What would be better >a 12 core CPU >a single core CPU

>>61494827
second because Amdahl's law

>>61494827
if we're talking about current cpus, I'd choose the second one, would be a great DF machine

>>61494827
>tfw no 7700k single thread performance cpu w/ ryzen 7 multi-thread performance

>>61494827
Depends on what the intended use is.

In most cases the 12 core will be better.

whichever one i decide is cooler imo
litteraly the only reason to chose ryzen over until is because of the aeshetics

>>61497029
What the fuck causes this kind of mindset?

>>61497104
not being a poor retard, idk?
save your money and buy what you want instead of slaving over price to performance graphs like a litteral monkey

>>61496981
that's next year on zen2

>>61494827
Dualcore with 6x

>>61497129
I don't save my money to waste it.

What's better a 100GHz 1-core CPU, a 50GHz 2-core CPU, a 25GHz 4-core CPU, or a 10Ghz 10-core GPU?

>>61497286
is IPC the same?

>>61497658
Yes

>>61494827
I'd go for the latter because only a minority of programs are multithreaded and I too often have to wait for a long process to finish while it's using only a gay 12.5% of my cpu...

Dual core could be more ideal on average though

>>61497196
>tfw have to Just Wait™ to build my already outdated system that can barely play anything

>>61497732
I'll have the 25GHz quad core one

>>61497755
>buy AM4 based system right now
>slap a zen2 or zen3 cpu when it comes out if it's really needed
>don't need to change motherboards nor anything else

Less cores = more stable
So less core + more frequency and cache is always better.

>>61497798
there is zen3?

anon you are my savior.

by the way what ram do you recommend for AM4 boards 16GB 3200 MHz

>>61497005
nope, tard

>>61497841
go on >>>/g/pcbg

they'll know more about which ram is best for ryzen than I know

>>61497892
anon i love you

What has more horsepowers a 12 core cpu or 12 raspberry pis?

>>61497917
12 core cpu easily, the rPi has a really weak ARM processor

>>61497942
What if it's a 12 core arm processor?

>>61497129
do you buy apple products by any chance?

>>61497956
even if you could build a 12 core arm cpu that has the same clockspeeds as a rPi cpu, it still sucks

>>61497967
no because I don't think they're that cool

>>61494827
How about a single core with twelve threads

Checkmate atheists

>>61498282
POWER8?

>>61496932
Exactly what I was going to post.

Someday, Urist. Someday.

>>61498282

Didn't SPARC have like 6 threads per core?

>>61494827
We only do 12 cores because it uses less silicon space than a CPU that runs 12 times faster.
The ideal CPU would be a single core CPU with flexible hyperthreading, where the OS can specify a percentage of timeslice per thread.
But that sadly would be a LOT more expensive than 12 cpus.

>>61499713
graphmeme 1THz cpus when?

>>61499772
As soon the scientists can have their two hands to study it, because you know, graphene is a very exciting discovery.

>>61497129
>save your money and buy what you want instead of slaving over price to performance graphs like a litteral monkey
monkeys don't litterally slave over price graphs. It's the sort of things humans do that make us better than monkeys.

Have you ever heard the expression "I may be rich but I didn't get that way by [something retarded]?

Unless you're making about 200/hr it's worth your time to investigate the pc components you purchase. I make sure to do all my pc shopping is done on the clock for this very reason.

>>61499837
i investigate them thorougly and then buy whichever one is absolutely the coolest and most badass / explodes jews

12 core cpu with 1 GB L1 Cache or a 64 core cpu with 64 1 KB caches

>>61494827
the single core with 12x better single threaded performance.
There's significant overhead to thread scheduling.

>>61499932
Depends on the application.
If you gonna do shader like stuff, option B will work better.

>>61499713
>We only do 12 cores because it uses less silicon space than a CPU that runs 12 times faster.
What the fuck is wrong with you, /g/. You were bad years ago, but it seems you've only gotten worse.

>>61499864
>and then buy whichever one is absolutely the coolest and most badass / explodes jews
>absolutely the coolest and most badass
I'm sure you'll remain wealthy with that mindset.

>>61499977
how about instead of looking at price/perf i look at price/coolness

>>61499944
For every extra instruction per cycle you add, you double the size of the CPU due the register renaming, instruction reordering etc, unless you go with the VLIW meme that dumps the responsibility for making the code not interdependent on the compiler.

File: havemybabies.png (66KB, 160x154px) Image search: [Google]

66KB, 160x154px

>>61499837
>I make sure to do all my pc shopping is done on the clock for this very reason

my hero

>>61499932
L1 cache? The 64 core 1KB caches is the better option.

>>61494827
Single core always, just avoid retarder programmer block single thread apps

>>61499996
That's retarded thought, you're talking about the "coolness" of a fucking processor, you're never gonna look at the fucking thing.

It's not even like checking performance is hard when CPUBoss and such exist.

You're just stupid.

>>61500096
Lol you're retarded; if you can't analysis a cpu arch by eye you don't belong on /g/

>>61500047
Increasing speed means increasing cycles per second, you dunce.

>>61500291
If your CPU executes more instructions per cycle it will be faster as well.

>>61500329
No. is RISC "faster" than CISC?

It might be more efficient at a specific task, but that's different than increasing a processor's speed.

Increasing speed = increasing cycles per second, which is in fact MORE difficult to do on a larger die because you run into latency issues, which is the complete fucking opposite of your assertion that faster = more silicon space.

>tfw no quantum computer to solve all possible states of your program for instant execution.

>>61500416
I think this discussion is trying to define absolute truths to something that depends with the situation.
In general, a 1 Ghz CPU that executes 2 instructions per cycle will be as fast as a 2 Ghz CPU that executes 1 instruction per cycle.
But then the details make this annoying, as in some cases, the 2 Ghz CPU is the better choice, but in others, the 2 instructions per cycle option is better.

But my original actual point was that if you do two CPUs that execute 2 instructions per cycle, you will use less die space than a CPU that executes 4 instructions per cycle.

>>61500515
>I think this discussion is trying to define absolute truths to something that depends with the situation.

You're "absolutely" wrong. You're confusing speed with efficiency.

>>61500552
Okay, let's just say its a word definition error here.
But it's still cheaper to do a 12 core CPU than trying to somehow make a CPU go 12 times faster or is 12 times more efficient, or a combination of both that yields the performance of 12 cores.

>>61496981
that's called ryzen 7 dude

>>61494827
The 12 core CPU would be better, or if single thread performance that high is possible a dual core with 6x better single thread performance. Sure the single core with 12x better single thread performance would end up being slightly faster in some tasks, but with multiple cores you don't have to worry about one process that happens to want a lot of processing power at the time hogging your entire CPU and making all the software on your machine hang.

>>61501017
A 12x faster single core would get everything done 12 times faster whether or not it's actually multithreaded. Something hogging your CPU would be very unlikely.

>>61501168
I already do some tasks that can max out all 4 core on my processor for about a minute if I wanted (but I normally won't use every core).

>>61497104
Autism