• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

What's the quickest processor currently available?

Sorry to jump into this thread but what the fastest laptop chip then? Assuming a "standard" chip again of course.
 
Last edited:
Surely the fastest chip is one in a delivery van ;)?

Joking, but why do you need a super fast chip? 3930k sounds like the CPU you want if you want fast, but it depends on your usage.
 
MaxP779 said:
I think they can go pretty far with bclk on 2011. I remember seeing a guy on this forum who has an overclocked i7 3820, the quad core locked chip.
Click to expand...

you can overclock the Xeons via the BCLK, but relative to the 3960X that's small fry, by way of overclock. Most applications can't take advantage of all those threads (unless you fold for instance)

elroberto said:
Wasn't there a quad CPU watercooled and clocked server system on the forums recently?
Click to expand...

yes. See Biffa's build log.... http://forums.overclockers.co.uk/showthread.php?t=18413298

Several people have built SR-X motherboard based systems for folding, and while they look tremendous, they are fundamentally constrained by the lack of overclock, leaving them as rather underwhelming builds.

Fastest chip depends on your application. In most instances the 3770K will keep up with the 3960X, it's only when you start multithreading and using MASSES of memory that the 3960X comes into it's own.

I'd recommend the i7 9X series. They are STILL competitive with AMD's offerings, cheap, and highly overclockable. New intels run cooler, and overclock easier. I'd invest in good cooling and push your 920 a bit harder rather than look to upgrade just yet.

The main reasons to upgrade PERFORMANCE BENEFIT wise from the new platforms are:

1. SATA 3
2. USB 3
3. PCI-e 3

the cpu's aren't that order of magnitude faster - i'd say, just much more efficient.
 
Last edited:
Interesting. Thanks for the additional information. I need to look into the cost/return of multiple cheap systems too - a diy cluster of cheap hardware vs one high end box.

For the curious, this would be for quick & dirty computational fluid dynamics work. Anything seriously intensive is likely to run on someone elses cluster, but there's normally a queue. A quick home computer would help with testing/prototyping.
 
JonJ678 said:
Interesting. Thanks for the additional information. I need to look into the cost/return of multiple cheap systems too - a diy cluster of cheap hardware vs one high end box.

For the curious, this would be for quick & dirty computational fluid dynamics work. Anything seriously intensive is likely to run on someone elses cluster, but there's normally a queue. A quick home computer would help with testing/prototyping.
Click to expand...

I used to do fluid dynamic simulations using RealFlow and my 920 OC to 4.2 was slow. Moved to a 990x and the difference was very noticeable with 12 threads. Now on a 3930k OC 4.8 and 12 threads and yet another step change.

Depends how much you want to wait and how deep into fluid dynamics you want to go.

The 920 could cope ok with small amounts of particles, so was ok for simple (very simple) modelling using Realflow and under 100k particles. Anything more than that and it started to creep.
 
If that's the case, then assuming the software you're using is well multithreaded, Jokester's idea in post #7 might actually be the best for you. 32 threads is insane :o.
 
Well, my 920 is definitely slower than I would like. I'll keep an eye out for a second hand x58 hex core, but will probably be moving platform.

The Xeon Phi looks staggeringly good. Approximately 64 x86 cores on a pci-e card. That pretty much kills the cluster idea, will wait to see how it progresses. Otherwise I suppose I could bite the bullet and embrace cuda, but I'd rather not.
 
It depends on budget then, but something like a quad opteron 61xx would be a beast (32 cores), perhaps 4 times quicker than the fastest single chip intel depending on compiler options and how parallelisable your code is. Four 6134s for example would only put you back £320 (ebay). Mobo choice is tricky though.
 
I would wait for Xeon Phi agaisnt cuda, unless you want to rewrite any code you might have, I would think it wont be too long untill it released.

Do you write your own code or just use off the shelf apps? If you write your own code make sure you use best performance compiler open source wont cut it if you care about performance
 
David_Vaughton said:
I would wait for Xeon Phi agaisnt cuda, unless you want to rewrite any code you might have, I would think it wont be too long untill it released.
Click to expand...

I didn't realise these Phi's were so efficient. I've read 1 TFLOPS for 300 W. That's 3.6 times more power efficient than a 3930K (120 or so GFLOPS for 130 W). Brilliant. All the HPCs will be using them soon.
 
You must log in or register to reply here.
Back
Top Bottom