MP 6,1 E5-1680v2 vs E5-2667v2 vs E5-2690v2

[UbiCrea]

Hey freinds...
Still upgrading slowly my Can.
I have three CPUs, two 8 cores (1680v2 & 2667v2) and just got a 10 cores 2690v2.
The 10c runs at 3ghz like the 1680 but the top speed is 3,6ghz instead of the 3,9 for the 1680. I wonder is there is a big difference? Which tasks would really benefit of a small 300mhz speed bump?
Thanks

 

[arw]

2667v2 vs. 2690v2: 300 MHz more base clock is still +10%. And up to 11% more turbo for single core tasks.
In general, a higher single core speed also makes the overall system feel more snappy (opening apps, etc).
But it highly depends on your apps if they make proper use of all cores/threads. Then 10 cores > 8 cores.

For my usage, I found most (CPU-based) Photoshop filters to top out at 8 parallel threads but to scale (more or less) linearly with the clock speed.
That's why I went back to said 2667v2 from a 2697v2 (12-core, 2.7 GHz base, 3.5 GHz turbo).
And I disabled Hyperthreading/SMT:
https://support.apple.com/en-gb/HT210108

 

[UbiCrea]

2667v2 vs. 2690v2: 300 MHz more base clock is still +10%. And up to 11% more turbo for single core tasks.
In general, a higher single core speed also makes the overall system feel more snappy (opening apps, etc).
But it highly depends on your apps if they make proper use of all cores/threads. Then 10 cores > 8 cores.

For my usage, I found most (CPU-based) Photoshop filters to top out at 8 parallel threads but to scale (more or less) linearly with the clock speed.
That's why I went back to said 2667v2 from a 2697v2 (12-core, 2.7 GHz base, 3.5 GHz turbo).
And I disabled Hyperthreading/SMT:
https://support.apple.com/en-gb/HT210108

Mmm, ok thanks.
I'm currious about disabling HT... on the link you gavem Apple says that there was a decrease of 40% in perf when disabled... what's your experience ?

 

[mikas]

I don't think the differences are that much if comparing to todays standards inside CPU generations.
If you plan to use 32GB memory modules, they would work faster with 16xx than they would with 26xx CPUs.
Here's one example of performance differences between 6,1 compatible CPUs: E5-2673 v2 is quite rare, but the thermals seems like exceptionally good. I haven't tested it, but it should be compatible according to conversations I have read.

 

[arw]

I'm currious about disabling HT... on the link you gavem Apple says that there was a decrease of 40% in perf when disabled... what's your experience ?

That’s only in regard to multi-core performance. Which is obvious as you go from 8 cores/16 threads to 8 cores/8 threads.
From my testing with disabling HT on my 12-core 2697v2:
- Geekbench 5 multi-core score: drop from 7244 to 6624 (so -9%)
- Cinebench R23 multi-core score: drop from 8119 to 6436 (so -21%)
But those remaining threads clock slightly higher with disabled HT so that applications that do not make use of more threads (than the available cores) finish tasks slightly faster.
We’re still only talking less than 10% faster (my real world measurements for Photoshop processing).
Which, as @mikas said, is peanuts in comparison to recent platforms - But it is a way to squeeze the absolute maximum out of our beloved 6,1s if your workflow allows it.
For thermal/power reasons I had also preferred a 2673v2 over the 2667v2 (110W vs. 130W) but they are hard to come by and relatively pricey.

EDIT: With activity monitor you can check the CPU usage during one of your main tasks before making any decision.
Attached Photoshop using ~800% CPU and Topaz using ~1700% CPU (HT enabled).

 

[UbiCrea]

Ok ok good thanks.
I'm using VMWare Fusion 13 with Windows 10 and it's super laggy with CPU at 100% many times. I've tried with 4 CPU, 8, 16, 2... and each time, it looks weird.
I was using VMWare on a PC before and it was super fluid... Any idea on how to solve this ?
Maybe it's using the HT as a CPU?