Current state of cMP Benchmarking

[usna92]

I am looking to start upgrading my cMP and want to do benchmarks at each upgrade to measure the effectiveness of what the hell I am doing. What are the common benchmarks that people are using. Right now I have GeekBench 4, Cinemark 4D, and BlackMagic. Are there others that I should be using instead?

 

[Synchro3]

Geekbench 2 and Geekbench 3.

 

[h9826790]

 

[usna92]

That looks like it covers just about everything. Are there any of those more reliable for general benchmarking than others? Give more consistent results, or is the consensus run them all, and see what comes?

 

[Fl0r!an]

It depends on what you want to benchmark.

Cinebench for example is heavily CPU limited, making it unusable as GPU benchmark. The same applies to Valley & Heaven when running with very low graphics settings.

 

[orph]

cinbench for cpu only (not gpu) https://www.maxon.net/en/products/cinebench/
OctaneBench for cuda https://render.otoy.com/octanebench/

but best benches are the apps that you actually use

there's no way to keep them spot on as there's lots of background stuff in osx + your background stuff + random changes, so anything in say 5-10% may just be random variation

 

[usna92]

I was looking for a basic overall system performance type of benchmark. I am using the graphic card I had sitting on a shelf, so I am not in the higher end graphic card use case yet. I am waiting to see how the RX world shakes out. Right now i have the R9-280X in there which probably meets my expected use cases for now, but might consider a newer card should prices and opportunities afford it.

 

[orph]

i dont think there is one benchmark that dose everything because there's to much, best to focus on what you do.

ie
if your ripping DVD's then handbrake is the best bench (if your using handbrake)
if your doing adobe PP then PP is the best
and so on

 

[usna92]

i dont think there is one benchmark that dose everything because there's to much, best to focus on what you do.

ie
if your ripping DVD's then handbrake is the best bench (if your using handbrake)
if your doing adobe PP then PP is the best
and so on

Fair enough. I will continue to run the numbers I have been running to keep some sort of consistent measurement as I make changes and then do a run of most of them when I am done tinkering for a while. That way I can get to measuring real world performance and look at specific use case improvements. The replies have been great. Thanks.

 

[SoyCapitanSoyCapitan]

Benchmarking upgraded cMP was fun 2 years ago. Now they're too crusty to think about benchmarking.

 

[usna92]

Benchmarking upgraded cMP was fun 2 years ago. Now they're too crusty to think about benchmarking.

That is probably true, for me it is just a way to measure relative improvements as I change things to determine if I am helping or hurting when I improve (fiddle with) something.

 

[AidenShaw]

Benchmarking upgraded cMP was fun 2 years ago. Now they're too crusty to think about benchmarking.

Agree. The cMP was an interesting alternative to the MP6,1 in early 2014.

Now, however, look at the posts here and see the number of hardware failures and "the old hardware doesn't work with the new software" issues. The cMP is becoming marginalized.

If I were considering spending money today on upgrading a cMP - I would probably drop the idea and consider

• an MP6,1 (horrible price performance, but good support (except for all the GPU-related and WindowServer crashes related to the ATI Dx00 GPUs))
• a Hackintosh (good price performance, but issues around upgrades and hardware support)
• Windows 10 (good price performance, good support, but mostly religious issues about running Windows)

 

[usna92]

I agree completely with regard to price/budget performance issues given time, etc. Eventually, my cMP will be finally EOL in the operating system (since it is already on the ragged edge of most of the newer features anway). Right now it does just what I need it to do at home, and with what I had on the shelf at home, could be customized at minimal expense to upgrade my home infrastructure. Were I looking to make a new or newish personal purchase, I would have to lean to iMac with some sort of massive external doc, given the age of the "n"MP in the market already. Moving away from PC Reference graphics cards in a pro machine was a mistake for Apple, and one they are going to need to correct if they decide this market is one they want to stay in. For me, right now, this is an experiment that is fun, allows me to tinker without exploding something really expensive if I screw it up, and is better than running my rMBP in clamshell mode as my home server which is what I have been doing.