12-core Zen2 ES spotted in UserBenchmark

twitter.com/TUM_APISAK/status/1088264161191022593

userbenchmark.com/UserRun/14076820

INTELFAGS ON SUICIDE WATCH

Attached: ayy.png (1277x203, 54K)

Other urls found in this thread:

userbenchmark.com/UserRun/14085581
twitter.com/NSFWRedditGif

>3.4 ghz
the absolute STATE

Also the latency is through the roof lul
Zen 2 confirmed a huge fucking flop

Weren't there 6700K ES with a base clock of like 3.0 GHz?

No! ES clocks are final! AMD is d-dead!

I sincerely cannot tell if this is misinformation by Intel, a bot or a combination of both. Every single zen 2 thread there's this same motherfucker saying the same thing. Without fail.

DELID THIS RIGHT NOW

Attached: intelseethingpajeet.png (813x1402, 324K)

Also the latency is through the roof lul
Zen 2 confirmed a huge fucking flop

they already showed chiplets working in cinebench and gaming faggot

no they didn't. it blue screened then everyone gasped

it happened again?

Attached: ryzendemo.webm (864x622, 275K)

yup, lisa su hung her head in shame then committed sepaku

Were you just joking?

I took it seriously because crashing AMD demonstrations is pretty normal.

yea I was joking.

>shit that never happened

>oh shit an engineering sample of a brand new arch crashes
>an unfinished product is unfinished

Attached: poointhecpoo1.jpg (700x466, 37K)

Attached: toddlers.png (189x202, 42K)

And there were Zen1 samples running at 2.8 GHz, I still remember how Intlel retards kept claiming those were the final clocks.

Attached: 1518621284086.jpg (425x312, 70K)

Single-channel memory appears to be used. And slow as hell.

Attached: rams.png (390x105, 10K)

Just wait for 10nm

Attached: 1527629778452.jpg (679x758, 54K)

Friendly reminder that Intel will still be faster with 14nm.

Might as well have taken a shit in the DDR4 slot

You mean 12nm+++++++++++++++(((?)))

Whats this?

Attached: 1547157197363.jpg (1052x518, 32K)

jesus christ m8 stop falling for bait you fucking retard

>tripfag filtered
feels so fucking good

>Whats this?
anudda shoa

>they already shown chiplets working in cinebench and gaming faggot
true but only 1 chiplet with a shit card (V-2) so we don't know what the CPU is capable of with a 2080ti in crap titles like AC:Origin/Oddyssey.
i'm only interested in performance of 2 chiplets on the package.

>Also the latency is through the roof lul
what? where?

Attached: 1541892583390.jpg (818x693, 91K)

This is obviously fake. Trust me goys

Attached: 1521165988625.gif (333x358, 1.83M)

*guys

fuck off let him nibble on it

i pray that those latencies get better

9900k for comparison

So the single core of a 3.6ghz turbo zen2 is 116
The 9900k usually turbos to 4.7ghz and it has a single core of 146

Doing some quick math we get a zen2 IPC of ~31.03 and a 9900k IPC of ~32.19

What do you guys think, is roughly equal IPC to the 9900k good?

Attached: nJZcCUQ.png (829x155, 28K)

Based

good enough
their SMT is substantially better anyway

>What do you guys think, is roughly equal IPC to the 9900k good?
same IPC with MOAR CORES and way lower tdp? thats amazing actually. well the price need to be low as well so AMD can snatch market share on desktop too.

would be fucking glorious shows how useless both are

According to MSI's CEO Intel doesn't give two shits about the desktop DIY market anymore and has refocused their company over laptops, servers and GPUs.

Intel never gave two shits
it´s as if people have already forgotten broadwell

This is an old video from Zen not Zen 2. Newfags I swear.

We don't know the single score turbo of this CPU.
It only states average boost, same how it says 4.7 GHz avg. for 9900K when we know it boosts to 5 GHz single core.

This Zen could've boosted higher than 3.6, there's no real way to tell.

this is with single channel memory though, keep that in mind

is that info legit though? the speed is retarded

newfag

looks like a 2666mhz kit run in single channel. I have a kit like that in my build.

my bad

see How hard is that gimping performace?

It would be interesting if somebody ran a test with Zen+ and a similiar Ram.

depends on how much they changed architecture I guess, zen+ runs the IF at half the RAM speed.

AMD deliberately run the chip so that if found these scores don't really tell you much about the real performance. They know everything about their design so they can use this to calculate real world performance and where to place their SKU's in the market.

Someone gimped their 2700X for comparison: userbenchmark.com/UserRun/14085581

Explanation for us brainlets please

Bros .... we got too cocky ...... it´s not looking good .....

Attached: img.png (805x268, 43K)

the 2700X in the link is running at the same RAM configuration, 2666MHz single-channel, for perspective

>latency issues again
whew

god fucking dammit

Attached: ahhhhhhhhhhhhhhhhhhhhhhhhhhhhhh.png (396x408, 165K)

What's the difference in floating point and integer performance?
What gets me more frames in fortnite?

> Base clock 3.7 GHz, turbo 3.65 GHz (avg)

How is the turbo on average slower than the base clock?

Attached: question mark 5.png (620x640, 260K)

Because he downclocked turbo to compare results you silly

>Taking your 2700x scores we get this:

2700x @ 3.7ghz = 113 pts

2700x@ 3.6ghz = 109.9 pts

ES @ 3.6ghz = 130 pts

130/109.9 = 18.3 % IPC improvement. WHAT!?

That would mean the ES shown at CES was running at 3.9ghz all core. No way.

It very well could have been running at 4ghz~ all core turbo.
The Zen2 core has a massive FPU compared to Zen1, the front end got totally revised. AMD themselves quoted an IPC uplift of up to 29% in some mixed FP workloads at the Next Horizon event.

delid this kind sir

...

faggot

Zen1/Zen+ - 8 DP FLOPS/cycle
Zen2 - 16 DPFLOPS/cycle
mainstream meme Lake - 16 DP FLOPS/cycle
Xeon Gold meme Lake with AVX512 - 32 DP FLOPS/cycle

2D3212BGMCWH2_37/34_N
2=ES1
D=Desktop
3=3rd revision
2=2 chiplets
12=12c
BG=105W
M=AM4
C=12c
W=12x512kB L2 + 64MB L3
H2=Mantisse
37/34=3.4GHz base, 3.7GHz boost

At these clocks it seems to have 116 SC and 1741 MC points, while 9900K is 158 and 1670 on max scores.
Also, it's on a single channel 2400MHz 4 GB stick of RAM

Attached: z8ytrTdsq_F_gAgrIuy1NiqPegn5kWNlIjOEX5WenFU.png (1024x705, 150K)

>Xeon Gold meme Lake with AVX512 - 32 DP FLOPS/cycle
for about a second and setting your house on fire
AVX512 is a massive throttling meme

130/113 = 15%. They were saying something like 15% IPC.

The 3.6 Turbo in benchmark is probably 3.7. There were many reports of IPC being ~13% higher. So AMD may perform slightly better here in this benchmark.

>tfw fell for the phenom meme instead of buying 2600k
>ftw fell for the fx meme
>tfw fell for 'JustWait™ for Zen2 bro' meme

why are we always so retarded, ayymdrone bros? Why can't we just buy superior Intel products and enjoy life, like the based Intelorlds?

>92ns

Attached: 070.png (633x874, 57K)

stfu you don't know how disappointing this is to me. I'm sitting on a 2200g waiting for ryzen 2 and it turns out to be shit. fuck my life

gaymen eternally BTFO

Thats the same ballpark as Zen1.
An 1800X with manual OC and 3200mhz DIMMs with decent times will be mid 80s in DRAM latency.

The test system in question was using a single 2666mhz DIMM. With faster memory the DRAM latency will likely still decrease.

Attached: aida64 compar.jpg (1100x538, 104K)

t-they are j-just sandbagging it .... j-just wait .... intel w-will get btfo for sure t-this time ....

Attached: 1548041468164.jpg (653x726, 99K)

AMD's gambit is that slightly slower DRAM latencies will be more than compensated for by doubling per-core L3. 32 MiB for 8c can hold a lot of working sets pretty comfortably.

Going to system memory is always painfully slow... can I get numbers for contemporary competitors just for a sense of scale?

>still nothing official
you fuckin fagets falling for this

Bulldozer 2.0 waitfags btfo

A-at least we got more cores ... r-right bros?

>slightly slower
>more than double the latency

sometimes I legit think this board is actually filled with unironic Intel fanboys
especially when I see some youtubers get called AMD shills for whatever vague reason

Having cores more than 8 is anti-semitic

vs. Zen 1/+.

MONOPOLY GOOD, AMD BAD

Attached: 1547452383852.jpg (645x729, 51K)

Right here:

sir do the needful and buy ryzen cpu and radeon gpu

you want a braindead wojak reply because you're about to get one bitch lasagna

Will we ever get Acht Acht core cpus?

...

NOOOO DELET. TOO GOOD FOR TO BE TRUE. OY

JUST BUY IT OK?? YOU WANT INTEL TO HAVE 100% MARKET SHARE? IDIOT!

>intel aspies thinking 25ns will make any fucking difference in the real world
who cares about leaks at all, all you need to know is that the intel shills have been on full damage control for 2 months and will be for another 5. Why would they bother running 24/7 damage control on a cpu that is supposedly "unusable" because it has 25ns worse latency.

Attached: thiskillstheintelfag.png (252x171, 56K)

Friendship with intel ended

seems plausible all the money will be made in the server market in the future

This kills the Poozen

Attached: intelaviv.png (103x58, 2K)

>Intel 12++++++++ will beat this
g will be on suicide watch

$300 b-die memory will fix it

Attached: 1548011527456.png (802x799, 41K)

this kills intels latency

Attached: $.jpg (640x349, 63K)

>posting the reflection

Attached: 1502143993622.jpg (444x444, 81K)

Intel shills are very retarded, its easier to point the obvious to normal people out for them.