Memory upgrade compatibility list for iMac 2019-2020

[wilberforce]

So just to add my findings with different combination of RAM in my 2020 iMac which arrived today:

2x4GB Apple RAM (mfr Micron!) (total 8GB): 2667 MHz, Geekbench 7060
2x4GB Apple RAM + 2x16GB Crucial RAM (Total 40GB): 2133 MHz, Geekbench 8220
2x16GB Crucial RAM + 2x16GB Timetec (Hynix) RAM (Total 64GB): 2133 MHz, Geekbench 8240
2x16GB Cruclal RAM (Total 32GB): 2667 MHz, Geekbench 8680
2x16GB Timetec (Hynix) RAM (Total 32GB): 2667MHz, Geekbench 8570

So no surprise, mixing RAM brands or sizes drops the speed to 2133 MHz and degrades the Geekbench score slightly.
Above scores are for the i7 processor, with activity monitor running. Don't read too much into slight variations in scores. (For example 32 GB Timetec scoring slightly lower than 32 GB Crucial - I only ran it once).

So I am using the 32GB Crucial RAM, returning the Timetec RAM, and putting the Apple RAM aside. I hardly ever use more than 20GB.

Note that the 2x16GB Crucial and 2x16GB Timetec have the exact same specs, same size, both dual ranked, both "made for MAC," so should theoretically play nice together, but no, when combined the speed and performance drop. Something to think about if you are considering adding some RAM now, and more RAM later.

 

[pldelisle]

So just to add my findings with different combination of RAM in my 2020 iMac which arrived today:

2x4GB Apple RAM (mfr Micron!) (total 8GB): 2667 MHz, Geekbench 7060
2x4GB Apple RAM + 2x16GB Crucial RAM (Total 40GB): 2133 MHz, Geekbench 8220
2x16GB Crucial RAM + 2x16GB Timetec (Hynix) RAM (Total 64GB): 2133 MHz, Geekbench 8240
2x16GB Cruclal RAM (Total 32GB): 2667 MHz, Geekbench 8680
2x16GB Timetec (Hynix) RAM (Total 32GB): 2667MHz, Geekbench 8570

So no surprise, mixing RAM brands or sizes drops the speed to 2133 MHz and degrades the Geekbench score slightly.
Above scores are for the i7 processor, with activity monitor running. Don't read too much into slight variations in scores. (For example 32 GB Timetec scoring slightly lower than 32 GB Crucial - I only ran it once).

So I am using the 32GB Crucial RAM, returning the Timetec RAM, and putting the Apple RAM aside. I hardly ever use more than 20GB.

Note that the 2x16GB Crucial and 2x16GB Timetec have the exact same specs, same size, both dual ranked, both "made for MAC," so should theoretically play nice together, but no, when combined the speed and performance drop. Something to think about if you are considering adding some RAM now, and more RAM later.

Thanks a lot for this analysis !!

 

[Mikeeee]

So just to add my findings with different combination of RAM in my 2020 iMac which arrived today:

2x4GB Apple RAM (mfr Micron!) (total 8GB): 2667 MHz, Geekbench 7060
2x4GB Apple RAM + 2x16GB Crucial RAM (Total 40GB): 2133 MHz, Geekbench 8220
2x16GB Crucial RAM + 2x16GB Timetec (Hynix) RAM (Total 64GB): 2133 MHz, Geekbench 8240
2x16GB Cruclal RAM (Total 32GB): 2667 MHz, Geekbench 8680
2x16GB Timetec (Hynix) RAM (Total 32GB): 2667MHz, Geekbench 8570

So no surprise, mixing RAM brands or sizes drops the speed to 2133 MHz and degrades the Geekbench score slightly.
Above scores are for the i7 processor, with activity monitor running. Don't read too much into slight variations in scores. (For example 32 GB Timetec scoring slightly lower than 32 GB Crucial - I only ran it once).

So I am using the 32GB Crucial RAM, returning the Timetec RAM, and putting the Apple RAM aside. I hardly ever use more than 20GB.

Note that the 2x16GB Crucial and 2x16GB Timetec have the exact same specs, same size, both dual ranked, both "made for MAC," so should theoretically play nice together, but no, when combined the speed and performance drop. Something to think about if you are considering adding some RAM now, and more RAM later.

Thank you for your analysis. I have 32 gb Crucial and 8 GB Apple (Micron). Very clear what configuration to do, and as you, I’m going with the 2 x 16 GB Crucial only.

 

[wilberforce]

So just to add my findings with different combination of RAM in my 2020 iMac which arrived today:

2x4GB Apple RAM (mfr Micron!) (total 8GB): 2667 MHz, Geekbench 7060
2x4GB Apple RAM + 2x16GB Crucial RAM (Total 40GB): 2133 MHz, Geekbench 8220
2x16GB Crucial RAM + 2x16GB Timetec (Hynix) RAM (Total 64GB): 2133 MHz, Geekbench 8240
2x16GB Cruclal RAM (Total 32GB): 2667 MHz, Geekbench 8680
2x16GB Timetec (Hynix) RAM (Total 32GB): 2667MHz, Geekbench 8570

So no surprise, mixing RAM brands or sizes drops the speed to 2133 MHz and degrades the Geekbench score slightly.
Above scores are for the i7 processor, with activity monitor running. Don't read too much into slight variations in scores. (For example 32 GB Timetec scoring slightly lower than 32 GB Crucial - I only ran it once).

So I am using the 32GB Crucial RAM, returning the Timetec RAM, and putting the Apple RAM aside. I hardly ever use more than 20GB.

Note that the 2x16GB Crucial and 2x16GB Timetec have the exact same specs, same size, both dual ranked, both "made for MAC," so should theoretically play nice together, but no, when combined the speed and performance drop. Something to think about if you are considering adding some RAM now, and more RAM later.

So I have a very interesting follow-up result, trying an arrangement suggested by mattmacpro in the order thread.
I put 16GB timetec (hynix) sticks in slots 1&2, and 16GB Crucial sticks in slots 3&4 (which you are not supposed to do.) I got this result:
2667 MHz, Geekbench 9001 !!
Does this mean I now have full speed and dual channel (because the sticks are the same size)?
How do I check if I really have dual channel?

 

[zestUK]

Hi wilberforce,

I returned my Timetec ram already, hence I would not be able to test it, would you mind to run a quick test on novabench and share with us the ram speed result of 16gbx2(crucial) + 16gbx2 (timetec) in A + A +B +B configuration and A + B + A + B as well please? Much appreciated.

the link below has my report on ram speed.

Memory upgrade compatibility list for iMac 2019-2020

Hi All, After reading as many posts regarding iMac 2020 ram upgrade as I can. I would like to share my little finding here. My mac is 3rd tier one. come with 2 x4gb Hynix ram. 1, I tested 2x16gb timetec kit. it works without mixing up with apple ram. once mix up with apple one. clock speed...

forums.macrumors.com

 

[pldelisle]

So I have a very interesting follow-up result, trying an arrangement suggested by mattmacpro in the order thread.
I put 16GB timetec (hynix) sticks in slots 1&2, and 16GB Crucial sticks in slots 3&4 (which you are not supposed to do.) I got this result:
2667 MHz, Geekbench 9001 !!
Does this mean I now have full speed and dual channel (because the sticks are the same size)?
How do I check if I really have dual channel?

You are are not in dual channel.
You can see in System Report. Your apeed might be 2666 mhz but you are not in dual channel mode.

 

[Neoaf]

OK so as my wanted amount of ram is 16 today I recieved second hand RAM removed from mac mini which is exactly the same ram as I had installed in my iMac.

I just put new sticks in empty slots.
I wonder if I have dual channel... Interestingly my GPU score is lower now?!

 

[pldelisle]

OK so as my wanted amount of ram is 16 today I recieved second hand RAM removed from mac mini which is exactly the same ram as I had installed in my iMac.

I just put new sticks in empty slots.
I wonder if I have dual channel... Interestingly my GPU score is lower now?!

Weird. Maybe just a matter of circumstances. Your mac should work fine. It has all the same memory modules. Memory transfert speed can show you are in dual channel mode.
Your Metal FPS is pushing down your score. Haven't you installed anything graphics intensive that could run in the background ?

 

[Neoaf]

Weird. Maybe just a matter of circumstances. Your mac should work fine. It has all the same memory modules. Memory transfert speed can show you are in dual channel mode.
Your Metal FPS is pushing down your score. Haven't you installed anything graphics intensive that could run in the background ?

there is almost nothing going on in the background. I also did geekbench but IDK if such result is ok or not because I didn't do it before adding RAM.

 

[ninecows]

New iMac 2020 arriving tomorrow

Forgive me my ignorance, but is it possible to only use one module or should you always (buy and) use them in pairs?

why?
2x16 Gb will take up two slots and would be money wasted if I later on wants to upgrade to the max 128.

I’m currently at 32 GB in my old Mac so anything less would feel wrong ;-)

Can’t really afford 2x32 right now.

 

[pldelisle]

Forgive me my ignorance, but is it possible to only use one module or should you always (buy and) use them in pairs?

Always in pairs.

why?

To benefit from dual-channel architecture.

2x16 Gb will take up two slots and would be money wasted if I later on wants to upgrade to the max 128.

So take 2x32 GB Crucial.

Can’t really afford 2x32 right now.

Then wait ?

 

[AppleLover87]

Hi AppleFans


iMac (Retina 5K, 27-inch, 2020)
3,8 GHz 8-Core Intel Core i7
AMD Radeon Pro 5500XT 8GB
Mac OS Catalina 10.15.6



I have tested 3 different sticks and this is the result :




CRUCIAL Total 32GB -- 2x 16GB Crucial CT16G4SFD8266 - DDR4 2666 1.2V CL19 - Geekbench 8890

CORSAIR Total 64GB -- 4x 16GB Corsair CMSX32GX4M2A2666C18 - DDR4 2666 1.2V CL18 - Geekbench 9030

SAMSUNG Total 32GB -- 1x 32GB Samsung M471A4G43MB1-CTD - DDR4 2666 1.2V CL19 - Geekbench 7730

SAMSUNG Total 64GB -- 2x 32GB Samsung M471A4G43MB1-CTD - DDR4 2666 1.2V CL19 - Geekbench 8810



I have tested 32GB Crucial and 64GB Corsair and CRUCIAL is the winner ! and its between 32 and 64GB.




I hope this helps

 

[Kb4791]

Hi AppleFans


iMac (Retina 5K, 27-inch, 2020)
3,8 GHz 8-Core Intel Core i7
AMD Radeon Pro 5500XT 8GB
Mac OS Catalina 10.15.6



I have tested 3 different sticks and this is the result :




CRUCIAL Total 32GB -- 2x 16GB Crucial CT16G4SFD8266 - DDR4 2666 1.2V CL19 - Geekbench 8890

CORSAIR Total 64GB -- 4x 16GB Corsair CMSX32GX4M2A2666C18 - DDR4 2666 1.2V CL18 - Geekbench 9030

SAMSUNG Total 32GB -- 1x 32GB Samsung M471A4G43MB1-CTD - DDR4 2666 1.2V CL19 - Geekbench 7730

SAMSUNG Total 64GB -- 2x 32GB Samsung M471A4G43MB1-CTD - DDR4 2666 1.2V CL19 - Geekbench 8810



I have tested 32GB Crucial and 64GB Corsair and CRUCIAL is the winner ! and its between 32 and 64GB.




I hope this helps

Many thanks for your testing.....

I have ordered 2x 16GB Crucial CT16G4SFD8266 and will discard the standard 8GB

 

[pldelisle]

Many thanks for your testing.....

I have ordered 2x 16GB Crucial CT16G4SFD8266 and will discard the standard 8GB

Although this kit works, this is not the good part number for certified Mac RAM from Crucial. See first post of this thread.

 

[wilberforce]

You are are not in dual channel.
You can see in System Report. Your apeed might be 2666 mhz but you are not in dual channel mode.

Where does it show in system report whether it is dual channel or not? All I know is the Geekbench score is good (repeatedly over 9000 for i7). This is what it shows for memory:

 

[pldelisle]

Where does it show in system report whether it is dual channel or not? All I know is the Geekbench score is good. This is what it shows for memory:

View attachment 947736

This is typical Dual channel memory speed. Ensure you get something like this.

Your system report is wrong. You installed your RAM incorrectly. It is not supposed to be in dual channel, so I don't know what's going on here ... One thing is sure is that installed this way, you are not in dual channel mode. And you are mixing Hynix RAM with Micron RAM. It will never work at full speed when properly installed.

Sorry I confounded you with another user.

 

[wilberforce]

View attachment 947739

This is typical Dual channel memory speed. Ensure you get something like this.

Your system report is wrong. You installed your RAM incorrectly. It is not supposed to be in dual channel, so I don't know what's going on here ... One thing is sure is that installed this way, you are not in dual channel mode. And you are mixing Hynix RAM with Micron RAM. It will never work at full speed when properly installed.

Sorry I confounded you with another user.

This is what I get from Novabench, which looks promising. I agree with you, I did not ever think this arrangement would work, which is why I am so surprised it apparently give very good benchmark scores! (for an i7). It is a real mystery what is going on:

 

[pldelisle]

But there is something wrong with your Metal 3D at 60 FPS. It's very low for a 5500XT.

 

[wavesm]

Hi AppleFans


iMac (Retina 5K, 27-inch, 2020)
3,8 GHz 8-Core Intel Core i7
AMD Radeon Pro 5500XT 8GB
Mac OS Catalina 10.15.6



I have tested 3 different sticks and this is the result :




CRUCIAL Total 32GB -- 2x 16GB Crucial CT16G4SFD8266 - DDR4 2666 1.2V CL19 - Geekbench 8890

CORSAIR Total 64GB -- 4x 16GB Corsair CMSX32GX4M2A2666C18 - DDR4 2666 1.2V CL18 - Geekbench 9030

SAMSUNG Total 32GB -- 1x 32GB Samsung M471A4G43MB1-CTD - DDR4 2666 1.2V CL19 - Geekbench 7730

SAMSUNG Total 64GB -- 2x 32GB Samsung M471A4G43MB1-CTD - DDR4 2666 1.2V CL19 - Geekbench 8810



I have tested 32GB Crucial and 64GB Corsair and CRUCIAL is the winner ! and its between 32 and 64GB.




I hope this helps

That comparison is not completely correct.

2* 16 cannot be compared to 1 * 32.

Also having all slots full is better than 2 available but on the other hand you cannot expand later.

 

[wilberforce]

But there is something wrong with your Metal 3D at 60 FPS. It's very low for a 5500XT.

Geekbench Metal Compute score is 42300 (with the "mixed" 64GB ram). Seems OK to me for 5500XT.

With only 32GB of Crucial ram (in the correct slots) the Geekbench Metal score was 43100. (I am not sure this slight difference is significant, as there is more variation from run to run.)

I cannot explain the 60 fps Novabench score. (Maybe capped by the monitor frequency?) All other gpu tests I run look fine.

 

[Ctg44]

All memory is 10% off today at Newegg.

 

[wilberforce]

This is what I deduce from my experiments:
In order to get full RAM speed and performance using 4 sticks of RAM in the 2020 27" iMac, apparently need to satisfy all of the following conditions:

1. All sticks must be the same size (GB) and rank
2. Sticks in Slots 1 and 2 must be identical (including brand and part number)
3. Sticks in Slots 3 and 4 must be identical

Of course, the most straightforward and sure way to ensure all 3 conditions are met, is to have all 4 sticks identical.

However, it is apparently possible to mix brands, provided they are same size and rank, which is useful to know if you cannot obtain sticks that are 100% identical to two sticks you already have.

It is a mystery to me why these 3 conditions need to be met. Not sure if it is intentional or if it is a bug. It is contrary to the common understanding that slots 1 and 3 must be identical, and 2 and 4 must be identical.

In any case, I will be returning the Timetec 2x16GB, and using the 2x16GB Crucial sticks alone. I don't need more than 32GB. I only did this as an experiment. YMMV.

 

[pldelisle]

It is a mystery to me why these 3 conditions need to be met.

Most PC motherboard have similar behaviour. Memory and motherboard relationship is picky.

 

[AppleLover87]

Although this kit works, this is not the good part number for certified Mac RAM from Crucial. See first post of this thread.

all the specs of this crucial memory is correct .... speed voltage timing, not because its not official "proved" its not good.

 

[pldelisle]

all the specs of this crucial memory is correct .... speed voltage timing, not because its not official "proved" its not good.

As I said :

Although this kit works

The correct, fully certified kit is CT2K16G4S266M . Up to you to chose it or not. Someone already tested CT16G4SFD8266 and it worked, so there is no problem. I don't know what is the difference between both modules. I prefer to take the certified kit. I don't see any reason not to take this certified kit over the generic kit.