Another benchmark Ryzen 2700x(8 core) beats i7 7900x(10 core) on IPC and almost matches multithread

Many benchmarks seem to indicate that the the new Ryzen 2700x has higher IPC than many of Intels offerings.

This this benchmark mean Intel is Finished?

Attached: AMD.jpg (781x767, 151K)

Other urls found in this thread:

anandtech.com/show/12642/amd-ryzen-2nd-gen-details-4-skus-reviews-19th
valid.x86.fr/bench/1
twitter.com/NSFWRedditGif

NOOOOOOOOOOOOOO

Attached: 7c2.jpg (267x297, 15K)

>2700X
>3.7GHz/4.3GHz
>only 15.6 more in single core performance against an old gen at 3.3GHz/4.30

>>only 15.6 more in single core performance against an old gen at 3.3GHz/4.30
They turbo to the same clock retard

Wow, it's almost as if I already knew that when I put the turbo values in my post you replied.

Big if true. My 1700 @ 3.8 gets 430 ST and 4559 MT.

Glad to know you are acknowledging your stupidity

You guys do realize the 7900x has a 4.5ghz max 1/2 core turbo, right? Though Sky-x has pretty bad single thread performance for intel.

Attached: 7900xark.png (1059x939, 93K)

seems to be only 2.5% away from a i7-8700k at the same clock speed

Attached: 1514940385153.jpg (638x629, 44K)

Don't worry. CPU-Z will get another update to 'fix' this discrepancy.

To be fair, have you ever seen any other tests that showed similar performance for Ryzen? It was a particular piece of code in their benchmark that ryzen just happened to be extremely good at. But nothing actually uses. It was beyond outlier status.

Though there are some very specific benchmarks on sisoft that ryzen seems to be a fucking monster at.

>8c/16t Ryzen nearly matches a 10c/20t Core iNein

Good god, HyperThreading fucking sucks

I wonder how much X470 impacts performance over X370.

Also Ian from AT mentioned that the reason there's no 2800X yet is under embargo.

Got a link to that? I'm going to be so pissed if they drop a 2800x uber bin shortly after launch.

No, I saw it on twitter a few days ago and I don't have a phone to search for it now.

Was it actually from Ian Cutress on his twitter account? I'm all the way back to march and haven't seen anything about it. Or maybe he got told to delete it.

I think it was Ian, it was a response to someone, said something about "2800X's missing under embargo until the 19th" or similar.
If the embargo is so close it's not likely to be a TR SKU

NO NO NO

This was supposed to be our year

Attached: 1494356435910.png (633x900, 162K)

>Be good at something
>OH NO NO NO NO NO we can't be having any of that

2800X is probably awaiting Intels response so they can BTFO of them. They learned their lesson from the 8700K.

probably yeah

If they can pull another 200MHz single core and 200-300MHz all core they'll be very close to the 8 core CFL, probably even better in some cases with tuned memory.

IT WAS OUR TIME


OUR TIME

Attached: 1504014579635.jpg (1053x671, 103K)

>what is price
>what is tdp

Waiting for it works fine for AMD since when it launches it'll be at its most optimal with fully featured BIOS and microcode improvements and the most memory compatibility.
Secondly they're gonna need to build a decent stock from good dies, that will take a while.

Lastly I think it won't actually cost more than $360, but it won't come with a cooler.

I have my doubts if even a 2800X will ever exists.

>
>>what is price
>>what is tdp

These are already public information:
Launch price and TDP

anandtech.com/show/12642/amd-ryzen-2nd-gen-details-4-skus-reviews-19th

>This $1000 CPU beats out this other $1000 CPU by 1%!
Literally who gives a shit?

>>This $1000 CPU
Ryzen 2700x will be about $ 330

Attached: 1500010756103.png (1228x1502, 944K)

>2800X exist
Why would it? The difference between the 1700X and 1800X were almost nothing. Just make the best and call it 2700X. Done deal.

SMT doesn't suck. It is very good at workloads and code that is hilariously parallel.

The issue that SMT suffers if there's resource contention (Intel never address this until meshed-topology in Skylake-X)

DELID DIS
DELID DIS
DELID DIS
DELID DIS
DELID DIS

>mine monero on an AMD processor
>using all threads improves hash/s
>mine monero on an Intel processor
>performance actually goes down unless I use the number of physical cores
Wew.

For reference, 8700k at 4.8GHz only scores under 4.4k multithreaded.

CPU-Z is going to have to update to gimp Ryzen harder again.

If this bench is true than there is only a 2,5% difference with a stock 8700k (508 score) on single core

see link for reference
valid.x86.fr/bench/1

Intel should be worried

Why should they be worried? They still got the 640x480 benchmarks which rely completely on memory latency, they'll win there!
Just get a few reviewers to find a board with shit default Ryzen timings and you're good to go.

It looks like the IPC increase is anywhere from 2-9% depending on application. That puts it VERY close to Coffeelake IPC, given that Zen1 was above Haswell and Kabylake and Coffeelake IPC is only like 10% about Haswell themselves.

It seems incredibly likely that Zen2 (not Zen+) is going to be significantly higher IPC than anything Intel has to offer. And likely clock close to 5ghz. And have higher perf/watt.

Had a small convo with a guy about Intel's next uarches, apparently everything after Icelake is a IPC increasing tock, besides Tigerlake, so Ice > Meteor > Adler.

Competition will be fierce, next 4 years will be amazing.

Icelake included as well, it's a tock.

I said HyperThreading, not SMT. HT is Intel's implementation of SMT and it sucks compared to Ryzen's.

Heres the I7 4790k at 4.2 ghz and a 1700x at whatever the fuck its boost is, it really does this time see that intel is actually even more fucked this time around than when ryzen first launched, Single core is similar to kaby lake and coffee lake but at a lower clockspeed

Attached: cpuz_2018-04-15_20-57-15.png (403x402, 16K)

Don't worry pal.
Zen2 is good as Vega just wait was. It totally lived up the hype

It's not our fault you only have the most cheap poolaris model and can't have glorious 8700k oc + 180ti sli

Haswell devils lake i7: 415
Ryzen 1: 370
Ryzen 2: 495
Skylake X: 480

INTEL BTFO

Attached: 1495358050273.png (1070x601, 495K)

Reminder that Ryzen has some AI branch prediction tuner that helps in benchmarks.

So does Intel (and they've done that even longer). Your point?

Shh into the ovens you go.

Because it's cheating!

t. Intlel shill

2700x official specs are for xfr2 4.35ghz
If there's any merit to that rumor then a 2800x would be a very nice surprise. Pushing it to the mythical 4.5ghz?

Attached: 13247509457687.png (1437x874, 369K)

guys seriously stop with the retarded hat wojak spamming. you don't know how it makes us feel

Wait a second, how is this 8700k working on a Z170???? HOW IN THE FUCK!!!!

Attached: 1522796911002.png (750x731, 531K)

I FUCKING HATE AMD!!

Attached: 1502831978496.jpg (700x565, 73K)

You need an unofficial bios

Thats cpu z not cinebench favelamonkey

Attached: 1502055755975.png (672x794, 343K)

>implying anyone but a diehard intel fanboy is going to go through any of this bullshit
Keep fucking dreaming

Attached: Delusional.jpg (938x120, 14K)

>2700X
>$1000?
state of shills, jessus christ..

He must've read that MT score and thought we were talking threadripper. That's the absolute state of corelets, they simply can't believe this. It's a sad state of denial.

Suicide watch

Attached: 1501991613334.png (882x758, 241K)

SHUT UP

Attached: 1520867957888.jpg (679x758, 52K)

JUST!

Attached: bk.jpg (921x865, 354K)

Holy shit, I was expecting just a simple die shrink, not such gains.
Imagine Zen2 on 7nm.

So zen2 potentially will be 4.8ghz stock ?

What the fuck this cant be, every year is now amd year

Attached: 1497689967244.jpg (638x599, 123K)

MAN I HAD IT WITH THIS FUCKING JEWS!!!!
I AM SELLING MY SHIT i7 6700k!!!!!
FUCK IT!!!!

is zen still ass on linux?
i just wanna build a comfy server

Attached: 1518073993045.png (432x616, 215K)

That board is probably the best for water cooling 8700Ks even better than most Z370s

No. Just make sure you have an up to date kernel and BIOS.

Still? Agesa updates came out like, 2 months after zen launch.

No, their current 14+++++++nm process lets them jack their clocks high enough to surpass the 2700X and a even higher clocked 2800X
Even without the higher clocks Intel still beats AMD at extremely niche and really extremely latency sensitive workloads, like HFT jew nigger shit
Intel won't be finished on every front until 7nm, unless they get their 10nm process up to speed or some other miracle happens

That's what happens when everything has been optimized heavily for the stagnant Intel arch's since Sandy Bridge, this optimization is quite significant when you compare it against a completely brand new CPU with zero optimization

HT doesn't really suck, it's just that Zen is able to execute more instructions at once than Intel
Since it's really hard to extract so much parallelism from most programs, running two at once really helps to make use of Zen's execution units

Great thread as always guys

Attached: ult.jpg (480x360, 11K)

You don't execute 2 instructions at same time it's impossible with ht or smt. You just have a pipeline with different locks waiting resources our running on CPU.

>Much gains
It's basically mediocre gains. Anything's for such shitty frequency is a gain

> 8 core matches 10 core in multithread

CAN SOME ONE EXPLAIN TO ME WHAT THE FUCK IS GOING ON HERE ?

lower clocks and not running the mesh at 3200 or higher

Processors have been superscalar for a long time now. I think even the pentium1 was superscalar. Multiple integer execution units can execute multiple instructions at the same time if the processor knows there will be no side effects from doing so and that the results do not rely on each other.

This is what Ryzen 1 should have been. But I'm still going to wait until 7nm Zen, only then Intel will be truly fucked

>Intel will be truly fucked
Doubt it.

this is sorcery, does Lisa Su acquired some alien tech?

Jim Keller is clearly not from this world

10nm will never happen. They'll just rebadge last year's CPUs with a tweaked 14+++++++++++++++++++++++++++++++++++++++nm, 10GHz clocks and a 300W TDP

Attached: rgfsdv.jpg (208x249, 6K)

Attached: 1522893678356.jpg (452x523, 22K)

>absurd overclock
>not showing cpu-z's benchmark

Attached: 8d6.jpg (645x729, 41K)

Attached: h_01933237-512x318.jpg (512x318, 52K)

Something something, Zen actually can run more (specific I think?) instructions per clock cycle, per core(don't quote me on this) but is bottlenecked with one thread on one core, but not with two threads leading to the almost linear scaling of Zens multi threading. Really I don't remember exactly, there's some good technical stuff on Anandtech about it from a user called "The Stilt". Posting from phone so fuck you find it yourself it's a interesting read if you like that stuff.

>"turbo"/"boost"
I really don't like how both AMD and Intel manage to fool tech press and others into showing "turbo" speeds. It's pointless and useless. My CPU is either close to idle or all cores are working at or near 100% load. "turbo" and "boost" (whatever they call it) only matters in synthetic benchmarks. Just tell me the IPC or each thread when threads are maxed out.

Now I might upgrade from my 4790k

Attached: 273.png (476x401, 165K)

Normies usually don't actually use their computers much, they either only do short spikes of load while loading facebook and scrolling through their feed or medium loads with a few spikes while gaming

6 million jews had the best time of their lives. Join them, user!

Attached: 1513339985381.png (460x454, 183K)

Attached: f05.png (693x460, 165K)

Thing about that is, for the most part code that's friendly to Intel is friendly to ryzen. But old software using old compilers uses unoptimal code paths treating it as older AMD CPUs. That's where in titles such as Rise of the Tomb raider when ryzen seen a significant performance uplift, this was mostly a simple code path update and a few tweaks. Some of those tweaks also helps Intel cpus as well.

You can look at Prime95 as a perfect example of this. Anything before 29.1 couldn't get ryzen beyond 80% load due to using unoptimzed code since the software had no idea what it was looking at, or what to do with it.

7900x as a 4ghz all core turbo, and can boost 2 cores up to 4.5ghz. Even if that 2700x is overclocked to 4.3ghz all cores, there's not a big enough clock speed difference to compensate for having 25% less cores, and single core is still at a disadvantage. It's mostly down to mesh and the 7900x's pitiful cache.

>shitposting amdrones still rampant
why can't they just wait for actual benchmarks that release in 3 days? what's the point in shitting up the board? if ryzen is so good then you should just wait for benchmarks.

>inb4 actual benchmarks are even better

INTLELFAGS ON SUICIDE WATCH