WINNING

>Zen 2 needs 4000mhz+ ram to preform well.
zen 2 is not even out yet, but you already know what ram it needs
whoa... insane insider info...

>he's still spamming it in every thread despite being btfo repeatedly

People who cared about single core performance were buying 9700k for gaming use cases which could outperform 9900k. This puts 9900k in an awkward position where it's overpriced and underperforming for both gaming and productivity users. If you a half and half balance of both then Ryzen is still ahead. If you multitasked while gaming guess what Ryzen is ahead again...

The real story is that the 9700k sku looks comfy for gaming. Their lower end lines still offer high speeds for light usage although they choke harder under heavy multitasking. Likely that 10c10t Comet lake CPU will become the new gaming kang.

9900K is truly BTFO for all time and that hotter than the sun part has no reason to exist getting BTFO by the 3700x which comes with a great stock cooler from a value perspective and the 3900x from a performance perspective.

Intel remains competitive in gaming (like they say) but what's shocking is how the professional market they had on lockdown really is better off not buying Intel unless they use something like AVX-512 or Quick Sync. Intel also really isn't looking so hot in the datacenter either.

I'm a bit disappointed that 32gb dimms are only available in ddr4-2666 cas19. So much that I'm starting out with 32gb and building up from there.

Bruh, 3xxx works with my B350 board

Lmao fucking what?

delid this
reported for antisemitic content

Zen 2 needs 128gigs of cl12 5100mhz+ ram to get anywhere close to Intels bottom barrel chips from last year in single threading
its over AMDrones

Attached: 1528705187294.png (900x980, 104K)

>Trusting something unfair and AMD biased as geekbench

>source: my ass

Multi-threading is the meme they tried to sell their Bulldozer garbage a decade ago. It has no or very fringe real-world application.

AMDtards never learn.

it's been over two years now and retards still think ryzen is the same as fx
pathetic

i'm fine with industrial chillers my little niggy. you may be too casual for that sort of thing and that's OKAY. not everyone is going to be bleeding edge.

>Multi-Threading has no or very fringe real-world application.
/3/ would like to have a word with you retard

Modeling software is inherently single threaded apart from a single use-case of rendering.

t. Blender user who had fallen for the AMD meme and is regretting it

it's sad but a 9700k 150 quid board and 80 quid ram ram will do a better job in gayms and still be cheaper than a ryzen x570 platform. amd bought into their own hype and will have to lower prices sooner rather than later, which will damage their brand but be funny af tbqh. anyone who doesn't give a shit about gamershit is better off buying a cheaper zen or zen+ cpu and a cheap x370 or whatever the poorfag platform is (b something or other).

OH NO NO NO NO NO

Attached: 1561232529597.jpg (1012x460, 63K)

What happened to the 15% ipc gain? Intel is still dabbing on them in the most important department: gaming

I might get a 9700k after the price cut
But let's see more benchmarks first

Bleeding edge of hardware exploits that is.

>3800x-benchmarks-leaked-crushes-intels-i9-9900k-in-multi-threaded-performance/
NOOOOOOOOOOOO

Attached: 1558950008166.jpg (1365x1280, 810K)

>Intel
>most important is gaming
That is the way it ended up because the enterprise demand is in heavy parallization.
Amazing how AMD is the thinking man’s choice these days.

*gets btfo by Xeon*

>xeon gets btfo by bulldozer because of security patches
kek

Telling their fans that the 3600 is 1% faster than the 9700k and in reality even the 3800 is by far slower, what a scam

Attached: image.jpg (1536x2048, 3.19M)

>inventing farcical scenarios in your head to defend random corporations

shit ram was used. wait for the actual reviews

>JUST wait for them to use 4666mhz ram
>intel btfo!

Oh no no no

Xeon is just a slow version of the flagship prosumer line.
So you pay more for a chip that performs less.
Simply just so you can get access to features Intel is keeping out of the i-series.
It is like they are destructively fragmenting their own costumer base for very shortsighted gains.

I think AMD said the sweetspot was 3733.

>reality doesn't matter

Now let's see that with mitigations applied.

>le ten second clock boost for my single core benchmark

>What happened to the 15% ipc gain?
intel boosts to 5 nigga hertz. clock for clock Ryzen should be faster.

How can Intel get crushed so hard even by Bulldozer jesus christ

>4666MHz
>does not know how a graph works
>does not realize that anything past 3733MHz will have worse latency.
clearly you don't even try to understand how to refute things.

I winder what this mean

Attached: 76yp2srrjb531.png (1121x844, 715K)

it means that you're an idiot who just sets maximum on everything, and then complains why it's not finely tuned. You're like the idiot who presses the Turbo button on a 486 computer.

ai.google/research/pubs/pub36448.pdf

>boasts about getting a 9900K to 5 GHz
>doesn't understand how to tune a processor.
>keeps sperging about how the an AM4 bios can go up to 4666MHz memory speed. but does not understand that sweet spot is at 3733Mhz.
You're the idiot who would probably overbolt a 9900K, thinking it's better with more voltage.

>2010
Maybe a decade changes some things.
It would be impressive if it didn’t.

You guys.. the pic is in response to the intlels who doesnt know about dividers and think half speed IF Is true if ram clock is high.

The intcels dont know about dividers.

feel free to point out what changed, an appeal to novelty isn't very convincing by itself

> I think AMD said the sweetspot was 3733.

Gains after 3200 MHz are miniscule, though. Thus why the Intel shill "you need expensive RAM" is horseshit.

Do we have any estimates from earlier Zen+ chips how high the latency difference is between 2133 MHZ and 3200 MHz?

Attached: Ryzen 3000 RAM latency.jpg (1810x1018, 651K)

The reason he may be referring to the 5000MHz RAM speed is probably because of the World record that was achieved with a Gigabyte AM4 board. then the UEFI option to run your RAM at 4000+MHz

But why? A cheap Hynix kit that does 3200+ and a 60 eur mobo will do just as fine, still be a lot cheaper with a Zen 2 chip than a 9700 or 9900.

Attached: 0.png (500x500, 184K)

3200C16 and 3600C18 are reasonably cheap. C14 and C16 respectively are very much not so.

I never understood this latency jewish meme, 2950X should have way higher latency and it's pretty much on par with 2700x in gaming

youtube.com/watch?v=3nIXahVEejM

Attached: 1541181531230.jpg (1280x361, 228K)

Latency impacts how long it takes the dimm to start shitting out data when requested. If you do huge numbers of very small reads this will eviscerate your performance. On the other hand if you're mostly doing big reads it's virtually inconsequential next to your raw throughput.

I'm impressed with Ryzen but it's a damn shame the single core performance is still lagging behind Intel. For that reason alone I can't bring myself to go back to AMD.

>running single-channel DDR4 rated at 2133MHz

Really, idiot?

Couldn't find 2133 MHz benchmark, but rough estimates would put it at 87+ns minimum, based on MHz scaling and diminishing returns.

The Ryzen 2000 has +12.5% higher latency on 2666 MHz compared to 3200 MHZ. Meanwhile, Ryzen 3000 has a ~ +20% higher latency on the same Ram frequency.

With a bit of napkin math, a 2133 MHz RAM on Ryzen 3000 would probably have somewhere between 29-38% higher latency compared to 3200 MHz, maybe even higher, but there some unknown variables.

Attached: Ryzen 2000 RAM latency.png (692x626, 19K)

what if memory speed doesn't matter that much with poozen mk3 and that's nearly the best one can get?

Who are you quoting?

even based on AMD's slides this can't be true, going to 3200mhz from 2667 already drops latency by 14ns, 2133mhz RAM might have latency in the 90s or even higher

I'm not talking about latency hypotheticals but resultant performance

what reason would there be for performance not to scale with memory latency? even intel's processors single threaded performance scales with higher memory speed/lower latency, albeit not nearly to the degree ryzen does since latency is lower all around on intel's platforms

next gen games will use at least 8 cores so if you buy a shitty intel thinking youre going to gain a few fps from single core performance youre in for a rude awakening in a few months

>albeit not nearly to the degree ryzen does
exactly, and amd has made improvements all around the stack so similar scaling could be possible

>Blender

Stop thinking every modeling software is the same as that heap of dogshit, freetard.

that response back makes no sense. how can you honestly say with a straight face you don't see a connection? you literately spent 300 million to hire new people then immediately fire 12,000 people. does he actually think intel just pulled 300 million out of its ass for new highers?

>8c16 vs 8c16t
>inferior single core performance but superior multithread
what the fuck? this is more proof tthat geekbench is flawed as fuck

possibly, i could see zen 2 not scaling to the degree zen 1/+ does but again, based on AMD's slides im not sure that's the case. look at this user's math here , it confirms what i was thinking earlier with 2133mhz RAM having 90ns or higher latency, and based on that graph and his own math compared to AMD's slides Zen 2 has slightly higher memory latency even at the same speeds. maybe that higher latency is negated by the infinity fabric changes but i'm not convinced, intel doesn't scale with memory speed/latency not because latency doesn't decrease by similar amounts at each step but because latency is already so low that those reductions are intangible

hol'up.
>the restructuring is being done to better align (and to align quickly) our company's structure to the company's strategy
>hire 300 million worth of gays, trannies, wahman, black people, and more pajeets
>fire 12,000 cis white men and cis asian men
>then claim the "restructuring" is to better align intel with their new "strategy"
>new "strategy" of
then says he doesn't see a connection? hiring 300 million worth of gays, trannies, wahman, black people, and more pajeets

Attached: 1559590090666.jpg (444x236, 18K)

im not sure they even used that money to hire them, all i know is that they straight up gave 300 million to political organizations like feminist frequency which is worse.

Actually AMD is winning in single core too. That's what makes this Zen 2 a Holocaust.

They used 3733mhz RAM in the on stage demos fag

No it's proof that Intel's SMT implementation is flawed as fuck

I assume it could be due to very aggressive single-core boost from Intel CPUs and due to the fact that HT is worse than AMD's SMT on Ryzen, but yes geekbench is hot garbage anyway.

AMD has more execution ports per core than Intel and a good SMT implementation to fill them. Geekbench multicore sprawns on all logical cores, so it makes perfect sense.

The power of infinity fabric.

Hmm

Attached: a03fc4cfecb30b89e1e864b8e1e3139e4da85961f250904a02160a86516dd304.png (856x936, 58K)

Are the official reviews coming with the 7/7 release?

the funny thing is that they're running it on Single Channel, 2133MHz, and the Inturds lapped it up like idiots, not realizing that the performance gets better with dual channel and tighter timings.

Yes, unless AMD went full retard and didn't send out review samples ahead of time. I sincerely doubt they would botch such an important launch though

>it confirms what i was thinking earlier with 2133mhz RAM having 90ns or higher latency, and based on that graph and his own math compared to AMD's slides Zen 2 has slightly higher memory latency even at the same speeds.

We lack data points, but at linear scaling, it would put Ryzen 3000 at ~112ns minimum latency (or 120+ns if the scaling is indeed magnified on Ryzen 3000) with shitty 2133 MHz RAM. Not sure how that compares to real-life scenarios, though.

If my estimates are right, then the 3800X should definitely be able to trade blows with the 9900K at the very least once non-shit RAM is used, and everything else is just shitposting and shilling. That's my takeaway from those "leaks".

Attached: Ryzen 3000 Benchmark Computex.jpg (2053x1025, 137K)

Nah, the reason the Intel one is worse is because hyperthreading is turned off.

Bigger L3. If the threads don't share much common data there's less time spent pulling shit from RAM.

where is that boost speed user?
its literally just below base clock

Thanks, I wondering if (hoping) we might get early reviews.

>4000mhz
amd's own charts say that you get the best results on 3733, also you still get very good speeds on 3000mhz ram wich is cheap as chips rn (73$ us for 16gb 3000mhz or 60€)

>Zen 2 needs 128gigs of cl12 5100mhz+ ram
meanwhile intel does not even support those amount or speeds

Naw

he does imply they did hire though with:
>spending to increase the diversity of intel's employee base is being done in order to ensure that we have the highest performing teams possible - only by bring in wide range of experiences and ideas can teams meet their highest potential
which is pretty retarded thinking. just because you're a wahman doesn't mean you bring anything unique to the table just for the sole fact you have a vagina. you can have two people with dicks but be the polar opposite of each other. same goes with vaginas. so it doesn't matter if its a nigger, chink, spic, whitey, wahman, man, mentally ill tranny, fagget, extra. now how much of that 300 million went to "donations" to virtual signaling groups and new hires based on sex and race is unknown.

If I remember correctly intel has ~42ns of latency at 3200mhz, it really make me wonder what speeds AMD was using those slides like the one you posted. if they're only using say 3200mhz ram then they really are sandbagging the results, but if they're using 3733mhz ram then the results can't get any better than what they showed us, which is them trading blows with intel like you said.

at the very least what we can take away from a combination of the leaks and AMD's slides is that Zen 2 does indeed have higher IPC than coffee lake, i really hope overclocking is good and we're not just limited to the PBO boost as was the case with Zen 1/+. a 5ghz 3800x would beat intel by a good margin if they're already trading blows, i hope overclocking in combination with the 1903 scheduler changes and everything else really bring out some extra performance over what AMD showed us

It would be nice if someone leaked some here. Can't get fucked by an NDA on an anonymous meme forum.

Wrong, Xeons are the better quality silicon. They are binned for energy efficiency and stability not "MUH MEGAHURTZ".

They are use in real-world workloads not shitty gamying a.k.a it makes you and your business $$$$. They are priced accordingly to this.

>whypepo good

Attached: typicalamerifat.jpg (360x480, 24K)

>They are binned for energy efficiency
lel

>SJWtards first false flag
pathetic