THE CITY STATE OF INTEL

>>61351781
Isn't this the exact same shit they were ragging on AMD for having to do to get optimal memory latency by staying on a single CCX?

>>61351781
So... Glued-together L1 cache?

>>61351800
Bingo

>>61351805
So their new mesh of bingbuses has some of the drawbacks of MCM but not really any of the benefits .. nice job, Intel.

>>61351781
OHOHOHOHOOHOHOHOHOHOO

File: Screenshot from 13-07-2017 12:45:38.png (761KB, 1493x838px) Image search: [Google]

761KB, 1493x838px

>>61351781
I got a feeling this company has a tail and head that don't listen to eachother.

NUMA NUMA YEH

http://www.realworldtech.com/forum/?threadid=169894&curpostid=169904

> It simply looks like EPYC is the better choice for lots of people. Better price, better performance, and a company that doesn't play stupid market fragmentation games.

t. Linus Torvalds

>>61351781
It's not strange, Intel recommended 2 NUMA node clusters for Broadwell-E due to the dual ring busses.

>>61351781
How does 4 numa nodes fit into a 6 memory channel arch? Feels like Intel didn't have the die space or yields for 8 channels

>>61351898
What's old is new again. https://www.youtube.com/watch?v=OE8WzYNRPNU

>>61351934
It's the hypocrisy.

>>61351781
WHY DOES IT NEED 4 NUMA NODES PER ::::CPU:::: IF IT'S A SINGLE DIE WITH A MESH INTERCONNECT

That shouldn't be happening unless the Mesh itself is partitioned

>>61351781
> Update 1: On speaking with Diane Bryant, the 'data center gets new nodes first' is going to be achieved by using multiple small dies on a single package. But rather than use a multi-chip package as in previous multi-core products, Intel will be using EMIB as demonstrated at ISSCC: an MCP/2.5D interposer-like design with an Embedded Multi-Die Interconnect Bridge (EMIB).


AHAHAHA
A
HAHAHAH
AHAHAHA
HA
AHA

Can anyone explain this to the layman?

what the fuck are you guys even talking about? how do i learn about this wizardry?

>>61351862
This is what happens when marketings does things without asking engineering.

I bet the engineering section is left flabbergasted at these idiots, when Intel does the fucking same and will just continue going more modular in the future

File: 55-1080.2427321958.png (88KB, 1500x844px) Image search: [Google]

88KB, 1500x844px

>>61352063
They got NUMA'd.

>>61352077
Is this going to be any more or less effective than MCM?

>>61352115
It's about as good as using silicon interposer.
Maybe a little bit worse.

>>61352115
We'll see in a few years.

File: a7f8a31517ad8fc601409be1ea99d3c63c9d4b94708c008f5eb3bfd80ed7a6af_1.jpg (90KB, 600x412px) Image search: [Google]

90KB, 600x412px

>>61351849
Don't forget the $7000 price tag you pay is for innovation

>>61351849
Innovation!
Now pay up, goyim.

juranga is going around damage controlling 4S+ Xeons against 2S EPYCs, lmao

>>61352188
Who?

>>61352194
You don't wanna know.

>>61352197
Sounds like some forumite is running around defending his Jewish masters after the gigantic failure that is SKL-SP launch?

>>61351781
>"It looks the same on the powerpoint slide, but they are very different". The place is Austin, where an AMD engineer is commenting on the slides describing the Zen and Skylake schematics. In Portland, the Intel representatives could not agree more: "the implementation matters and is completely different". "We have to educate our customers that they can not simply compare AMD's 32 core with our 28 cores".

>>61352079
>>61352092

Found the Intel marketing team.

Didn't mean to tag>>61352092

Meant for

>>61352078

>>61352267
it's time to stop posting

>>61352359
you're not funny

>>61352407
Kike.

>>61351862
After looking at the number of people who really do not fully understand the entire architecture and workloads and thinking that AMD Naples is superior because it has more cores, pci lanes etc is surprising.
AMD made a 32 core server by gluing four 8core desktop dies whereas Intel has a single die balanced datacenter specific architecture which offers more perf if you make the entire Rack comparison. It's not the no of cores its the entire Rack which matters.
Intel cores are superior than AMD so a 28 core xeon is equal to ~40 cores if you compare again Ryzen core so this whole 28core vs 32core is a marketing trick. Everyone thinks Intel is expensive but if you go by performance per dollar Intel has a cheaper option at every price point to match Naples without compromising perf/dollar.

>>61352448
DAMAGE CONTROL
A
M
A
G
E

C
O
N
T
R
O
L

>>61351902
>and a company that doesn't play stupid market fragmentation games.

I'm not familiar with the processor market, what did he mean by this?

File: just segment my shit up.jpg (338KB, 1360x2300px) Image search: [Google]

338KB, 1360x2300px

>>61352513
This happened.

>>61352533
That doesn't even mention the memory and multiple RAS features disabled randomly for each tier

>>61352448
>$0.13 has been deposited in your account
>Thank you for shilling for Intel!

>>61352255
No, I literally have only amd stuff and hate intel, but i'm interested in digital hardware design...

>>61352533
JUST FRAGMENT MY SHIT UP

File: intel.png (744KB, 750x1268px) Image search: [Google]

744KB, 750x1268px

DELTE THIS

>>61352077
GLUED TOGETHER

>>61352560
AND YOU STILL NEED TO PARTITION IT INTO 4 NUMA NODES FOR OPTIMAL™ GENUINE™ INTEL™ PERFORMANCE.

>>61351902
>[email protected]

>>61352563
OURS AREN'T GLUED, THEY'RE STITCHED!!

t. Intel Spokesperson

File: d5c.jpg (68KB, 600x737px) Image search: [Google]

68KB, 600x737px

OY VEY IT'S A SHEKEL SHOAH!

>>61352448
> AMD made a 32 core server by gluing four 8core desktop dies whereas Intel has a single die balanced datacenter specific architecture which offers more perf if you make the entire Rack comparison.

Sounds more like you're the one who doesn't understand "the entire architecture and workloads".
Also the fact that you use the Intel meme term "gluing" truly means that you know fuckall nothing and just goes by what Intel shitposts in their slides.

> Intel cores are superior than AMD so a 28 core xeon is equal to ~40 cores

By what definition? This magical "entire rack performance" variable you mentioned briefly?
Why are they superior? In what way are they superior? How did you reach this conclusion?

> if you compare again Ryzen core so this whole 28core vs 32core is a marketing trick.

Marketing trick by whom? AMD for pointing out that they have more cores? Or Intel for arbitrarily benchmarking server processors against consumer processors?

> Everyone thinks Intel is expensive

Well I mean... they are. For the performance and functionality you're getting they're hella expensive.

> but if you go by performance per dollar Intel has a cheaper option at every price point to match Naples without compromising perf/dollar.

Again, what performance? Where are the independent tests and reviews proving such?
Also what cheaper options? Judging from the SKU list, even a 4-core Xeon costs almost 4 times as much.

File: 1386064260873.jpg (71KB, 688x688px) Image search: [Google]

71KB, 688x688px

>bingmesh

>>61352755
but entire rack performance is a lot better on amd, you could cram ~1200 cores, ~80TB of RAM and 140 gpus on a 42U rack

>>61352777
>4S+

Neat, lets ignore how Intel seems to target under 9% of the market since 2S an 1S make around 90% of it.

Not very smart for the bottom line.

>>61351781
We take all competitors seriously, and while AMD is trying to re-enter the server market segment, Intel continues to deliver 20-plus years of uninterrupted data center innovations while maintaining broad ecosystem investments. Our Xeon CPU architecture is proven and battle tested, delivering outstanding performance on a wide range of workloads and specifically designed to maximize data center performance, capabilities, reliability, and manageability. With our next-generation Xeon Scalable processors, we expect to continue offering the highest core and system performance versus AMD. AMD’s approach of stitching together 4 desktop die in a processor is expected to lead to inconsistent performance and other deployment complexities in the data center.

>>61352808
4S+ is a shitty meme that scales like a turd, only retards and niche users buy this stuff

>>61352824
To expand on this, 4S+ is mainly only used in universities and academia.

It's a fucking mess, even Google and Amazon mainly use 2S racks

>>61352844
and to think that some people are only using 2S because they need the additional pcie lanes

File: 1489010760579.jpg (12KB, 184x184px) Image search: [Google]

12KB, 184x184px

>>61352448
>AMD glue their dies together
lmao, at least AMD don't glue their heatspreaders to the die with jizz.

File: desperate-fragmentation.png (2MB, 1360x2300px) Image search: [Google]

2MB, 1360x2300px

With so much fragmentation you can raise the price in 200$ steps from 200$ to 13,000$. But you still do not get 32 cores.
Good thing Intel gets called out for their bullshit.

Meanwhile AMD makes it easy for buyers and provides competitive and even better products at a good price.

>>61352844
>>61352824
>>61352872
Can anyone explain this to a normie?
I though those numbers denoted the size(height?) of components in a rack

>>61353177
1/2/4/8S is the number of sockets per mobo.

>>61353183
Ah alright ty
I have only ever seen 2S then

>>61352448
Is this a spicy new pasta?

>>61351781
i didn't know domestic systems had numa.