Question About New Mac Pro vs Old Mac Pro Rendering Applications

[KBS756]

I currently have a 2009 Mac Pro with 2 x x5680's 12 Core 3.33 GHz

Was wondering if the new Mac Pro 12 Core 2.7 GHz will be able to post the same or better times in applications such as 3D rendering in Lightwave (CPU Based) as my existing Mac Pro or if it would be a step back of sorts.

Thanks!

I mostly see benchmarks that don't necessarily saturate all the cores so was wondering if anyone had some input on the way these 2 stack up thanks!

 

[sirio76]

This give you an approximate idea of the performance against the old 12core models(also with upgraded CPU).
http://www.barefeats.com/tube01.html
The 12core nMP seems to be significantly faster than any stock old 12core and even slightly faster than upgraded models. Don't know if lightwave support some forms of distributed rendering, but generally is much more cost effective to invest in a 6/8core model + one or more cheap pc nodes.

 

[flat five]

in the scenario you've layed out, you'll probably see slight increase in rendering times.

but, you'll probably only see those increases if you sit there with a stopwatch etc.. as in, I doubt you'll feel/notice much difference at all.

(but if you get a new 12core and keep the old one as a rendering node, then you'll notice the difference )

 

[Spinland]

Was wondering if the new Mac Pro 12 Core 2.7 GHz will be able to post the same or better times in applications such as 3D rendering in Lightwave (CPU Based) as my existing Mac Pro or if it would be a step back of sorts.

Great to see another Lightwaver here!

I don't have a Mac Pro yet to compare, am awaiting my custom hex core order, but based on years of distributing LW renders over multiple machines/cores I believe you'd probably do better with your current rig. Unless you force LWSN to use multiple cores per frame you are going to 100% see faster cores do better than slower cores.

For example, a quad core 2.0GHz i7 will do a single frame faster than a dual core 2.66 GHz i7, but if I split things up so the quad core does two frames at a time (four threads per frame) and keep the 2.66 dual core on one frame (again, four threads), the 2.66 will win.

If you let all 12 cores work a single frame things might get murky depending on how LW plays nicely with Turbo Boost, but if you distribute your renders so, say, you're working on 12 or 24 frames at a time, your faster existing cores will generally win out.

----------

(but if you get a new 12core and keep the old one as a rendering node, then you'll notice the difference )

Oh, man...the idea of having two 12 core Xeons as render nodes makes me salivate. I need to land some more animation gigs.

 

[KBS756]

Great to see another Lightwaver here!

I don't have a Mac Pro yet to compare, am awaiting my custom hex core order, but based on years of distributing LW renders over multiple machines/cores I believe you'd probably do better with your current rig. Unless you force LWSN to use multiple cores per frame you are going to 100% see faster cores do better than slower cores.

For example, a quad core 2.0GHz i7 will do a single frame faster than a dual core 2.66 GHz i7, but if I split things up so the quad core does two frames at a time (four threads per frame) and keep the 2.66 dual core on one frame (again, four threads), the 2.66 will win.

If you let all 12 cores work a single frame things might get murky depending on how LW plays nicely with Turbo Boost, but if you distribute your renders so, say, you're working on 12 or 24 frames at a time, your faster existing cores will generally win out.

----------



Oh, man...the idea of having two 12 core Xeons as render nodes makes me salivate. I need to land some more animation gigs.

Thanks for the information was very helpful, may hold out till i can afford to buy one without selling my old one in that case.


I am a student right now so I haven't tried working lightwave across multiple machines but it is something I hope to do in the future. Thanks!

 

[echoout]

(but if you get a new 12core and keep the old one as a rendering node, then you'll notice the difference )

That's exactly the approach I took with keeping my 12-core MP and adding a 16-core z820. The two Team Rendering C4D files together is fantastic.

 

[riggles]

As a MODO user, I've found the Geekbench 64-bit multithreaded benchmarks to be at least a decent starting point when comparing Macs for rendering speed estimates. So, I believe the newer Xeon might have slightly faster renders, perhaps depending on how much single threaded pre-computing or tesselation is taking place (it turbos up to 3.5GHz), but it will cost you almost $8,000 to gain that little bit. You could sell your current Mac Pro to help offset the cost, but you still have to ask yourself if it's worth it. For me, with two X5670s, it's certainly not.

Edit: You could go the route of keeping your existing MP and using as a render node, yes. But I just think that if you want more rendering power and stay with Mac, buy another 12-core old Mac Pro for half the price of the new one and get similar rendering performance.

 

[fig]

I'm amazed there's any LightWave users around at all anymore

riggles, hoping that Modo will be nicely optimized for the nMP with the relationship we've seen between The Foundry/Luxology and Apple.

I'm really waiting for someone to run some tests with Maya, Modo, etc., so we can start to see what the real world performance of the new machines actually is. Benchmarks are nice but don't tell the whole story.

 

[riggles]

riggles, hoping that Modo will be nicely optimized for the nMP with the relationship we've seen between The Foundry/Luxology and Apple.

Me too!!!

Unfortunately, for timing sake, I had to buy a workstation recently and couldn't bet on eventual OpenCL support for MODO since there was no word on anything from them. Went the used Mac Pro route, which should hold me over for a couple years at least. Especially with Octane coming soon

 

[flat five]

That's exactly the approach I took with keeping my 12-core MP and adding a 16-core z820. The two Team Rendering C4D files together is fantastic.

thought about updating my 1,1 to an 8core and doing the same thing but i decided against it..

too much hassle (no OS support / space/ wires&converters (fw800->TB (the wifi is too slow.. if i could put it in a closet with 802.11ac wifi then maybe)

i'm going to donate it instead.

 

[echoout]

thought about updating my 1,1 to an 8core and doing the same thing but i decided against it..

too much hassle (no OS support / space/ wires&converters (fw800->TB (the wifi is too slow.. if i could put it in a closet with 802.11ac wifi then maybe)

i'm going to donate it instead.

I love the idea of a 4,1 an upgradable render node. So many deals to be had on 2009s, and should get really interesting in the next year. My 2 labs at school will be selling off 26 of them in one pop.