i7-4770K -8 MB or i7-4930k -12 MB ! Short on budget

[gabmass]

Hi there,

I really need your advice. I'm so stuck between two CPU choices. with different sockets !

 

i7-4770K -8 MB

LGA 1150

4 Cores / 8 threads

 

 

 

i7-4930k -12 MB

LGA 2011

6 Cores / 12 threads

 

 

the price almost gets doubled from where I intend to buy, is it worth it ? the advantages for 4930k are having 2 more cores and 4 more threads. plus being able to support 64 GB Ram.

so for architectural vray and 3ds Max modeling, how different are these 2 from each other? does it make significant difference ? How much faster could my Rendering Time get with 4930k rather 4770k?

I'm not a real advanced user but in a year time I plan to start taking up animation courses as well.

 

(I heard 3ds Max and vray are not multi-thread supporting programs, if true so I find 4930k a real waste of money )

 

Here are detailed comparisons

http://ark.intel.com/compare/77780,75123

http://www.anandtech.com/bench/product/836?vs=552

 

thanks in advance

Edited November 11, 2013 by gabmass

[komyali]

If you ask me I would go 4930k, 2 more cores you will feel the difference... and both of them you can overclock, buy strong cooler! I had quad core computer and now I am using six core it is big difference

[Stephen Thomas]

3ds max and vray are most definitely multi-threaded during render time, so you will see a marked increase in render speed with 6 cores over 4.

[bulenthalibram]

If you have short budget, 4770K is preferable, because LGA 2011 motherboards are also more expensive.

But if you have the money you should definitely pick 4930k, except the extra cores it supports 4 memory channels, which means, if you make a config with 4 memory modules, you will have 2 times faster communication between the RAM and CPU.

[gabmass]

ok then so I'll go for 4930k

[numerobis]

...4930k, except the extra cores it supports 4 memory channels, which means, if you make a config with 4 memory modules, you will have 2 times faster communication between the RAM and CPU.

theoretically... yes...

 

practically...

http://www.overclockers.com/gskill-ripjaws-z-ddr3-2133-quad-channel-ram

 

http://www.tweaktown.com/articles/4416/intel_x79_quad_channel_and_z68_dual_channel_memory_performance_analysis/index3.html

Edited November 16, 2013 by numerobis

[joelmcwilliam]

Have a look at this. It might help with your decision.

 

http://nl.hardware.info/reviews/4585/2/48-desktop-en-66-mobiele-processors-getest-in-cinebench-115-resultaten

 

http://www.hardware.fr/articles/905-4/performances-applicatives.html

 

http://prohardver.hu/teszt/az_intel_core_i7-4930k_es_i7-4820k_tesztje/rendereles_tomorites_cpu.html

 

http://www.kitguru.net/components/cpu/zardon/gigabyte-ga-z87x-oc-motherboard-review-w-intel-i7-4770k/17/

 

Btw. In singlecore performance, like when you are modelling the i7 4770 (K) is better than the i7 4930k (hexacore).

Edited November 16, 2013 by joelmcwilliam

[RyderSK]

 

No difference whatsoever

 

 

Glad you posted this, HW suggestions always turn into so much weird bro-science.

[Dimitris Tolios]

RAM hasn't been the bottleneck in modern CPUs for ages now...

Most people won't see a difference past DDR3 1600, tho 1866 is so cheap now that doesn't make sense getting 1600 in most cases.

None of this difference will be seen in rendering times ofc. RAM is dealing with very fast operations, so only highly repetitive, non-CPU limited stuff benefit from it.

i.e. the CPU has to be able to "finish up" the set it got from the RAM and forward it so fast, that the time wasted waiting for the next package is "meaningful"...that means that the CPU manages to push more than 10-12GB of data per second (DDR3 1333-1600 speeds), which simply doesn't happen that often at all.

 

In mainstream computing, the only reason to go above 1600 and notice performance gains, is gaming, and in most cases in games that already score pretty high fps. You have to have one of the fastest CPUs - probably already overclocked - and one or more of the very best GPUs to actually make the memory subsystem become the "bottleneck" of the whole operation.

 

Another scenario that might benefit, is when you are using the intel IGP (the graphics built into the CPU) or an AMD APU processor. Since those GPUs don't have any significant amount of memory built in, everything is stored in the main memory, and GPU tasks are far more often memory limited than CPU tasks.

Edited November 18, 2013 by dtolios

[gabmass]

wow this was really useful.

could you make it somehow easier to perceive ? let's say we have a 3D scene we are going to animate plus light, shadows and effects. with 2 scenarios that the viewport to work in is continuously updating by GPU with every single click and change we make. another scenario for non-gpu viewport updating ( I have no idea what makes it if non-gpu - maybe you know some other methods other than gpu). if so, how does RAM work in between CPU AND GPU for both scenarios? I'm trying to figure out the importance of memory for "highly repetitive, non-CPU limited stuff" ! I hope you get this right, I kinda think it's all confusing maybe just because I am.

 

RAM hasn't been the bottleneck in modern CPUs for ages now...

Most people won't see a difference past DDR3 1600, tho 1866 is so cheap now that doesn't make sense getting 1600 in most cases.

None of this difference will be seen in rendering times ofc. RAM is dealing with very fast operations, so only highly repetitive, non-CPU limited stuff benefit from it.

i.e. the CPU has to be able to "finish up" the set it got from the RAM and forward it so fast, that the time wasted waiting for the next package is "meaningful"...that means that the CPU manages to push more than 10-12GB of data per second (DDR3 1333-1600 speeds), which simply doesn't happen that often at all.

 

In mainstream computing, the only reason to go above 1600 and notice performance gains, is gaming, and in most cases in games that already score pretty high fps. You have to have one of the fastest CPUs - probably already overclocked - and one or more of the very best GPUs to actually make the memory subsystem become the "bottleneck" of the whole operation.

 

Another scenario that might benefit, is when you are using the intel IGP (the graphics built into the CPU) or an AMD APU processor. Since those GPUs don't have any significant amount of memory built in, everything is stored in the main memory, and GPU tasks are far more often memory limited than CPU tasks.

[thomasoxvig]

I just bought the 4770k, but had the exact same dilemma. Building a CGworkstation for rendering and modelling leaves the big question: If the 6 cores is worth the money over the 4 cores?

 

First of all remember you are probably upgrading from an old system, which will make the new setup perform very well anyway. The price of the 6 core is expensive, and in my oppinion the best value for money is the 4770k. I'll be doing a lot of practicing both modelling and rendering, but most for smaller personal jobs /open competitions (i'm not depending on very fast rendering). I don't think there is a need to pay that much for a 6 core, instead save the money for something else. And in 3 years from now you might upgrade again, and the HW will be a whole different league

 

Dimitri has done a briiliant job on his site, and he adviced me on this one as well. I will definitely recommend his site(direct link to midrange WS): http://pcfoo.com/mid-range-workstation/

As he told me; its not the machine who do the job, its the creative mind behind. People build CGarchitecture 10 years ago that were just as difficult as the things we are dealing with today.

 

Cheers,

Thomas