E3 slide leak

This implies that they have a 10nm product that's not DOA. Ice Lake-U is a paper launch.

Intel will become 10nm superpower in 2030, you'll see

BASED AMD INNOVATING AND LEADING IN THE SEMI-CONDUCTOR SPACE.

all those cores and nothing outside or rendering and VMs to take advantage.

AMD is really e-peening here. hell not even the 8700k is hampered by anything in gaming.

game devs need to get onboard or intel and amd need to cut the "more coars" shit and focus on ultra fast quads cores.

> @
You need to go back.

Imagine 4k video encoding while streaming
Imagine the workloads

>I don't understand physics.

Inturds really are dumb.

This tbqh. The last paradigm defining thing to come out was speedmonster quad core 6700k.

the 8700k is already choking on plenty of stuff compared to a 9700k

Physics don't matter

Attached: 1539296460877.jpg (474x415, 37K)

Just like how Intel destroyed their name with the Pentium 4 by trying to go beyond 4GHz?

You're the ones who don't realize that Intel is only throwing 5GHz because they're limited.

The same fucking logic worked with the Intel bruteforcing Pentium 4's with MOAR GHz when they were faced with Athlon 64 X2's? Right Inturds?

>still can't reach 5Ghz
WHY BROS, IT WAS OUR TURN.

>not 5Ghz
guess adored was wrong again.

how is a 6700k in any way impressive it's barely faster than a 4790k

>doesn't know that AMD reached 5GHz first.
>Muh Nigguhertz

AHAHAHAHAH HISTORY REPEATS ITSELF

Doesn't need 5Ghz. It already curb stomps the 9900k at 4.7.

Abd look at the TDP. 105W for 16 cores.

5GHz superpower by 2038

amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189

amazon.ca/Best-Sellers-Electronics-CPU-Processors/zgbs/electronics/1233052011

amazon.co.uk/Best-Sellers-Computers-Accessories-Processors/zgbs/computers/430515031

amazon.com.au/gp/bestsellers/computers/4913338051

amazon.de/gp/bestsellers/computers/430177031

Attached: 1543585518490.jpg (960x720, 116K)

>trusting AMD's TDP
kek, remember 9590?

why stop 4.7 ghz? why not 5,6 or 10?

The sky's the limit when you're photoshopping

>16 cores
>gaming cpu

lmao

kek, remember 9900k?

>nigger calls kettle black.

>companies never market to retard gaymurs with more money than sense

Yup. And the 9590 not over clocked took the rated 125w. When you over locked it to 5+Ghz it took 250w.

Your point?

40 hardware security vulnerabilities has been delivered to your Intel CPU. Thanks for defending Intel online

/Intel Damage Control Team

>Imagine being a technology grognard
256k is all you will ever need!

>tfw you realize that games being developed for the PS5 and Xbox "Two" are designed to maximize 16 threads.

Inturds got fucking worked. Imagine buying 8/8 CPU's and realizing that new games ported from the PS5 and new Xbox are optimized to make use of 16 threads

QUICK BROS OVERCLOCK 9900K TO 6GHZ AHAHAHAHAAHAAHAHAHAAH

Attached: 1558853140201.png (800x618, 937K)

meh, I'll believe it when I see it
not sure if games will keep scaling very well beyond what's being done already

Imagine thinking they could brute force a super clocked 8/8 to equal a power sipping 8/16.

you realize that API's are moving towards parallel computing and that both Vulkan and DX12 are doing optimizations to improve multi-core performance.

What you need 16 cores for. I’d rather prefer 8 cores + navi apu.

>3.5 GHz base
>4.7 GHz boost
How

buy a PS5

Nice
No way my shit x470 k4 will run it so I'll just stay with my 2700x until 5ghz+ is possible

yes, but they also relieve overhead, and amdahl's law exists

If it is videocardz then it is 100% true, literally only site I trust with leaks in tech industry.

9590 was not rated at 125w - it was rated for 220w which is roughly what it pulls under stock conditions. This massive leap in TDP vs the 8xxx chips is why only a handful of motherboards supported them (mostly asus) precisely because it wasn't a bullshit number.

Can't wait for 240hz ips to come out and jizz all over it hnng get one of these fuckers at 5ghz+ oc

699$, calling it now. AMD will take a premius but not too much.

Attached: 1547697586525.jpg (218x250, 7K)

Low boost duration is the simplest answer. Intel does it that way as once you start loading all cores you can't maintain TDP at such high clocks for very long. Its why AVX fucking annihilates Intel TDP numbers and chips will throttle below base clocks to stop them from melting as it is so stressful.

kek. you niggers won't buy an AMD cpu running at 1Ghz while killing the 9900kelvin at 5.2Ghz in gayman benchmarks

>Frequency doesn't matter!11

Good thing about gaming fx boards is that they had great caps. Mine still run fine since 2013 on the original caps.

It really doesn't because if it did. Intel would've not switched it up and just tried to reach 10GHz and burned their fabs trying to do it against the Athlon 64.

Also, Imagine trying to justify Ice Lake's clocks vs your current lineup next year.

I say 999. Threadripper was a 1200 dollar chip.

MOAR MEGAHURTZZZZZZZZZZZZZZ

Attached: 3452.png (450x650, 75K)

>one intel core at 5GHz is worth 2 amd cores at 4.7GHz
* in amount of power draw (higer is better)
ayy lmao

what a beast

Bruh that 72MB of cache and 4.7Ghz single core
>but I bet it all-core turbos to 4.3Ghz with PBO

so this basicly confirms that 5.0 ghz is pretty much feasable

likely 749$, half the cost of intels 16-core

No.

4.6 All-core. AMD just needed to reach a single-core parity with Intel. Better Multithreading would do the rest.

@71343051
no (you) for you!

Attached: 1384916822674.jpg (640x470, 58K)

The 3.2C could be overclocked to over 4GHz.

HOLY SHIT ITS OVER FOR INTEL HAHAHAHAHAH

Attached: 1533848965742.jpg (1024x768, 333K)

Attached: It'sOver.png (1296x900, 371K)

yes even fucking 2xxx ryzen on LPP had a 300-350mhz headroom
4.7ghz

Amdahl's law is complete bullshit for the specific realm of graphics pipelines, game engines, etc. There are certainly algorithms which must be run in a serial fashion, but a vast majority of the slow parts can be trivially parallelized if the developers know what they are doing. Ray tracing is a great example of something that scales linearly basically out to infinity.

oh no, you sound like that retard youtuber moore's law is dead

find me a 4.6ghz 2700x lmao

Attached: 1457294956960.jpg (801x1500, 177K)

>there's still a market for threadripper

Attached: 1556325887287.jpg (586x580, 346K)

>16 cores 32 threads
>Gaming
Wut???

>THREADRIPPER 3 48C? NONONOONO

and you sound like a retarded NEET who failed out of college after a sad attempt at an engineering degree.
>muh academia theory concepts applied universally without any regard for context

Future proofing for multi platform ports from the PS5 and the new Xbox.

Because if you bought kaby lake 4 core shit in 2017 you got fucked hard.

>focus on ultra fast quads cores

Attached: 1522192439102.png (1300x2000, 307K)

kaby lake is great - EOL in under a year and no upgrade path.

the context was specifically related to workloads that aren't particularly parallelizable
add more "cores" to a gpu and performance scales almost linearly, yes, but you're just being a gigantic retard in not realizing what dx12 and vulkan do with more threads

AMD is already implementing infinity fabric on Apple GPUs

~$660 would be same price/core as 3700x/3900x so something between 649 and 699 seems kikely

based

Attached: Lisa Su.jpg (757x627, 87K)

*likely

700+ is more likely

Attached: Muh 5ghz.png (1920x1080, 666K)

They've already implemented it on Vega.

single core boost probably

>3.9 beats 4.8 ghz
based

16 core 4.7ghz 105w tdp
This is too good to be true. Convince me I'm wrong Jow Forums, please.

Attached: 1508688011779.gif (342x284, 1.05M)

That's an interconnect between the GPUs not on them. It's on the card, like crossfire but for HPC GPGPU

We don't know the price, probably around 700 burgers

>”making use of 16 threads” means the same as actually gaining benefit from
16 cores

kek

>irrelevant game
>irrelevant CPUs
>800x600
you're not even trying

It's much lower all core boost speeds. Until you over clock it, then you're pushing 200w through 16 cores at 4.7Ghz all core.

more PCIe lanes and memory channels are still a big benefit for HEDT workloads