Mac M2 Render Speed Test

[Pat Stanford]

Don Seidel Render Test Results

 

What is a better value calculation than the number of days to complete 1000 renders divided by the machine price?

 

Edited January 31, 2023 by Pat Stanford
Table Updated 11:17AM PST 1-31-23

[Pat Stanford]

Values in the table in italics are my best guess. Please let me know actual values so I can update.

[line-weight]

Glad to see 16GB M1 Mini winning on value 👍

[Pat Stanford]

I believe the calculation has smaller numbers as better. 😉

 

[John S. Hansen]

its not better to pay a more???  

So higher number is best - paying least for performance...

[ashot]

Render Time 15:00

OSX Ventura 13.2

VW 2023 SP3 (Build 684728) (64 bit)

2017 iMac Pro purchased for $6,600 from Apple

64GB RAM

3.2 GHZ 8 core Xeon W

Graphic card:Radeon ProVega 64 16GB

[Don Seidel]

HA ! $2000 buys you a respectable 4th place. 

[zoomer]

2 hours ago, Pat Stanford said:

Values in the table in italics are my best guess. Please let me know actual values so I can update.

 

My Ryzen and Mini Prices are pretty accurate.

[zoomer]

2 hours ago, Pat Stanford said:

I believe the calculation has smaller numbers as better. 😉

 

 

So we all have to buy M1 Mac Studio Ultras from now on ?

[Don Seidel]

1 hour ago, zoomer said:

So we all have to buy M1 Mac Studio Ultras from now on ?

or buy 2 M2 Pro Mini's . One can render while you continue work on the other. Wouldn't that, by definition, beat the fastest machine on the market? 😄

 

You get what you pay for, mostly. Everyone's needs are different. I always buy the most machine I can reasonably afford and never look back. Of course everyone wants a killer machine, but do you need it even if you could afford it? If my projects get that big and that complex, yes. Otherwise, nope...ego is talking. The Aston Martin will get you there in style, faster than the Toyota Camry. But the Camry will still get you there. 

 

M2 Mac Studios are probably 3-4 months out

[arqteran]

on Mac Studio M1 ultra, 48 gpu model... 6:16 vectorworks result, 6:57 from the moment you hit update...  

Edited April 5, 2023 by arqteran

[E|FA]

12 hours ago, Don Seidel said:

There is a BUG on the time clock at the bottom of the screen. Sometimes (not always) if you click away to another program or desktop, the stopwatch on screen reports a much longer time than reality...you have o render again to see correct time.

 

That improves my speed all the way to 38:59.  I think I'm still the "winner".  I'll post the hardware cost if & when I can find it.

 

On 1/30/2023 at 11:18 AM, E|FA said:

39:55 - 2019 iMac 3.7 GhZ 6-core Intel Core i5, GPU Radeon Pro 580 X 8GB, 40GB RAM

 

OS 13.2, VW 2023 SP3

 

[line-weight]

I don't understand the "value" figure.

 

For example M2 Max macbook pro gets 0.0016 and i9 iMac gets 0.0028. They both cost about the same but M2 Max is loads faster. This suggests lower number = better.

 

But ...

 

M1 Macbook Pro (0.0051) and M1 Mac Mini (0.0110) take roughly the same time for the render, with M1 Mac Mini being half the price. This suggests higher number = better.

[Don Seidel]

2 hours ago, line-weight said:

I don't understand the "value" figure.

The “value” factor, the start of this topic, was to show how much processing power you can get vs the money spent. Any laptop costs a premium for the feature of portability. You also have monitor, keyboard and mouse all in one. So it’s less relative value for the cost…. Generally you can buy a more powerful desktop for less money than the laptop. But people are willing to accept the laptop premium because they must have or really want a portable system.

 

Headless Macs, on the other end, don’t come with monitor, keyboard or mouse. ( It’s funny Apple describes “accessory kit” as included with the Mini and Studio Macs… the only accessory is a power chord). So while there are needed peripherals to operate the machine, the collateral cost can be mitigated by using 3rd party products, or extra used products one may have around.

 

so the Mini, as intended, is a budget solution. One can always spend more, and have a more powerful, faster system.  But can one get a budget Mac that will work well with VW ? Clearly the answer is yes for many users, now that the M2 Mini has arrived.

 

 

 

 

Edited February 1, 2023 by Don Seidel

[Don Seidel]

13 hours ago, Don Seidel said:

The Aston Martin will get you there in style

…when it’s not in the shop! 😜

[line-weight]

1 hour ago, Don Seidel said:

The “value” factor, the start of this topic, was to show how much processing power you can get vs the money spent. Any laptop costs a premium for the feature of portability. You also have monitor, keyboard and mouse all in one. So it’s less relative value for the cost…. Generally you can buy a more powerful desktop for less money than the laptop. But people are willing to accept the laptop premium because they must have or really want a portable system.

 

Headless Macs, on the other end, don’t come with monitor, keyboard or mouse. ( It’s funny Apple describes “accessory kit” as included with the Mini and Studio Macs… the only accessory is a power chord). So while there are needed peripherals to operate the machine, the collateral cost can be mitigated by using 3rd party products, or extra used products one may have around.

 

so the Mini, as intended, is a budget solution. One can always spend more, and have a more powerful, faster system.  But can one get a budget Mac that will work well with VW ? Clearly the answer is yes for many users, now that the M2 Mini has arrived.

 

 

 

 

I get the basic idea of value - just meant that the numbers in the table don't seem to make sense.

[Jeff Prince]

29 minutes ago, line-weight said:

I get the basic idea of value - just meant that the numbers in the table don't seem to make sense.

Agreed.  Look at the two iMacs, the “value” calc would indicate larger is better in that instance.

[bcd]

fascinating post Don, no interest here on the community at all!  😉

 

I'm seeing:

14:14 on MBP M1 Max 64GB

13:04 on  Alienware x15 R2 i9 32GB RTX 3080Ti 8GB $2700

 

I agree the easy part is comparing price/speed.

Deciding value is another story - but I do think 1000 renders/price  (t/$) isn't quite it.

 

[Christiaan]

One thing that's very difficult to factor in is RAM. 128 GB RAM may have added $1100 to the cost of the 2019 iMac i9, but it also helped keep Vectorworks running the whole day, rendering SLVPs.

[Pat Stanford]

6 hours ago, line-weight said:

M1 Macbook Pro (0.0051) and M1 Mac Mini (0.0110) take roughly the same time for the render, with M1 Mac Mini being half the price. This suggests higher number = better.

The price of the MBP includes the keyboard and screen while the Mini is only the base box.

 

The value calculation determines the number of renders you can finish in one day as the numerator. So the Numerator increases with the speed of the machine. The demoninator is the price of the machine. So the denominator increases with the price of the machine. Which is normally related to the speed of the machine. Maybe I need to break out laptops separate from stationary machines so we don't get confused by the "extras" included in the laptop price.

[line-weight]

7 hours ago, line-weight said:

I don't understand the "value" figure.

 

For example M2 Max macbook pro gets 0.0016 and i9 iMac gets 0.0028. They both cost about the same but M2 Max is loads faster. This suggests lower number = better.

 

But ...

 

M1 Macbook Pro (0.0051) and M1 Mac Mini (0.0110) take roughly the same time for the render, with M1 Mac Mini being half the price. This suggests higher number = better.

 

Above^^^ in bold is what doesn't make sense. Those two conflicting observations.

 

If numerator = speed and denominator = price, then a high "value rating" number = better.

 

If a high number = better, then the M2 Max MBP should be getting a higher value rating than the i9 iMac. But it doesn't in the table. I think a calculation is going wrong somewhere.

[Pat Stanford]

I agree that the Value calculation does not work well. Can you suggest a different option? I have tried $/Render, and Renders/$ both give misleading information.

 

Here is a version of the table that just shows the number of renders per day for each system instead of the Value.

 

[zoomer]

3 hours ago, Christiaan said:

128 GB RAM may have added $1100 to the cost of the 2019 iMac i9

 

But who would buy RAM from Apple,

especially for on of their last last (and easy) user upgradeable offers.

 

Around the time I build my PC, I did a similar table.

Also mainly for CPU Rendering only, R21, later R23, for iMac Pros, Mac Pros,

Ryzens, my Trash Can.

And abandoned after I added my Mini (Native and Rossetta2)

 

Columns were :

- number of cores

- estimated CPU prices

+ Cinebench single core (CAD)

+ Cinebench multi core (Rendering)

> Cinebench Points (multi core) per Core (Scaling)

> Cinebench Points (multi core) per $ (Value)

 

Aim to do one again for M1s, M2s and Ryzen again soon.

[Christiaan]

That 128 GB RAM was purchased from MacSales.com at the time.

[zoomer]

Basically we could also just use our Cinebench Results for this task.

(RW Rendering = C4D Rendering and single core points should resemble

overall App speed)

I think Cinebench's clock is trustworthy out of the box..

 

What Cinebench does not (opposed to 3D Games) is shifting much memory.

So there is not much advantage for 3800 GHz RAM vs my slow and outdated

2400 MHz RAM, although a Ryzen in real world would profit very much from

highspeed RAM.

So a bit unfair for not using M Max and M Ultras high memory bandwidth.

 

But none of all Apple ARM SoCs looks like it was build for any old school

CPU Rendering anyway.

And that Apple ARM works great with VW overall, you can already see by

our weak device "M1 Pure" performance.

Edited February 1, 2023 by zoomer