Twilight Render User Forum

Fletch wrote:According to Intel's website, this cpu only has 4 cores and 4 threads, so it is going to be slower than most 4 core machines which are hyperthreaded, meaning it has 4 cores, but 8 threads... such as a Corei7.
RAM is great for multitasking, that is, you can render, and still do other work. But it does not actually compute anything for the rendering. It only keeps the information in instant access memory.
So it is the number of threads in the CPU that is most important.
Look at other computers earlier in this thread, and see how many cores/threads they have.

Thanks now I understand....is the Frequency and the TDP important?

Also I don't know much about those things... I had to google Hyperthreaded.... but a 6 cores 12 Threads is going to be faster than a 4 cores Hyperthreaded?

6 cores hyperthreaded to 12 threads will be faster of course than 4 cores hyperthreaded to 8 threads, yes.
The speed of the cpu is important. 8 threads at 2.2Ghz will render more slowly compared to 8 threads at 3.2Ghz.

But even so, 8 threads at 2.2Ghz (17.6 total) is better than only 4 threads at 3.2Ghz (12.8) if they are similar in architecture. Remember that the latest release multithreaded cpu's render better/faster than similar number of threads for older cpu architectures.

TDP - thermal design power does matter when it comes to your machine and cooling. It's more important in laptops. Either way, lower TDP is probably better, because it's easier to keep cool. If it has high TDP, (corei7 runs pretty hot when rendering - over 70C) then be sure you have a good cooling system/airflow to keep your system from throttling (slowing) down the Ghz while rendering for long periods of time.

intel core i7-6700 cpu 3,40Ghz
32Gb
Nvidia Quadro K1200

Windows 7 Pro
Dell precision T3610
Intell Xeon CPU E5-1650 v2 @ 3.50GHZ
16GB 1333Mhz

If you check page 2 there are the result of the same computer with my old processor (4 cores 4 threads) with the new (6 cores 12 threads) it's about half time

|
|
SketchUp 2016 Windows 10
|
|

I'm not sure if this is of any interest, but:

Specification
System 1
Intel i5 2500k (4 core, no HT)
GigaByte GA-Z68XP-UD4
1333Mhz 16GB RAM (non-ECC)

System 2
Xeon E5-1650v4 (6 core, 12 threads)
Asus X99-E WS
2133Mhz 16GB RAM (ECC)

Consistencies:
64-bit O/S (Win 7 on System 1, Win 10 on System 2)
Sketchup 2017 (same version on both)
Twilight Render (same version on both)
No changes to model settings made, just 'as it came').

Results
System 1:
Low: 22 secs
Med: 1min 17 secs
High: 5mins 51 secs

System 2:
Low: 10 secs
Med: 33 secs
High: 2mins 31 secs

System 2 screenshots

bikerchris wrote:I'm not sure if this is of any interest, but:

Intel i5 2500k (4 core, no HT)
vs
Xeon E5-1650v4 (6 core, 12 threads)

Results
System 1:.................. System 2:
Low: 22 secs.............................10 secs
Med: 1min 17 secs.....................33 secs
High: 5mins 51 secs...................2mins 31 secs

Nice comparison!
Shows that increasing thread count by factor of three does not necessarily reduce render times proportionately, but it definitely gives much quicker render times.

Fletch wrote: Nice comparison!
Shows that increasing thread count by factor of three does not necessarily reduce render times proportionately, but it definitely gives much quicker render times.

Glad you like! I do have an 8 thread laptop, I might give that a go too - as an in-betweenie with a very similar CPU speed, it might be nice to see.

Specification
System 3A
Intel i7 4810MQ (4 core, WITH HT, 2.8Ghz / Turbo up to 3.7Ghz)
Laptop form factor (Dell Precision M2800)
1600Mhz 8GB RAM (non-ECC)

System 3B (I disabled Hyperthreading for fun)
Intel i7 4810MQ (4 core, NO HT, 2.8Ghz / Turbo up to 3.7Ghz)
Laptop form factor (Dell Precision M2800)
1600Mhz 8GB RAM (non-ECC)

Consistencies:
64-bit O/S (Win 7 on System 3)
Sketchup 2017 (same version on both)
Twilight Render (same version on both)
No changes to model settings made, just 'as it came').

Results
System 3A:
Low: 16 secs
Med: 0min 53 secs
High: 4mins 15 secs

System 3B:
Low: 18 secs
Med: 1min 03 secs
High: 4mins 57 secs

Observations: In System 3 testing the CPU speed fluctuated and settled on 3.4GHz most of the time. Not due to thermal throttling. Using Core Temp 1.4.