View Full Version : late 2013 mac pro LW benchmarks vs early 2009 mac pro.

03-04-2014, 11:54 PM
2014 mac pro:
32 gig ram
8 core
512 gig drive
2 @ fire pro 500
content directory installed onto promise raid in raid 10 config. (mirrored and striped)

2009 mac pro:
24 gig ram
8 core 2.26 ghz
7200 rpm drive.
NVIDIA GeForce GT 120 512 MB

LW content directory from the LW 11 install.

bottom line: time cut by a third. see pic for details (values are in seconds)

03-05-2014, 03:01 AM
Is that the trashcan mac pro?

03-05-2014, 03:09 PM

03-05-2014, 03:11 PM
had a comp vs the last aluminum mac pro with 12 cores:
Mac pro 2010 12 core:

HDRI spheres: 52.3

Radiosity box_MC: 16.4

Radiosity box: 6.3

Radiosity_reflective things: 12

Radiosity things: 3.4

Raytrace test 01: 25.8

Raytrace test 02: 5.4

Teapot raytrace radiosity: 44.7

Textures: 1.5

Tracer no radiosity: 30.0

Tracer radiosity: 28.4

Z buffer sort: .05

just about the same, but the 12 core edges out the 8core in a few areas.

03-06-2014, 01:31 PM
well the marbles scene (Lightwave 11-6 Content<Scenes<Benchmarks) is the one everyone benches against and my new Mac 12 core is beaten by my old 2011 12 core Mac tower by around a minute on that, the new one is a way better machine in every respect but it won't be till they bring out a 16 core that most people should consider upgrading

03-10-2014, 10:17 AM
I really hope the next MacPro after this one at least offers the option of dual-socket CPUs. I know Apple are really betting on GPU's displacing CPUs for most computational tasks in the long run, but right now the lack of CPU oomph is really annoying for 3D artists

03-10-2014, 12:16 PM
Yes and no, I came to the realisation that for my day to day rendering where my average render times were between one and two minutes on a twelve core whether it be new Mac or old tower, going sixteen core would be nice but would not make a huge difference, and for animation needs one sixteen core is little better than one twelve core either, thing about the new Mac Pro is it makes a great render node, it's small kicks out next to no heat and make no noise, have two on the desk and can't wait to get rid of the two towers in the room when I replace those later in the year, For me it is not the fact that they are no faster than the old ones it is whether open CL will gain the traction that Nvidia's Cuda enjoys at the moment

03-10-2014, 02:22 PM
A dual socket 16 core machine (2 x 8 core) would be perfect for me. And yes, hopefully this hardware will help push for greater things for OpenCL. I'm pretty sure that's exactly why Apple picked AMD instead of Nvidia for their GPUs. They are an oddly principled company, in their own way. They don't mind trying to push the market in the way they think is best for everyone (see also the original iMac - which popularised USB)

03-24-2014, 03:22 PM
Quick question- how does Layout like the Firepros? Layout can get so sluggish on my 2012 Mac Pro its infuriating.

03-24-2014, 05:54 PM
would not honestly say its that much different, VPR – slightly better, few million ploys in a scene – can't really tell the difference, 2 million instances load time – as abysmal as ever, I don't think OGL is that well supported on AMD by NewTek so its crap whichever AMD card you have

03-24-2014, 10:43 PM
also, LW is very CPU bound.