Who can't wait for the 16 and 32 core Zen chips...

Who can't wait for the 16 and 32 core Zen chips? Intel's 20 core (10 per CPU) dual Xeons aren't that far ahead in a benchmark that scales into the hundreds of cores.

Other urls found in this thread:

semiaccurate.com/2017/01/19/amd-infinity-fabric-underpins-everything-will-make/
twitter.com/AnonBabble

what pc workloads will actually use more than ~8 threads? like 0.1%?

>tfw no 256 core cpu to go along with my 16gib of rams

its not fair bros

VMs, with a 20+ core I can serve a fully functional desktop to 5-10 people with varying performance profiles from one computer.
Not to mention all the stuff that could be hosted.

Video encoding, 3D rendering, CAD, and general parallel computing.

But all you care about is gaymen, right?

If i have money to buy it yes

Is this how you justify your lack of interest? By espousing your lack of knowledge?

Unlikely to serve you if you're asking.
The market for these high core processors is a $100 billion annually one and it's continuing to grow, they have their uses.

I really hope those 16 cores aren't costing more than $1000

>I don't have a use for this and I don't know what you'd use it for so I must inform others that this is uninteresting.
I'll bet you're fun at parties.
It'll be interesting to see how this whole infinity fabric thing works. If this is to be the replacement for HyperTransport, then it has some big boots to fill. HyperTransport was a massive step forward when it first came out and while each new revision was evolutionary rather than revolutionary, they were all important developments none the less.

Yes, my interest in server chips is purely based on the bus they're using.

Hardware level multi-user GPU virtualization as well. Get a couple of the AMD dual cards and you can have a gaming pc in every room in the house.

Kafka, Hadoop, Spark, Cassandra, Openstack, Docker, Mesos, GDB, MAP, SLURM, Cactus, Chimera, Gadget, AcuSolve, OpenFOAM,

Generally a lot of shit.

Infinity fabric is based on hypertransport. They just added on to it.

They even call it hypertransport 2.0.

It new socket right? hoping it using the same am4, one can dream

how well does this actualy impact gaming though? not much so dont fall for it. no optimisation

>It'll be interesting to see how this whole infinity fabric thing works.

It's lost now in the annals of Google, but before Ryzen was released they said they managed to get equal latency across all cores and caches even on a 2S system.

Which is probably blowing hot air but AMD has a penchant for surprises

There's no space on AM4 to fit in that many pins needed for 16+ cores, it's gonna be LGA with 3500+ pins or something , either SP3 or SP4

Nah, it'll use gigantic LGA monster called SP3.

>They even call it hypertransport 2.0.
Source? That would be really weird since HyperTransport 2.0 was an actual thing back in 2004 and 3.1 was released in 2008.
>they said they managed to get equal latency across all cores and caches even on a 2S system.
I remember the hype around it. Seeing how it actually behaves in a multi socket system is a different matter though. It'll be interesting, to say the least.

I want 16c HEDT Zen so I can just spawn an 8c W10 VM arbitrarily for random gaymen and not have to close anything else down.

Also, 4ch DDR4, ~64 PCIe lanes, and presumably ECC would be nice too.

Do am4 and probably am4+ socket stuck at 8core ?

The 7nm Zen chips will probably go up to 12 cores on AM4*
You're not getting more cores on mainstream platform like AM4 before that.

I think I'll be fine with 8/16 for the lifetime of AM4.
still waiting for good mITX mobos though.

MOAR BING BUS :DDD

Even on 7nm there's still gonna be lack of pins on AM4, those 10+ cores will probably all be on AM4++
You're gonna be fine with 8 cores on the desktop until 2020 so I don't see the problem

>Video encoding, 3D rendering, CAD

thanks for making my point for me, 1% of general purpose use cases.

> general parallel computing

kek

the reason many-core cpus still haven't taken off is because you average joe shmoe has no use for CAD apps and CFD simulations. niche use cases like games have massively parallel gpus, cpu's are left in the dust.

Who the fuck cares? By your definition Intel should close its Xeon lineup because the average faggot doesn't buy it, well duh.

I'm not buying a CPU for others, but for myself.

>thinking $1000+ CPUs and 2S motherboards are for the average joe

You're on the technology board, please act like less of a retard from youtube.

>a lot of shit.

a lot of shit that's used by 0.001% of world's computer users

with Microservices, Service Bus, Docker, Virtualization, etc., it's easy for even retard tier single core software and solutions to benefit from 20 core servers

What are you even trying to argue here? Nobody ever said these are mass market chips

it's called 'economies of scale' you dumb cunt. it's not about YOU, it's about selling millions of units.

I'm not running a company, I don't care about that.
What's with you? Why do you care what people spend their money on?

>On the surface it sounds like AMD has a new fabric to replace Hypertransport but that isn’t quite accurate. Infinity Fabric is not a single thing, it is a collection of busses, protocols, controllers, and all the rest of the bits. Infinity Fabric (IF) is based on Coherent Hypertransport “plus enhancements”, at a briefing one engineer referred to it as Hypertransport+ more than once. Think of CHT+ as the protocol that IF talks as a start.

>Hypertransport lacked a lot of things that quite frankly were not on the roadmap when it was designed in the DEC Alpha days. It wasn’t really scalable as later Opterons brought into focus, had no real QoS capabilities, wasn’t very flexible, had no inherent security features, and lacked any authentication mechanism. If you throw in higher performance and greater efficiency you have the list of why AMD needed to redo their bus to get Infinity Fabric.

semiaccurate.com/2017/01/19/amd-infinity-fabric-underpins-everything-will-make/

Charlie goes into some detail about IF

How about compiling code?

Most code compiles in seconds anyway so 4 cores is better

So full of shit, who are you trying to market here?

...

Run wangblows ganooloonix and MacOsX at the same time on the same computer and switch between them at yhour whim?

>Most code compiles in seconds

hahahaaaohwow you dont work in the godforsaken software industry do you phew

i guess you have never written anything other than fizzbuzz ?

:^)

Pretty much all server workloads, but hey, you can't expect a Sup Forums gaymer kid to know that.