When will it come out? will it be good?

yet it runs games well

this guy gets it

Attached: 4L_nDbyhtCb.png (500x1250, 179K)

exactly what you would expect slightly better than a 3700x
clearly more CCX and even multiple dies doesn't do anything bad for performance
it's not like threadripper with weird NUMA issues

The game(s) just don't use more than 6 cores. No one said more dies would hurt performance, except maybe more load on the IMC.

brainlet post

dudes in the thread are talking about inter CCX issues when there obviously are none

>CPU with hugely more IPC loses
also decade-tier memory latency

Attached: 4L_ziAQjzaU.jpg (822x777, 207K)

if there were inter CCX or inter die issues then the 3900x would perform worse in games than the 3700x

It performs better due to less traffic per core on average.

or maybe it performs better because it clocks slightly higher and there's one game in the sample that can use the extra cores a little bit

Yet still loses to lower-IPC processors.

yes because it clocks lower and the specific IPC gains don't seem to apply much to games
not because of any inter CCX or inter die issues
this is the point I have been trying to clear up
the 3950x will be slightly better for gaming than the 3900x which is slightly better than the 3700x
there are no weird threadripper NUMA problems this time

>paying $750 for a cpu
Jesus christ you people are retarded shills.

>4.6 3900x
>4.9 9700k
The .3GHz difference should mean the 3900x would win. Lower latency wins, it turns out.

3900x does not run at 4.6ghz for longer than a few milliseconds the clock in a real game workload is more like 4.2 or 4.3

For the price of $340 you could buy a 9700k, B-die RAM, and have nearly the best CPU performance for low latency gaming.

this pic is showing me that the 3900X delivers a similar gaming experience as a 9900k.. for 30 dollars less...

plus 3900X crushes productivity charts

Attached: pricy.png (1548x732, 491K)

This, but a 9900k would be more future proof.

Attached: gh.jpg (1006x602, 210K)

obviously the 9700k is faster in games this is what the benchmarks show
I do not care about what the better deal for gaming hardware is I am clearing up stupid statements people like you make about technical details
the reason skylake runs well in video games is primarily because it clocks higher not any esoteric point about CCX or multi die

>Paying $150 for only 8 imaginary cores and 4mb L3 when you could've gotten 4 cores, 16 threads and 52mb gamecache

uhm....

Attached: i9k.jpg (1627x624, 248K)

gross....

What the fuck, you retarded desperate shill nigger, you can get the 9900K for 480 on Amazon and even less with deals on other sites. The 3900X absolutely costs more, and performs worse in almost every game.

FUCK WHY DOES EVERYTHING HAVE TO BE SO EXPENSIVE IN AUSTIN?!?!

wat

>similar performance
>looses to a 3 year old 5 core, 12 thread 8700k based of an arch that is now 5 years old, going on 6.
by your own logic you should buy a 8700k. you will never need the multi-tasking of the 3900x and by the time it will be useful, zen 5 and intel 14nm+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++^200th based off skylake will be out delivering the same performance but with a 6 core.

Good 9900K's don't exist since they're being siphoned off for the 9900ks sku.
9900kf is a better buy if you're OCing for this reason.

Buying a 3900x exclusively for gaming is obviously retarded. Might as well save the money and get a 3600 unless you're a professional gamer or have a 2080 Ti.

Attached: vs.jpg (1920x1080, 528K)

You can't go wrong with more cores and cache on a proven uarch so 9900k isn't pointless

Yeah of course, if you're actually a professional and need the productivity performance then a 3900X would be a good buy, but user specifically tried to make an argument around the supposedly similar gaming performance and "lower" price of the 3900X.

They're using ultra settings and probably testing mostly movie games, in real games the difference would be larger.

HT/SMT are fucking garbage for gaming.

I'm just saying for better smoothness, Intel would be the better choice due to lower latency and lower jitter due to no CCXs.

You can't bargain with shareholders bub.

idk much about computers, but the 3900X still seems to have a lot more value for your money.. but if gaming is your focus then I do recommend Intel

>t. not a shill

That's 90% of the home computer market because of gaming.

>1280x720

there is literally no evidence for this smoothness and latency and jitter thing
it makes no sense on a technical level
if you wanted to prove it it would be very easy just take a 3900x and run it in 2+2+2+2 mode and compare it to a 3700x
if the result is worse then CCX is a problem but it is clearly not

>HT/SMT are fucking garbage for gaming.

They said that back in the old 4C/8T days, New consoles are supposedly going to be neutered ryzen 8 core 16 threads. Hence why i said future proof.

>3900x and run it in 2+2+2+2 mode and compare it to a 3700x
There's test out there comparing 4+0 and 2+2 configurations of the 2700x.
2+2 performed better as each core has access to twice the L3 capacity.

The 3950x will have Zero stock and run 4.2ghz all core at most.
4.7ghz boost on one of the 16 cores for 20-30 seconds because the rest are defective chipLets that failed Epyic binning.

4.2ghz 16 cores is pretty sick it's going to bench very well against x299

My 2700 got .6-.7us interrupt to DPC latency on average whereas my new 9700K gets .3us on average. Games run infinitely smoother and the gap between average FPS and the lows is significantly smaller. Not to mention less input lag.

The day you get better performance with HT on is the day you need a new CPU.

doubt
even the 3900X barely reaches 4.1GHz all core

x299 already can't compete outside of avx512. Mesh cripples its ipc and the core's power consumption caps its peak clocks as well.

You have smoother gameplay because your CPU is faster. There is no technical detail the 2700x lags behind on other than FPS.

>what is memory latency and ccx latency

Attached: 1565903177324.png (1268x673, 207K)

What is with all the intel shilling here? Wait for the 10 to verify if lower clockspeed are design or bios problems as said by ayymd.

I thought we already dismantled your ignorant claims
the performance differences are entirely reflected in FPS
the FPS difference is almost entirely attributable to the clock speed difference
the 9700k is faster than any AMD CPU in games because it clocks higher than any AMD CPU
inter CCX and inter die issues do not exist outside of threadripper

I know, I just like posting charts.

>what are cpus in completely different price brackets

Disabling a CCX on zen+ drops the interrupt latency by roughly .2us, to around .5us from .7. I'm sure they improved the latency of zen 2 by a lot, but to claim there are no latency issues with zen is something a retarded fanboy would say.

if the interrupt latency affects performance it is reflected in FPS
latency exists especially memory latency but it is not the primary factor in lower zen performance
the simple fact that zen runs around 4.2ghz and skylake runs around 4.8ghz is the primary source of the performance difference
the idea that any kind of CCX or memory latency would cause more jitter or input latency has no technical backing at all
if you want more frames in your video games you should buy a 9700k because it is objectively faster single threaded than any zen CPU
claiming there is some difference other than FPS in input lag or jitter is very stupid

we just need 32gb of cache and do away with ram. : >)

I don't think there's excessive Intel shilling, people are just tired of the non-stop, unreasonable AMD shilling.
It's cool that AMD finally made some good CPUs again, but people act like there's zero reason to buy Intel anymore when that's obviously untrue.

When will you dumb niggers get past the "muh fps" meme? It's an irrelevant metric past a certain amount. The consistency in timing between the frames is what matters, and lower latency will result in better consistency. Even at 200 FPS capped, input response and smoothness would be better on Intel than AMD.

there is literally zero technical backing for the idea that any kind of latency would cause frame rate instability

I'd rather have more than less. Gives you that fuzzy feeling like my cpu isn't going to be dogshit in the next couple of years.

those 16 core thread rippers are slower than a 3900x or at best, the same in multi. you can go wrong with more cores as you will never get your value out of them unless you can use them which you, and 99% of everyone else on /g won't. and that cache isn't useful. that cache is there to make up for its latency and bandwidth bottleneck. without that much cache zen 2 would be much more worse off. it takes 4x-6x more cache to just match intel with its small 2mb of l3 cache per core.

look, if you're a retard, then by all means, buy something that's slower in 99% of the workloads you do for the same price as intel. you get your more cores that never will go fully utilized and be wasted but you'll be happy. since you're a retard. but if you care about your money, then buy something thats more practical. by something that has the right amount of cores + single threaded performance. and right now, 8 cores are the sweet spot. you either buy a 9900k for $500, or a 3700x for $320. or a 3600x if you're on a budget. but go ahead and call me an intel shill. you're a retard after all.

Attached: jeJNud9.jpg (608x369, 96K)

How retarded are you?
>higher latency
>average deltas between frames are less consistent

Bro you know the post you're replying to is talking about why the 9900k is good right? You look real retarded right now.

give me one slight indication of any technical reason why latency would affect frame time deltas
protip there is none you made it up

>I am incapable of abstract thinking and original thought
Back to youtube you go.

we are having a discussion and you have absolutely nothing to back up your claim
if you want to fall back on insults and rage posts then go ahead it will only make you look stupid

Attached: ram.jpg (1920x1080, 364K)

Why does it say lower is better on an fps chart lol

They admitted it was a fuck up on their behalf. People didn't have trouble letting them know.

Alright good, i thought i was being stupid and missing something there

You would need a pretty good reason to warrant 32 threads.

Watch it still lose to the 9900K in gaming.

Is anyone saying otherwise?

So, equal single threaded and Israeli rapingly good multi core?

Only after engineers figure out how to make a 1 terahertz processor : >)

They can just put the cache on a pci card and just plug and play. Boom 32gb L3

That's retardedly interesting.......

hmm I wonder if anyone ever had an idea like that

That's a great idea. I'd name something like that Random Access Memory

CPU needs assets to prepare a frame for the GPU, has a local LLC miss either from pajeet code data locality or simply because games are so big and complex nowadays, stalls for hundreds of cycles waiting for it. In the meantime the GPU finished rendering a frame and is now n-1 frames behind the CPU instead of n. Rinse and repeat and eventually the GPU catches up to the CPU and the whole graphics pipeline stalls

Think outside the box, Like apple.

>Latency matters
Show proof
>I have no proof, just "abstract thought" :^)

The state of mental gymnastics that intards have to make to justify their purchases. Last year you had some reviews to back up your stupid claims. But now, you literally have to make stuff up to shill your obsolete POS.

Run this benchmark on any ryzen system and look how high your latency is. Makes perfect sense.

Attached: memory.jpg (540x518, 97K)

You know memory latency is measured in nanoseconds and frames are measured in milliseconds right? It would take thousands of full on cache misses to affect frame time appreciably. You would need to find a reason for some frames to have thousands of times more cache misses than the next frames to justify the idea that memory latency affects frame time stability. You need about 30000 cache misses for a Skylake CPU to be 1ms faster than a Zen 2 CPU.

Okay, and what's the real world consequence of this? Again, with proof.

Bitchin' ass frame rates of coarse.

Attached: 1562636197585.png (1920x1080, 731K)

Attached: mp,550x550,gloss,ffffff,t.3u5.jpg (528x550, 45K)

Check out the percentage differences between 1% lows and average there. The 8700k and 2700x are slightly more stable than the others and the 7600k is somewhat less. The 3900x is slightly more stable than the 9900k. Kinda puts a dent in the whole latency jitter theory.

>3900x is on par with 2700x
he must've got something wrong
this is newer

Attached: WWZ.png (1373x1413, 49K)

I honestly wonder what the gaming performance will be like being based on a 4x4x4x4 setup compared to something like the 3700X/3800X with a single die. I own a 9900k myself which is great for gaming, but falls behind for stuff like rendering.