Why does an Intel platform from 2011 have better memory performance than Zen 2 from 2019?

Why does an Intel platform from 2011 have better memory performance than Zen 2 from 2019?

Attached: DDR3.png (539x518, 108K)

Other urls found in this thread:

arxiv.org/pdf/1902.07609.pdf
twitter.com/NSFWRedditVideo

Why did AMD have to create a real 64-bit instruction set for your Yiddish friends?

>AMDpoos can't make up an excuse
pathetic

Here have a you I feel bad for you.

This is one of the shittiest threads I've ever seen on Jow Forums.

> be AMD
> forced to pay licensing fees and royalties to intel for decades
interesting.

>needs quad channel to compete
>less cores, same clocks, more power
Nah OP I really don't think your dad is going to start respecting you if you fight back at night.

Why do all Nvidia cards have higher DPC latency than a 1st gen GCN card?

I get better RAM latency than that on X58 with 1800Mhz, the fuck are you doing.

SO Intel's Ring buss of fire is still more efficient than AMD's infinity fabric.
Ring buss however doesn't scale well past 10 cores while AMD's fabric can scale 64 cores.

I asked yesterday for proof regarding the real world consequences of these latencies, but got none. Is this a meme or just mental gymnastics done by Intel shills?

it does matter, just not as much as Intel shills like to think.
it's not much of an issue now.
look at the 1800x vs 2700x vs 3700x.
the latency drops every generation by quite a lot.
but the 1800x was a shit CPU due to the horrid CCX latency.

>quad channel vs dual channel
>shit tier cache bandwidth
incel cope

Why Zen 2 only has two memory channels when Intel platform from 2011 has four? Why Zen 2 cannot hit even close to 4.5GHz all-core overclock when Intel platform from 2011 can?

Attached: hmm.png (680x521, 115K)

Zen 2 has eight memory channels.

>Made the entire Intel HEDT line-up at the time obsolete
>Shit CPU

pick one

>look at the 1800x vs 2700x vs 3700x.
>the latency drops every generation by quite a lot
Zen 2 has higher memory latency than Zen+ did though...

And?

Why Intel from 2011 has better memory performance, even when Zen2 from 2011+8 has more memory channels?

It doesn't. Here's a pic from a Zen 1 CPU.

Attached: der8auer-Epyc7601-OC-02_7556E908CCE24113AC7A417D8F892818[1].jpg (1360x765, 177K)

What an abnormal latency... I hope you didn't think greater latency is better, did you?

We are talking about Intel from 2011 vs Zen2, don't bring Zen1 here.

You do realize that the extraordinary amount of cache that Zen 2 has means that RAM latency is less important, correct? There's a reason that two $10000 Xeons can't compete with one $4000 Epyc.
lmao

Latency doesn't matter for normal desktop systems and it hasn't mattered for almost 20 years now. Pre-fetching and beefy L2/L3 cache easily mask any potential shortfall from it.

It is a different story if you are dealing with HPC/clustering computing stuff a.k.a anything beyond two sockets/chips.

It is just another round of cope by computer illerate /v/tards and shills hailing from a certain team.

My motherboard only has four slots and OS says only two channel. How I enable other six channel?

Attached: dynamite.png (850x680, 267K)

I know, but shills still make two threads each day. I wanted to know where they get their bullshit from. There is nothing to support their claims outside of CPU-Z benchmarks and games where Intel wins due to higher frequency and 10 years of exclusive optimization.

x79 is quad channel. X58 is triple channel.

More memory channels is gonna introduce latency.

...

Install scrot.

>tfw feels bad

Attached: cachemem.png (539x518, 92K)

Zen 2 Ryzen is dual-channel
Zen 2 Threadripper is quad/octa-channel depending on the chipset
Zen 2 Epyc is octa-channel

Consumer CPUs are typically just dual-channel, notice how the OP pic is a Xeon.

DPC got nothing to do with graphics, only CPU and Chipset

Wrong, you can Google it. It's a super well known issue with Nvidia hardware.

Don't know about games where this is usually a topic of discussion, but some scientific applications are sensitive to RAM latency, so much so that you sometimes even get a regression in performance moving from DDR3 to DDR4 due to the trade-offs made with increasing bandwidth in the latter

arxiv.org/pdf/1902.07609.pdf

ESL go to hell

>forced
Noone really forced them to literally copy intel's technology - that's how AMD grew to be what it is today. Started making their own x86 designs at one point, but it all started with blatant copying.