Memory upgrade troubleshooting in 2020 27-inch iMacs

[MacKarmaTech]

Really great and highly informative thread. Thanks to all that contributed to furthering our overall knowledge of this critical and very particular memory upgrade specifics of the 27" 2020 iMac's

I kinda believe that Apple just gave an at somewhat obsure, but increasingly apparent (to memory uprade customers) middle digit to everyone who purchased one of the last ever "Intel" CPU iMacs - by only including an embarrassingly minimal 'base' 8gb (2x4gb's) of memory in all iMac configurations. Since this one of the very few Apple manufactured computers that even allow any of it's internal components to be "user upgradable"...they probably figured (correctly) that if anyone wanted to purchase them - especially if as a BTO, top-of-the-line spec'ed i9 iMac model....that either the consumer would go 'all in' and willingly succumbed to Apple's infamous price gouging for these (last ever possible) iMac memory upgrades and ordered ONLY from Apple themselves....or.... the reality that they would eventually have to deal with a thread like this one explaining why they should have paid the piper (Apple) in the first place. By Apple's caluculations: you either buy your preconfigured 27" 2020 iMac with 8 gb's (4x4gb's) of useless, insultingly inadequate memory, or buy a properly (BTO) upgraded 64 or 128gb memory ) via (AAPL) for many hundred's of $$$ more...AND..It's all pure gravy to them

Next up...The everything soldiered (SOC) M something iMac where any though of discretionary components or upgrades becomes all but just a distant "memory".

 

[mikehalloran]

Next up...The everything soldiered (SOC) M something iMac where any though of discretionary components or upgrades becomes all but just a distant "memory".

Too many active threads about that now — no need to continue this one into that territory.

In any case, time to stick a fork in it — the original topic is done, answered and scrutinized to death. Anyone who doesn't understand should just reread it from the beginning.

Putting this thread into the Ignore status for myself. Life's too short to care about any more responses.

 

[eXistenZ78]

Hi everyone
First post here, I read a lot of pages on this thread but not everything, sorry if I'm about to ask something already explained previously but I didn't found a clear answer.

My 2020 5K iMac comes with 2x4Gb modules.
I bought 2x16Gb Crucial modules CT2K16G4S266M.
As everyone else in this tread noticed, if all 4 ram modules are installed they work at 2133Mhz.
So I decided to uninstall the 4Gb modules, keeping just the 2 Crucial.
I'm a graphic designer so I mainly work with Indesign, Photoshop, Illustrator, and Lightroom, no video editing, no 3D rendering. So I think 32Gb is enough I think.

Since I'm still in time to send the 2x16Gb modules back to amazon, my question is:
There's a difference between a 16-0-16-0 configuration and an 8-8-8-8 configuration?

If there's no difference I keep the 2x16 modules, but if there's a performance improvement I can make the change and put 4x8Gb (CT2K8G4S266M) on for the same money.

Thank you

PS: I bought 2666Mhz modules based on Apple's specs, but searching on Geekbenck I found several iMacs 2020 with 3600Mhz ram on them. https://browser.geekbench.com/v5/cpu/9813813
How it's even possible?

 

[PBG4 Dude]

Keep the 16-0-16-0 in case you want to add more later. No reason to go to 8-8-8-8 as the iMac only has a dual channel setup.

 

[CWallace]

PS: I bought 2666Mhz modules based on Apple's specs, but searching on Geekbenck I found several iMacs 2020 with 3600Mhz ram on them. https://browser.geekbench.com/v5/cpu/9813813
How it's even possible?

You can put faster modules in, but they will run at whatever maximum speed the chipset supports.

The Intel chipset maxes out at DDR4-2933 per their specsheet and that is probably what it is actually running at in an iMac as someone posted their benchmarks on the Amazon page for the Corsair Vengeance DDR4-3600 modules and they noted it was a bit better than stock Apple RAM.

 

[wilberforce]

Hi everyone
First post here, I read a lot of pages on this thread but not everything, sorry if I'm about to ask something already explained previously but I didn't found a clear answer.

My 2020 5K iMac comes with 2x4Gb modules.
I bought 2x16Gb Crucial modules CT2K16G4S266M.
As everyone else in this tread noticed, if all 4 ram modules are installed they work at 2133Mhz.
So I decided to uninstall the 4Gb modules, keeping just the 2 Crucial.
I'm a graphic designer so I mainly work with Indesign, Photoshop, Illustrator, and Lightroom, no video editing, no 3D rendering. So I think 32Gb is enough I think.

Since I'm still in time to send the 2x16Gb modules back to amazon, my question is:
There's a difference between a 16-0-16-0 configuration and an 8-8-8-8 configuration?

If there's no difference I keep the 2x16 modules, but if there's a performance improvement I can make the change and put 4x8Gb (CT2K8G4S266M) on for the same money.

Thank you

PS: I bought 2666Mhz modules based on Apple's specs, but searching on Geekbenck I found several iMacs 2020 with 3600Mhz ram on them. https://browser.geekbench.com/v5/cpu/9813813
How it's even possible?

Essentially no performance difference between 2x16GB and 4x8GB. Best to keep what you have and you have the option to add another 2x16GB later. I have the exact same RAM as you.

The geekbench score you linked is for a Hackintosh (Fake Macintosh). The motherboard is identified as Acidanthera, which is a dead giveaway. Also, when you see results for several machines that look very similar, they are often actually a single machine being tested multiple times by one person.

You are welcome to ask as many questions as you like, whether or not they have been asked before. I for one think that is perfectly OK.

 

[T4R06]

I think Big Sur is the issue. I just got my 27" 2020 i7 with 8gb and instantly upgraded the ram to 32gb single module. then I added the 2 4gb supplied. so, in short its mixed.

like the other complain, yes my score went dropped. 8gb is performing much better vs 40gb. but now I upgraded my OS to Monterey the 40gb now is more higher that the average. see screenshots below. the RAM I used was the TEAMGROUP Elite DDR4 32gb single for $113

 

[minik]

I think Big Sur is the issue. I just got my 27" 2020 i7 with 8gb and instantly upgraded the ram to 32gb single module. then I added the 2 4gb supplied. so, in short its mixed.

like the other complain, yes my score went dropped. 8gb is performing much better vs 40gb. but now I upgraded my OS to Monterey the 40gb now is more higher that the average. see screenshots below. the RAM I used was the TEAMGROUP Elite DDR4 32gb single for $113

View attachment 1872675

I ran Geekbench 6-7 months ago with 2x16GB under macOS 11.2.3, the score was 1289 single-core and 8958 multi-core. Later one with 64GB memory from Apple, the multi-core score was slightly lower. They were all Core i7.

 

[Fearless_Felix]

I bought my 27’ iMac (2020) last year and upgraded the RAM to 64GB straight away. The RAM was bought from Crucial, a 64GB kit (2 x 32GB) DDR4-2666 SODIMM, product number CT19090527. It has been working perfectly since then.

However, yesterday I decided to try putting back the original 2 x 4GB RAM modules to increase the total RAM to 72GB. The two 4GB modules were inserted into slots 1 & 2 and the two 32GB modules into slots 3 & 4. System Report showed 72 GB RAM installed at 2667 MHz, all seemed fine.

The before & after Novabench scores were much the same, however Geekbench & PugetBench Photoshop Extension showed a dramatic loss of performance.


Geekbench (version 5.4.3) showed the CPU Multi-Core speed dropping from 10187 to 7930. Things were just as bad with PugetBench, the score dropping from 1023 to 874. I removed the two 4GB RAM modules & returned the 2 x 32GB to slots 2 & 4. The speeds are now back to their former values.

I’m running Mac OS Monterey 12.01.

 

[wilberforce]

I bought my 27’ iMac (2020) last year and upgraded the RAM to 64GB straight away. The RAM was bought from Crucial, a 64GB kit (2 x 32GB) DDR4-2666 SODIMM, product number CT19090527. It has been working perfectly since then.

However, yesterday I decided to try putting back the original 2 x 4GB RAM modules to increase the total RAM to 72GB. The two 4GB modules were inserted into slots 1 & 2 and the two 32GB modules into slots 3 & 4. System Report showed 72 GB RAM installed at 2667 MHz, all seemed fine.

The before & after Novabench scores were much the same, however Geekbench & PugetBench Photoshop Extension showed a dramatic loss of performance.


Geekbench (version 5.4.3) showed the CPU Multi-Core speed dropping from 10187 to 7930. Things were just as bad with PugetBench, the score dropping from 1023 to 874. I removed the two 4GB RAM modules & returned the 2 x 32GB to slots 2 & 4. The speeds are now back to their former values.

I’m running Mac OS Monterey 12.01. View attachment 1920564 View attachment 1920565

Yes, because you had 8GB in channel A (slots 1+2), and 64GB in Channel B (slots 3+4). This means that 56GB of your 72GB RAM was only running in single channel mode, which halves its effective transfer speed.
To get full dual channel operation, always have equal amount of RAM in Channel A as in Channel B (which is what you originally had, with 32GB in slot 2 and 32GB in slot 4).

 

[Fearless_Felix]

Thanks for your explanation.

 

[eftaefta]

I have noticed this problem on mine too.

I have always upgraded the RAM myself on iMacs, and my 2019 model has two 4GB Apple sticks and two 16GB Apple sticks, and registers 40GB at 2667 MHz without an issue.

My 2020 model is stuck at 2133 MHz.

This is despite them both using Crucial RAM of the same specs, in addition to the Apple RAM.

I've played around with the sticks and these are my findings:

The 2020 iMac registers 2133 MHz using the same RAM sticks that the 2019 registers at 2667 MHz.

I took the sticks out of the 2019 model and placed them in the 2020 model, in the correct formation, and they reported 2133 MHz. So the 2020 model is more particular about RAM.

The 2020 iMac has a slightly different RAM arrangement:

View attachment 942848
2020 iMac (above)


View attachment 942855
2019 iMac (above)
​

This change is probably tied to the implementation of the T2 controller?

Putting all four 16GB Crucial sticks in the 2020 model still registered 2133 MHz.

This surprised me as they are all the same brand and spec. So I rearranged the sticks to what to me seems to be the 'wrong' way - my inherited 2019 Crucial sticks in Slots 1 and 2 (rather than 1 and 3 or 2 and 4 as you would normally expect), and the newly-bought Crucial sticks in Slots 3 and 4, and low and behold it is now running at 2667 MHz and 64GB.

The 2020 iMac is fussy even between different batches of the same RAM manufacturer and specs, not just between Hynix and Crucial. Get around this by keeping each batch/manufacturer to its own channel.

I am not sure what (if any) performance hit will come by mixing DIMMs in this way. I have always been told to install them in pairs in the first and third, or second and fourth, slots.

In the 2020 iMac (and only the 2020 iMac), the requirements for using 4 DIMMs are:

1. part numbers in slots 1 and 2 must match, else speed drops to 2133
2. part numbers in slots 3 and 4 must match, else speed drops to 2133
3. Total size (GB) in Channel A (slots 1+2) must equal total size in Channel B (slots 3+4), else it will not run in full dual channel. (The requirement for dual channel has nothing to do with matching part numbers).

 

[eftaefta]

I bought my 27’ iMac (2020) last year and upgraded the RAM to 64GB straight away. The RAM was bought from Crucial, a 64GB kit (2 x 32GB) DDR4-2666 SODIMM, product number CT19090527. It has been working perfectly since then.

However, yesterday I decided to try putting back the original 2 x 4GB RAM modules to increase the total RAM to 72GB. The two 4GB modules were inserted into slots 1 & 2 and the two 32GB modules into slots 3 & 4. System Report showed 72 GB RAM installed at 2667 MHz, all seemed fine.

The before & after Novabench scores were much the same, however Geekbench & PugetBench Photoshop Extension showed a dramatic loss of performance.


Geekbench (version 5.4.3) showed the CPU Multi-Core speed dropping from 10187 to 7930. Things were just as bad with PugetBench, the score dropping from 1023 to 874. I removed the two 4GB RAM modules & returned the 2 x 32GB to slots 2 & 4. The speeds are now back to their former values.

I’m running Mac OS Monterey 12.01. View attachment 1920564 View attachment 1920565

In the 2020 iMac (and only the 2020 iMac), the requirements for using 4 DIMMs are:

1. part numbers in slots 1 and 2 must match, else speed drops to 2133
2. part numbers in slots 3 and 4 must match, else speed drops to 2133
3. Total size (GB) in Channel A (slots 1+2) must equal total size in Channel B (slots 3+4), else it will not run in full dual channel. (The requirement for dual channel has nothing to do with matching part numbers).

 

[jordii]

Different variation on this theme.

Per this thread, I added 2x16GB Crucial RAM modules (DDR4 2666MHz, CT2K16G4S266M) to my iMac20,1 in slots 2 and 4, and discarded Apple's stock 2x4GB modules.

"About this Mac" shows "32 GB 2133 MHz DDR4" and below is a screenshot from system report. The module labels do specify 2666.

 

[jordii]

Resolved.

Swapping in 2x16g OWC memory brought me to 2667MHz

 

[Xylopyrographer]

Adding to the info pool...

Upgraded a 2020 iMac 27" i7 from 64GB to 128GB.

In the 64GB configuration, slots 1 and 3 each had a 32GB stick from the same mfg with the same P/N (call this mfg A). Total memory reported as 64GB at a speed of 2667 MHz.

Installed the two new 32GB sticks from a different mfg with their same P/N (call this mfg B) into slots 2 and 4. Total memory reported as 128GB at a speed of 2133 MHz.

Removed the sticks from mfg A and moved the sticks from mfg B into slots 1 and 3. Total memory reported as 64GB at a speed of 2667 MHz.

Reconfigured with mfg A in slots 1 and 2, mfg B in slots 3 and 4. Total memory reported as 128GB at a speed of 2400 MHz.

Didn't try mfg B in slots 1 and 2, mfg A in slots 3 and 4.

Quite the game of whack-a-mole.