/tech/ - Technology★

>>>/g/

>>802625

>sages a thread at the top of the catalog

You didn't think sage was a downvote did you?

>tfw still no compelling reason to upgrade from my 6700k

RIP intel to THREADS xDDDDD

>carlos.jpg

>buying literally backdoored CPUs in 2017

>buying a CPU from a brand that has a track record of unethical and illegal business practices

Not that AMD is any better but fuck Intel,

A socket change a year keeps the goyim in check.

>>802655

*goyim in fear

You didn't even get the phrase right. And Skylake and Kaby Lake both use the same socket

>>802659

Sure, but Coffee Lake doesn't.

>>802670

Because muh memory speed boost or some retarded shit

bet it has extra botnet that consumers love because it lets them watch tv

>>802643

>Not that AMD is any better but I have autism so fuck that other company in particular

>>802670

Coffee lake does use the same socket but it will fry if you put it in a kaby lake motherboard. Intel didn't even change notches on the cpu to prevent it. Also they're changing chipsets once again for no reason other than shekels.

>>802720

>but it will fry if you put it in a kaby lake motherboard.

Source? I know the socket is the same while the pinout is different, but I figured it would've just failed at the BIOS/UEFI with a CPU error POST code. Does it really fucking fry boards?

>>802674

It's not even 7%

Taking figures from here: https://images.anandtech.com/graphs/graph11859/91901.png

... the 8700K is 5.6% faster than the 7700K

BUT.. they're using the stock multiplier, which is 47x for the 8700K and 45x for the 7700k

Since they're both unlocked, there's no actual reason to do it that way. The actual clock that you would end up using depends on whether CL is more overclockable than KL, which is not something we know yet.

When you take out the artificial multiplier change, the actual IPC difference is about 1%

Unless it turns out to be an OC god I'd say Coffee Lake is pretty much a scam.

well well Intel had to finally get off the ass and make a proper cpu after screwing their customers for decade, and its still fucking hot like the i9, its the performance king thats for sure

AMD will probably still stay the best-price-for-performance-king for while thou

>>802723

OK I was just being a bit dramatic, I don't think anyone has tried it yet.

What's the point? AMD already said they're keeping their sockets around at least until 2019, and Thread Ripper and AM4 are going to get two more iterations of Zen before a possible change.

If you go with AMD, in two years or whatever you just buy a new CPU and drop it in your existing motherboard. Or you can get Intel and either throw your motherboard away after a few years, or be stuck on the same quad core for 5 years because Intel hasn't innovated enough to give you a reason to upgrade.

>>802729

wow amazing. 4% faster clock speed, 5.6% faster performance, and all results were done by a marketing site that has Intel as a client.

wtf I love Core i9 now xD

Just don't buy anything Zen+/Navi and before, AMD is in a transition phase.

Thank you based Lisa Su

>>802624 (OP)

X86 architecture is dead since the 80386.

POWER9 is the only sane choice that we have:

https://www.raptorcs.com/TALOSII/

>>802781

What is it compatible with?

>>802781

>The architecture with the highest performing chips and is the choice architecture for supercomputers, servers, and engineers is dead

Wew lad keep LARPing

>>802760

>>802729

Intel single core performance was already ahead of Threadripper as it was and now they have a high core count chip for consumer prices AND significantly lower TDP

Face it, AMD got #Jew'd

>>802809

After all this bullshit about 14nm++ they turn it in and the IPC is virtually identical? Yes it has two more cores. We knew that. Any expectations of improvement otherwise have been totally dashed. I fail to see how this could be worse for intel.

PS I haven't bought an AMD since the Athlon 64

>>802624 (OP)

>What do?

I still buy the CPU that doesn't downclock itself 1GHz if I try to use its main new features

>>802823

10nm is fucking hard to pull off apparently, and it's why the only 10nm chips right now are RISC processors from Qualcomm Samsung and Apple

I don't expect 10nm to be that big a jump from 14nm in terms of TDP though. At nodes this fucking small you really start seeing diminishing returns. The jump from 65nm to 45nm was huge in comparison

>>802624 (OP)

Get fire extinguisher ready

>>802782

x86 emulators.