vs intel 9th generation with the 9900k.
zen 1.5 refresh brought:
>3% increase in IPC
>200-300mhz increase in clock frequency
>10% average overall increase in performance
all according to AMD themselves. so to go off this lets compare the 2700x vs the 1800x:
>1800x - 3.6ghz base / 3.7ghz XFR all core clock
>4.0ghz max boost / 4.1ghz max XFR boost
sustained boost for the 1800x is 3.7ghz XFR. 4.1ghz is only reserved for SHORT, only a few seconds, bursts in LOW load usage. followed by a quick drop down to 3.7ghz sustained. so for something like cinebench, a 1800x will operate at 3.7ghz in the single threaded bench.
>2700x - 3.7ghz base / 3.8ghz XFR all core clock - 4.3ghz max boost - 4.4ghz max XFR boost
with PBR 2 / XFR 2 the behavior of both changed for zen refresh. they were modified to behave something more similar to intel's turbo boost. rather changing in steps, it changes in a more gradual slope. max boost frequencies are still reserved for short bursts in low load usage, but as it goes down, it goes down gradually before staying at a frequency it can sustain that are within its boost parameters. thread load, power consumption, silicon binning, voltage, and heat. this results in an average XFR all core frequency of around 4.0 - 4.1ghz in single thread. 3.8 - 4.0ghz in multithread (depending on severity of the load)
so now lets take a look at cinebench single threaded results:
>pic related
cinebench results from techspot.
>2700x - 178 score
>1800x - 159 score
a difference of 3%. obviously this greater than the 3% increase in IPC AMD claimed. so what caused the increase? obviously, clock frequency. if we use an average of 4ghz for the 2700x, this gives us a 300mhz increase in frequency over the 1800x's 3.7ghz. this 300mhz increase falls in line with amd's claim for zen refresh. this 300mhz increase results in being a 9% increase over the 1800x frequency. if we do armchair math, 9% + 3% = 12%. this falls directly in line with the increase the 2700x has over the 1800x in cinebench. 3% came from IPC alone, while 9% came from clock frequency.
we can actually help prove this by looking at the r5 2600 vs the 1800x.
>2600 = 163
>1800x = 159
>3% difference
the r5 2600 lacks XFR and its max PBR boost is 3.9ghz. considering 3.9ghz is not sustained all core, this gives the average all core in single threaded scenarios being around 3.7ghz with PBR. a 300mhz over its base clock of 3.4ghz. we know the 1800x has to be running at 3.7ghz and the 2600 must likely be running at 3.7ghz as well because the difference between the 2600 score and 1800x score is -3%-. this gives us a near apples to apples comparison where the increase with the 2600 comes solely from IPC improvement alone. so we now have a pretty good idea where the performance increase between zen refresh compared to ryzen 1 came from. with the 2700x we know 9% came from clock frequency while the remaining 3% came from IPC. both combine gave it a solid 12% increase. not to shabby. what's important? this 12% falls in line with amd's claim of an average of around 10% increase in performance over zen 1 with zen refresh.
so now lets compare AMD's claim of 15% increase in IPC over zen with zen 2. if we do some armchair math and add +15% to the 2700x score in cinebench we get:
>178 + 15% = 204
lets compare to the 9900k:
>216
.
There's nothing good when one of the competitors is finished because guess fucking what: it ruin the competition
This article is a mess.
that now puts the 2700x 6% behind the 9900k. that's not to shabby considering the 9900k single thread clock frequency is 5ghz. that would be a full 1ghz difference between it and the 2700x with its 4.0ghz frequency. if we compare to the 8700k that actually puts the 2700x AHEAD as the 8700k scored 198. an increase by 3%. the 8700k operating at around 4.7ghz in single threaded loads so a 700mhz frequency increase over the 2700x, a 18% increase in frequency, only to lose to the theoretical 2700x by 3%. so what does this mean? clock for clock, AMD now holds the single threaded crown over intel.
So now lets figure out the clock frequency needed to be 3% ahead of the 9900k in single threaded performance. This is the 3% claim AMD stated for the 3800x vs the 9900k in AMD's cinebench single threaded result.
So lets add another 300mhz, 8% increase, to the 2700x 4.0ghz to bring it to 4.3ghz. first though lets try to find a baseline for the average increase in cinebench single threaded score done by increase in clock frequency with processors based on the same generation.
i7 8700k vs i9 9900k:
> 4.7ghz vs 5.0ghz = 7% difference
>198 vs 216 = 9% difference
a 7% increase in frequency gave a 9% increase in score
r5 2600 vs r7 1800x:
>3.7ghz vs 4.0ghz = 9% difference
>163 vs 178 = 9% difference
9% gave a 9% increase
i7 7700k vs i7 8700k:
>4.5ghz vs 4.7ghz = 5% difference
>192 vs 198 = 3% difference
5% gave 3% increase
i7 8700k vs 9700k
>4.7ghz vs 4.9ghz = 5%
>198 vs 214 = 8%
5% gave a 8% increase
r5 2600x vs r7 2700x:
>3.9ghz vs 4.0ghz = 3% difference
>173 vs 178 = 3% difference
3% increase frequency for 3% increase perfomrance. i averaged 3.9ghz with the 2600x because its specs are 100mhz less than the 2700x. so i just subtracted 100mhz off. so overall we see a % increase in frequency either gave same increase in perfomrance or an increase by 2% - 3% over the frequency increase percentage. with the exception of the 7700k vs the 8700k which only saw a 3% increase in score compared to its 5% increase in frequency. so if we calculate an armchair average increase lets go with increase in frequency = increase in score.
going back to the hypotehtical 2700x, if we add 8%, the increase percentage in its frequency to its score, we go from 204 to 220 score. 216 vs 220 = 2%. if we up the percentage increase to 9%, which still offers 4.3ghz at now 4.36ghz, we get a score of 222. a 3% increase over the 9900k 216 score.
Now we know the 3800x specs a 4.5ghz turbo boost. but amd really didn't go into detail much about XFR with zen 2. if XFR behaves the same with zen 2 as it does with zen refresh with XFR 2, then we know max turbo boost isn't sustained frequency. so more than likely, its actual sustainable boost, even for single threaded, is lower than its rated boost speed. as its max boost is just its burst boost. so from this we can conclude an estiamte that amd's demo 3800x in cinebench was most likely running around the 4.3ghz range. That means, AMD even with a 700mhz frequency deficit, still managed to come out ahead of intel at 5ghz by 3%. That means, hypothetically from armchair math, that a 4.2ghz zen 2 would match a 5ghz intel coffee lake core. And intel would need to be 4.8ghz to match a zen 2 core at 4ghz.
So now the question is, clock for clock, how far behind is intel vs zen 2? well we know the 7700k at 4.5ghz recieves a cinebench single threaded score of 194. the 8600k at 4.3ghz single gives 179. hold on for a second, the 2700x recieved a score of 178 at 4ghz. 300mhz increase and all intel recieved was 1 point higher. that means skylake architecture was already behind zen refresh in IPC, at least in cinebench. if we look at the i5 8400 we see at 4ghz it scored 168. 10 points behind the 4ghz 2700x, and only 5 points higher than the 2600 at 3.7ghz. that means, at 4ghz for both the 2700x and the 8400, clock for clock, zen refresh, at least in cinebench, has a lead of 8% in IPC. with zen 2 offering 15% over zen 1, that means zen 2 is at least 23% faster than coffee lake. at least in cinebench.
now i know some might be thinking how can that be so. zen was losing in games and other single threaded bound titles, horribly. well, its because zen does have a bottleneck. latency. in latency sensitive scenerios zen suffered. pair up that latency bottleneck with lower frequencies than intel amd suffered greatly because of it. but with zen 2 now upper the frequencies well in the 4ghz range with improvments to its latency bottlenecks makes zen a brute force to be wrecken with. intel is in trouble and right now they're best saving grace is their high clocks because those high clocks in the upper 4ghz to low 5ghz rnage is whats going to allow intel to match amd at lower clocks. but no more suprassing amd, and no more being the gaming king.
>wccftech
Constanza.jpg
Literally the National Enquirer of the tech world.
I only play games, in that realm AMD still is and will always be irrelevant.
>you lived long enough to see Jow Forums make currytech threads
see
But all those posts are in some kind of monkey negrum pidgin speak. "force to be wrecken with" indeed. By the way, you are obese.
reckon, you happy now, nigger.
Its good. Maybe we will get non Jewish competition. Jews dont deserve business of any kind
There's Qualcomm, Loongson, VIA, MCST, and a fuckton other big players already emerged on the market. No Inturd? Good-fucking-riddance, I say.
It'll force Intel to ditch 10nm and push 7nm (which will be comparable to TSMC's 5nm).
>longsoon
Fucking kek. It's not enough that you named arm architectures, you named one of the biggest pieces of shit in existence.
>knows nothing about CPU architecture
>insists that pathetically weak processors are competitive against x86
You couldn't look dumber if you tried.
daily reminder: intel and amd threads are posted by marketing kikes and chongs representing these companies
oh fuck off you marxist cunt.
you need to loose some weight, marketing lard.
Oh sweety, hillary lost. get over it
Can you change Intel to incel? PNG is out there.
I'm too lazy. Here you go:
im running a 8086K ultra edition from the der 8auer
seeing threads like this
never reached 50°C
feeling good
you sound like you didnt get one
Got'em
Kys, unironically.
8086k clocks from 4Ghz to 5Ghz
mine to 5,2
stay mad poorfag
AMD monopoly > Intel monopoly
1.08 x 1.15 = 1.242
Thats how you should calculate.
This but unironically
*reckoned
THANK YOU BASED AMD INTEL POZZED HOUSEFIRES BTFO
amazon.com
amazon.ca
amazon.co.uk
For fucks sake, you fa/g/got. Stop posting links all the time and start posting this instead:
That image is old. frenchfags are sucking Intel dick now
dude, thats just an shitpost. if fucking AMD survived goddamn Bulldozer then Intel will have no fucking problem. yes, they will bleed marketshare and money like crazy but for the benefit of us all: more cores are mainstream now, 6c/12t is basically toaster tier shit. we can't lose, server market can't lose
Well, the thing is that Intel's situation is different from what AMD's was in that Intel is, at its core, more of a semiconductor fabrication company than a chip design company, and so if they cannot unfuck 10nm and miss on 7nm they can become GloFo but worse very quickly because they have no other customers and the operating costs of their fabs are enormous.
This, a big part of Intel's revenue goes into R&D 10nm and shit, and they STILL seem to struggle really hard with that node while TSMC and Samsung are now officially ahead
>more of a semiconductor fabrication company than a chip design company
thats true man. but remember that AMD bled money for those cocksuckers GF too, for no good fucking reason actually. maybe bad management which all of us would totally agree
>if they cannot unfuck 10nm and miss on 7nm
original 10nm is dead. the (((new 10nm))) isn't any better, just look at the initial Ice Lake offerings for laptops/mobile: 4c/8t (do they actually use HT now?) at 1.1Ghz base 4Ghz boost? ROFLMAO what a scam, everyone is laughing at Intel right now. 18% IPC improvement claimed, tested with fucking 3733Mhz RAM without mitigation patches applied vs old ass Skylake.
>big part of Intel's revenue goes into R&D 10nm
i unironically believe that this so called 10nm process is just a blunder, not very different that 14nm++++++. just for the press to report and to show shareholders that the obvious dumbsterfire is not a dumbsterfire. Intel need to switch to 7nm right now, leave all the old shit behind. 10nm on paper looked amazing but they done goofed, no on takes them seriously.
that shows, that all accross gamers are buying the 8700K. its the best for gaming
The image is exact representative of the links.
>all gamers
>buy Inturd
Here's a good chunk of harsh reality for you:
>How
back to lauratard furfag
AMD claimed 15% over Zen+/Ryzen 2000, not Zen 1
Aww, c'mon! How do I always get recognised on here? Jow Forums is one of the boards I post least on.
Anyway, /tkg/ told me to go away. Need somewhere to post Lauras.
Mostly because everybody's waiting for ryzen 3k to release, just check any hardware forum / website
Source: i'm a frenchfag
>the the 8auer
here you go user
The thing is, intel will respond to it eventually, it's how intel work.
Instead of having this smooth linear improvement scale, they just do the fastest chip they can, then they fucking sit on it and do nothing until someone REALLY kick their asses.
Then they go back to the drawing table and create the best chip on the market again... and sit on their asses, waiting for the next kick.
So what will happen now is that they will try everything to not have to innovate. They will overclock their chips to the maximum, they will even do moar cores, but when everything fail, they will cry, do a tantrum and finally fucking come up with a brand new architecture that will BTFO the Zen 2.
(((((architecture)))))
Time for stronger botnet.
>mos as fast as 8088 at less than half the clock
intel is finished and bankrupt
The 6502 was also less than half the price of the 8088.
Only thing about those benchmarks. I have always had a suspicion they do AMD out of the box, but OC 9900k, 9700k etc as it is generally accepted the K chips will be OC. Sometimes you see a 4.2ghz OC AMD bench, but even that falls behind my numbers.
I have had closer to 9900k scores close to 2000 (9700k does not get that close) on cinebench purely messing with ram timings, turning on PBO, undervolting 0.050 and cooling with an NH-D15s (with 120mm on front). This is my personal experience, yet I see "tech reviewers" get far less results with apparently more knowledge. There are guys liquid cooling surpassing that.
This means to me, whatever they are spouting about the 3000 series, it has a little more potential in the real world, because these idiots aren't using the best settings. Yes I know we are talking single thread, but it goes to show, if those benchmarks are often off by up to 200, what else do they have going wrong?
DELID
I
S
Not much
Stock 8700k is as fast as the *new* ryzen 3000
COPE I3
COPE I5
COPE I7
LOLOL
You mean it's just as fast at lower clocks? Damn AMD is fucking based, best IPC
The 8700k has a baseclock of only 3.7 GHz
If i recall the *new* ryzen 3000 are above that
Huh? What's going on?
8700k is OC'd to tell
You can add another 20% on that and you will see the 8700k oc results
8700k boosts to 4.7 on its own, the fact that Ryzen matches it at lower clocks is a testament to Ryzen's superiority.
Are you retarded? Not even the all core boost is 3.7, it's like 4.4 or something. 5ghz would give it a nice 15% at the most in all core workloads.
Based on the 2700x the new ryzen should be able to get about a 10% OC over its all core boost too
I'm Laura user, I've got a folder full of Lauras and I like to edit/post them (usually where relevant).
is just a Laura hater, probably remembers me from my Laurafox threads or the time I was talking about the Laurem.
>Laura
Who/what?
It depends on whose writing. Usman accurately brought up AMD trying to replace GCN over a year ago, among a whole essay of accurate navi details
Laura (of the Coastal Fox Tribe) is a character from the Twokinds webcomic. She's an autistic fox girl with a stutter that causes calamity and heartbreak due to her being manipulated/making poor decisions.
There's a general on /trash/ dedicated to the comic here: (don't go there though, they're mean now and turned against Laura).
Anyway, the Twokinds general memed her into a Laurem (Laura Harem). Lots of art and stuff ensued.
AMD never used jewish trickery like Intel.
If you don't know about the three decades of Intel scams, unfriendly competition, payd "unbiased" reviews, false advertisement, blatant lies and coverups you don't deserve to post in this board.
Black people, just like Intel processors, run faster but they're way too pozzed. Ryzen is a white man, valuing security, balance, efficiency and intelligence.
1 Zen core = 2.5 Intel cores.
hothardware.com
INTEL POZZED HOUSEFIRES BTFO
Now put 3200mhz ram on an 8400. Nice gimpmark