[Irinikus]

I've started this thread as I'm going to be releasing a series of videos showing the various differences in performance between the Entry and Max Spec versions of the Mac Pro 6.1, as I now have very nice examples of both of these machines.

This is my first video showing the Multi Core CPU performance difference between these two machines:

[Irinikus]

[Shiunbird]

I was considering migrating my FreeBSD fleet to trashcans but the rack mount kits are so expensive.

I ended up getting a bunch of dells and modding them with noctuas. I 3D-printed brackets and installed a bunch of exhaust fans (basically doubling the number of fans), and they are running nicely, eating 50W each at load.

And my classic Mac Pro has 4 nvmes (via sonnet card), so I can't justify the change either. So, well, I save money!

[Irinikus]

[Irinikus]

At this time I'm unfortunately not able to compare the GPU compute performance between the two Trashcans's, as firstly Blender 4.2 won't install on Mac OS Catalina, although I was able to get Blender 4.0.2 to install it requires Mac Os 12.3 in order to enable GPU rendering in the Cycles Render Engine!!!

I was however able to run Cycles with CPU rendering in Blender 4.0.2 and got the following result:



The D300's on the Entry Level machine are significantly faster!

I did compare the Heaven Benchmark performance between the two machines and got the following results:





According to this result, the D700's aren't significantly more powerful than the D300's, they just have 3 times the video RAM!
However, it would have been interesting to see if the difference is the same when it comes to GPU Compute with both D700's enabled?

[Shiunbird]

You inspired me to do some testing.
I couldn't get GPU in Cycles enabled in macos, but it works fine in Windows.

I got a cMP with dual x5675 (2x 6 cores @ 3.06) and a Vega 64.
The 3.3 Blender splash screen demo scene renders in 27 minutes on the CPUs and 7 minutes on the Vega 64 under the newest Blender in Windows 10.

edit: enabling both CPUs and GPUs raises the time to 11 minutes.

[Irinikus]

You inspired me to do some testing.
I couldn't get GPU in Cycles enabled in macos, but it works fine in Windows.

I got a cMP with dual x5675 (2x 6 cores @ 3.06) and a Vega 64.
The 3.3 Blender splash screen demo scene renders in 27 minutes on the CPUs and 7 minutes on the Vega 64 under the newest Blender in Windows 10.

edit: enabling both CPUs and GPUs raises the time to 11 minutes.

Cool! 
I have a solution to get Monterey running on the 12 Core machine! I’ll install Monterey onto one of these and select it as the startup disk when I want to run Monterey!


The 4TB OWC SSD’s absolutely staying in the machine, as it’s an extremely rare item and thus more valuable than the machine itself!