cinema and dual core?

[Per-Anders]

Yes a dualcore 2.8 is a lot faster than a single with ht.

Yes C4D supports dualcore and all the rest. If the chip and OS support HT then you can also HT each core to get a little more performance out of it.

[moka.studio]

I have a dual core Athlon.
Performance is ca. 1.86 that of a single core equivalent.

www.imashination.com shows you some test results for DualCores, with one and multi processor rendering.

[dann_stubbs]

I know this is kinda more a hardware question, but does dual core in the intel range give significant speed increase. For example would a dual core 2.8ghz be alot faster than a hyper-threaded p4 2.8ghz?

And does c4d even support dual core for rendering? Oh yeah and can you hyper thread each core to get even more performance? :D

you can use HT on a dual core only if the chip supports HT - it is a hardware thing not just a software setting to choose. the extreme editions of the intel dual core chips are the ones with HT.

so with a dual core extreme edition (HT) it would look like 4 processors to the system - but HT is not as fast as 4 real processors or two dual core chips.

dann

[designbytes]

Yes a dualcore 2.8 is a lot faster than a single with ht.

Yes C4D supports dualcore and all the rest......

Hi,

would this be the case with the new G5 dual cores as well?

I am trying to patient and wait 'til someone has these in their hands before ordering one, but don't know if I can hold out! ;-)

thanks

[Srek]

G5 Dualcore or Pentium / Athlon Dual Core makes no difference at all to CINEMA. You can expect about the same speed increase as with a dual processor system, namely a speedfactor of up to 1.85 during rendering.
Cheers
Björn

[moka.studio]

You can expect about the same speed increase as with a dual processor system, namely a speedfactor of up to 1.85 during rendering.
Cheers
Björn

nope, 1.86 ;)

I ahve to say, the speed increase on my dual xeon was 2.12...
So dual cores are not as efficient. But they are more efficient if you look at the render speed to price ratio...

[dann_stubbs]

nope, 1.86 ;)

I ahve to say, the speed increase on my dual xeon was 2.12...
So dual cores are not as efficient. But they are more efficient if you look at the render speed to price ratio...

do your xeons have HT? i've seen varying levels of dual core improvement ratios - regardless there is some sort of overhead for multiple cores or multiple processors on one motherboard.

dann

[moka.studio]

Yes, the Xeons where Cinebenched with HT.
The ration is taken from the same test, so HT was enabled for the singleCou and Dualcpu tests.
It always seemed baffling to me that you woul dget a score over 2.0 when using a dual processor machine (I guess it has to do with sharedroutines?)

[Venkman]

Yes, the Xeons where Cinebenched with HT.
The ration is taken from the same test, so HT was enabled for the singleCou and Dualcpu tests.
It always seemed baffling to me that you woul dget a score over 2.0 when using a dual processor machine (I guess it has to do with sharedroutines?)

I believe the "over 2.0" increase has to do with the hyper threading. I bet if HT was turned off that dual xeon would have been closer to the typical ratio.

[moka.studio]

I believe the "over 2.0" increase has to do with the hyper threading. I bet if HT was turned off that dual xeon would have been closer to the typical ratio.

actually, you may be right, since normally the HT threading is shown on BC tests only on the multi CPU score. So that the single CPu score on a dual processor machine may be the
Single processor without HT, and the Multi processor score may be both processors +HT for each of these. that would seem to make sense...

[Thalaxis]

actually, you may be right, since normally the HT threading is shown on BC tests only on the multi CPU score. So that the single CPu score on a dual processor machine may be the
Single processor without HT,


Not may -- it IS ignoring HT. HT doesn't do anything with only one thread running.


and the Multi processor score may be both processors +HT for each of these. that would seem to make sense...

Yup. It's an illusory supralinear performance improvement. Just don't let thet marketeers find out or we'll never hear the end of it ;)

[Tyler_Durden_83]

yep, cinema completely supports multicore hardware. it would be weird if it didn't, as rendering is one of the simplest things to make multithreaded. anyway, i know this post isnt exactly about it, but do not buy a dual core intel system. this is not the place to point out how and why the intel dual core solutions are far from being an exemple of flawless engeneering result, so i'm gonna stick with a simple value comparison. These are the results from a cinebench topic from an italian c4d forum:

Workstation Dell Dual Xeon 3,2 GHz (HT active = 4 threads):
Rendering (Single CPU): 596 CB-CPU
Rendering (Multiple CPU): 278 CB-CPU
Multiprocessor Speedup: 2.15

Apple g5 MHz 1800 (2 threads)
Rendering (Single CPU): 259 CB-CPU
Rendering (Multiple CPU): 463 CB-CPU
Multiprocessor Speedup: 1.79

Apple G5 MHz 2000 (2 threads)
Rendering (1CPU) 287 CB-CPU
Rendering (xCPU) 509 CB-CPU
Multiprocessor Speedup: 1.77x

and my results with a dual opteron dual core @2.2 GHz

Dual Opteron Dual Core (4 threads)
Rendering (Single CPU): 310 CB-CPU
Rendering (Multiple CPU): 1013 CB-CPU
Multiprocessor Speedup: 3.27

While in the image are the times to render another scene, including all the desktop dual cores from intel and amd. My time in that scene is 27 seconds.
Finally these are the times to render a testscene posted there (if you want i can attach the scene):
INTEL PENTIUM 4 3,2 HT ---- 4min 20 sec
G5 dual processor 2Ghz ----- 3 min 55 sec
Dual Xeon 3.06 GHz ---- 2 min 45 sec
Dual xeon @ 3,2 ---- 2 min 17 sec
Dell PowerEdge quadri cpu (4 Xeon 2.5 GHx HT, 8 threads), 4Gb di ram ---- 1 min 30 sec
my configuration ----- 53 sec

The conclusions you should get from my post are:
1)cinema is heavily multithreaded and gets major benefits by multicore, and the best desktop segment cpu is the p4 840 ee as its the only desktop cpu that can execute 4 threads
2)if u want a multicore you should see that a dual opteron dual core @ 2.2 4 threads and a time has beated a quadri xeon with ht capable of dealing 8 threads and @ an highter frequency. altho there are many other factors to consider when judging the hardware, this is still undeniably something you might keep in mind
3)the price, i didnt consider the prices of each solution, and probably the most powerfull ones are too expensive unless your cousin has an hardware shop ;), so consider my post as a prove to show you how much it supports and gains from multicore, not as a guideline to the best power/price system to buy

[Thalaxis]

A simple ruleset of thumb look something like this:
if you're on a tight budget, dual-core P4
if you're primarily concerned with OpenGL performance, the highest-clocked Athlon64 you can buy
if you're on a decent budget and want a good balance between OpenGL and raytracing, Athlon64 X2
if your budget is pretty high, go for a dual Opteron, and dual-core procs if you can afford it
if you're spending someone else's money, go for the quad-Opteron box, and quad-Opteron add-on board,
and go with dual-core Opterons

:D

[chromecity]

Tyler, you've got a typo in your CB entry for the dual Xeon. It says "278" when it should in fact say "1278". That had me scratching my head for a while until I did the math with the mulitiplier...

[Tyler_Durden_83]

ops, sorry about that, but i just cut and pasted the results from the italian c4d forum i wrote about, so the real owner of the machine must have made the mistake in the first place. anyway i'm sorry about the headache ;)

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.