Best CPU for Mac Pro 6.1

[Zen_Arcade]

I have upgraded several 6,1 to the E5-2697 v2 12 core CPU and none of them ever had an issue with fan control, wifi, bluetooth, or anything else.

I second flygbuss’s suggestion to check the fan ribbon cable.

 

[tpivette89]

Hi All

I have just found this thread through google as I have been trying to see if I have the best processor that I can get or if there is anything that I can upgrade to. I currently have the 2690v2 with 3745 Single-Core score and 28285 Multi-Core.

Judging by this thread the 2697 V2 will be a slight improvement with around 32226 Multi. Would this be stable with fan control?

Im afraid this is all a bit of a new area for me, so would I notice any difference if i were to upgrade? I mainly use the computer for Photoshop (albeit quite complex files) and illustrator. I do find that it slows down as I can end up with lots of files open at once.
Could anybody advise on the difference of upgrading to 128gb RAM from 64?

Cheers

Yes, the E5-2697 V2 will work just fine in a 6,1. It was the top tier CPU Apple installed from the factory, so there should be no issues, especially with fan speed altering software.

Whether or not you will notice a "difference" depends on your usage. While running your workload, open Activity Monitor and double click on the CPU Load section at the bottom of the window for a display of all cores/threads and their current usage. If things are maxed out or close, then you may benefit from 2 more cores. Since I'm not intimately familiar with Photoshop, I can't say for sure, however.

I have heard that you take a speed hit (stock being 1866mhz) when going from 64GB to 128GB. Again, though, I have no way of confirming this as I do not have a 6,1. If you need to know if 64GB is limiting your work, again, try opening Activity Monitor during a Photoshop session and see how much RAM is being utilized.

 

[crystalidea]

I installed 2667v2. solved.

 

[crystalidea]

solved, all good.

 

[joncresci]

Sorry to resurrect this thread. But I just installed a 2697 V2 that I got from ebay. Mac booted but no USB ports and Bluetooth was really flaky, cycling between available and unavailable rapidly. Reseated the fan assembly cable the same thing. Reinstalled the old 6core chip and all is working.

Bad chip? or bad install... I am trying to decide if I should try again with another vendor. Am I missing something about the install? I am a protools user so i was hoping to get the multicore performance. I am on OS10.13.6. Any help would be great.

 

[crystalidea]

Most likely a bad install, can't imagine that USB doesn't work because of 'bad' cpu.

 

[joncresci]

Very good point. I couldn't understand it. I'll try it again tomorrow after I get some work done and see if I can replicate it.

Should I be doing a fresh install of MacOS? Can't imagine that would be the issue either.

 

[crystalidea]

Because of my bad install, the fan was not spinning up, so take care.

Should I be doing a fresh install of MacOS?

No.

 

[joncresci]

Got it. Thanks!

 

[MisterAndrew]

Why is the E5-2667 V2 preferred over the E5-1680 V2? The benchmarks show the E5-1680 V2 is faster. It also has a higher max tcase (85°C vs. 74°C) so it seems it would be less sensitive in the thermally challenged trashcan. And then what about the 110 watt E5-2673 V2? Any benefits in terms of thermals on that one? Intel doesn't list the max tcase for that one. Perhaps it's just best to stick to the E5-1680 V2 as Apple originally installed?

Intel product specifications

Intel® product specifications, features and compatibility quick reference guide and code name decoder. Compare products including processors, desktop boards, server products and networking products.

ark.intel.com

Geekbench results for the Intel Xeon E5-1680 V2: 870 single-core, 6659 multi-core

Geekbench results for the Intel Xeon E5-2667 V2: 751 single-core, 6268 multi-core

Average of the only 2 Mac Pro 6,1 Geekbench results for the E5-2673 V2: 816.5 single-core, 5948 multi-core (https://browser.geekbench.com/v5/cpu/899803 & https://browser.geekbench.com/v5/cpu/1079402)

The Userbenchmark site shows the E5-2667 V2 to be slowest out of the 3:

Userbenchmark for E5-1680 V2: 76.5% (87th of 1221) -> 22% faster than base 4 core E5-1620 V2

Userbenchmark for E5-2667 V2: 67.7% (153rd of 1221) -> 8% faster than base 4 core E5-1620 V2

Userbenchmark for E5-2673 V2: 69.3% (137th of 1221) -> 10% faster than base 4 core E5-1620 V2

 

[crystalidea]

Don't believe those GeekBench numbers. I consistently get these results with 2667v2, see my last result:

MacPro6,1 - Geekbench

Benchmark results for a MacPro6,1 with an Intel Xeon E5-2667 v2 processor.

browser.geekbench.com

It must be compared to:

Mac Pro (Late 2013) Benchmarks - Geekbench

2667v2 has 300Mhz higher base frequency and 100Mhz higher turbo frequency.

 

[MisterAndrew]

Okay, thanks. It looks like non-Mac Pro builds are skewing the results in the benchmarks I linked above, maybe due to overclocking. For example, when I look at the E5-1680 V2 in Mac Pro builds only on the userbenchmark site the benchmarks are much lower. 70% and lower.

UserBenchmark: Apple MacPro6,1 Compatible Components

[automerge]1587838606[/automerge]
It looks like I'll be going with the E5-2667 V2 then. The 12 core benchmarks sure don't look good to me. I don't see any reason to go with the E5-2697 V2 unless I really needed the extra cores. Multi-core is almost just as good with the 8 core and much higher on single-core.

 

[MisterAndrew]

Your E5-2667 V2 benchmark isn't actually super far off from the 8 core in the new Mac Pro. 902 vs 1032 for single-core and 6515 vs 8046 for multi-core. I mean, the new one is better, but not 3-4x better value IMO.

MacPro7,1 8 Core Geekbench

 

[tpivette89]

I have a E5-2667 V2 swapped Mac Pro (6,1) with 32GB of RAM... Geekbench scores as follows:

MacPro6,1 - Geekbench

Benchmark results for a MacPro6,1 with an Intel Xeon E5-2667 v2 processor.

browser.geekbench.com

Would like to test with the best 10-core (E5-2690 V2) and the best 12-core (E5-2697 V2), but at the moment, I do not have access to either of these processors. Also, would love to test with 64GB of RAM, but again, I don't have access to the available hardware.

Anyone???

 

[MisterAndrew]

Since that 12-core is the standard one from Apple it should be easy to find benchmarks. The 10-core would be in-between that one and the 8-core. After that 8-core single-core performance decreases and multi-core performance increases.

 

[crystalidea]

12 Core scores 7k in multi-core while 2667-v2 6.5k

Mac Pro (Late 2013) Benchmarks - Geekbench

 

[tpivette89]

Sometimes it's hard to discern Trashcan scores from the Hackintosh ones. Would like to have actual results from forum members.

I have the 2667 CPU and haven't come close to 6500 in multi-core scores in Geekbench 5 (I have 32gb of RAM). Mine are in the 6200 range. So something in the 7000+ area would be a decent improvement for a 12-core.

The cost for 64gb of RAM (4x16gb) is about the same, if not a little more, than the 10-core CPU. Which one would be the better upgrade?

Being that I'm bored and have money to burn, I will be ordering a 2690 10-core CPU and testing it vs the 2667 that's currently in my machine. Will determine if the increase in multi-core scores is worth the trade-off in single-core performance.

 

[crystalidea]

@tpivette89 maybe your silicon is a little worse and/or you have less RAM and/or I've got faster NVMe SSD and/or eGPU eats some CPU bits.. I don't see much sense in 8 -> 10 upgrade, because you'd loose single core scores and doubtfully get decent multi-core scores increase.
Single core score is more important in daily tasks while multi makes rendering/encoding faster. If you really need the last one, 12 is the way to go. 8-core is the most balanced choice I'd say.

 

[MisterAndrew]

I've been looking over the Intel data sheets for these processors and it looks like the E5-2667 v2 is indeed the highest performing processor in the lineup. However, that comes with a major downside. It has the very worst thermal properties. It requires more airflow than the other chips. So the fans should be kept at a higher speed with that processor installed. That's not an ideal scenario for those who value silent operation. And due to the thermal constraints of this Mac I think it's best avoided. It looks like the 10 core E5-2690 v2 would be fine. It appears to have slightly better thermal properties than the stock 12 core. Intel lists the same thermal properties for the E5-1650 v2 and E5-1660 v2 so the latter may be a good upgrade for those who want the best single thread performance. For 8 core I think it's best to stick with the E5-1680 v2.

 

[tsialex]

I've been looking over the Intel data sheets for these processors and it looks like the E5-2667 v2 is indeed the highest performing processor in the lineup. However, that comes with a major downside. It has the very worst thermal properties. It requires more airflow than the other chips. So the fans should be kept at a higher speed with that processor installed. That's not an ideal scenario for those who value silent operation. And due to the thermal constraints of this Mac I think it's best avoided. It looks like the 10 core E5-2690 v2 would be fine. It appears to have slightly better thermal properties than the stock 12 core. Intel lists the same thermal properties for the E5-1650 v2 and E5-1660 v2 so the latter may be a good upgrade for those who want the best single thread performance. For 8 core I think it's best to stick with the E5-1680 v2.

What you found on the Intel documentation that got you scared of E5-2667 V2?

While I don't recommend anyone to consider installing the E5-2697W V2 since it's a 150W TDP Xeon, with the E5-2667 V2 you have to dissipate the same from the Apple stock 12 core, E5-2697 V2.

 

[crystalidea]

@tsialex is right, the 2667-v2 has the same 130W TDP spec as every official CPU installed in MacPro6,1. I do confirm the fan works absolutely fine (actually in favor of silence than thermals as with many other macs, but this can be solved with the app).

 

[MisterAndrew]

Well looking at the thermal charts, it appears it's preferable to have higher processor-to-ambient thermal resistance and case-to-local ambient thermal characterization parameter values. The processor-to-ambient thermal resistance for the E5-2667 v2 (along with the E5-2643 v2) is the worst at 0.186. It's 0.242 for the E5-1680 v2. The other chart shows it requires 26 CFM of airflow, while the E5-1680 v2 requires 21.8. Of course it shows the heatsink is larger for the workstation CPU, but it shows the case-to-local ambient thermal characterization parameter value is lower for the E5-2667 v2.

 

[crystalidea]

This can in theory give +1C, but who cares?

 

[MisterAndrew]

This can in theory give +1C, but who cares?

How were you able to calculate 1 °C?

 

[crystalidea]

Unless one can prove by doing extensive testing (both 2667v2 & 1680 v2 in the same MacPro6,1) that temperature actually differs, it's not even 1° C but 0.1° C for me: 2667-v2 is ice cold in my MacPro.

Edit: if interested, I can stress him out, just propose a scenario.