Given the fickle nature of these 2020 iMacs it's hard to say. If, in the end, the four sticks are truly identical, you should have 2667MHz speed and dual-channel. Somebody reported dropping down to 2133MHz with four "identical" sticks of RAM from the same company, but bought a few months apart. But, go for it! Pull the Apple RAM, install your first pair in the same slots, then throw your next set in the empty slots when it arrives. Good luck!
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Memory upgrade troubleshooting in 2020 27-inch iMacs
- Thread starter KrazyKanuck
- Start date
- Sort by reaction score
Glad to hear you're up and running at full power! What a shame about the Apple RAM. Knowing Apple, they probably charged all of us $600 for those first 8GB. Perhaps it's a firmware thing? Onwards though!
I placed two orders with Amazon about a day apart (one order for each pair). They came in a little, Crucial-branded plastic clamshell, which seemed respectable enough.
This is wonderful! Given the trouble so many of us have been having, this is a fairly definitive summary of our experiences. Thank you for spending your time doing this!
Yes, that’s the retail packaging version you.received. If you order factory direct, they ship differently (or at least they used to).
It believe it used to be micron.com, and was a godsend when they started selling factory direct. Ordering on the phone was common.
I built my first pc in the very early 90’s, and quality memory (SIMM) was very hard to find. The ones that had the tin leads were tough to find - it wasn’t good to mix tin connectors on the motherboard with the gold leads on the SIMM because there was concern of corrosion with he two different metals. I am showing my age... Sorry for the walk down memory lane - back to the regularly scheduled topic
Thanks for the good wishes
So how can I confirm speed and dual channel? Considering it also seems the springs are a wee bit delicate, I'd only like to pop these in once and not have to do trial and error comparisons.
Micron is the company. Crucial is one of their retail brands.
Unless you are a VAR or OEM ordering directly, never buy RAM or SSDs under the Micron name. These are sold without warranty of any kind. The price to OEMs and VARs is low enough that they are expected to eat the occasional failure—OEM deals are normally structured that way. Micron is very upfront about this on their web site—if you have a problem, contact your vendor as there is no warranty from Micron. Unfortunately, you can find Micron branded RAM and SSDs all over Amazon and eBay etc.
Their Crucial brand has excellent customer service That I’ve experienced first hand.
I bought 20 Crucial MX300 2TB SSDs for clients. Every one of them developed issues during the 3 year warranty period. After emailing SMART reports, I received RMAs and shipped them off to their service center in Nampa, ID. The MX300 being discontinued, I received MX500s by return mail. As it should be.
The Micron 1100 is the exact same unit as the MX300. Imagine how much money I’d be out if I had bought 20 of those.
Which is exactly half speed. Would you mind doing the the test with the RAM set up for dual channel( same size sticks in slots 1 and 3 and other same size sticks in slots 2 and 4) and see what Geekbench sees as the speed? My running theory is it is a bug with the system report.
That sure would be a nice, easy fix. See MMidnight's benchmarks a few posts back. He ran five different configurations, and the one where he left the Apple RAM in the original slots, but added 3rd party to the open slots (ie. the 2133MHz configuration) he was scoring less in Geekbench than with the Apple RAM removed. This is making me think that it may be firmware or something at the hardware level.
Ah didn’t see that. That is odd how it scored lower. Still curious to see if Geekbench sees the RAM speed at 2667 MHz despite the system report saying 2133 MHz.
Why does About This Mac say "Your Mac contains 4 memory slots, each of which accepts a 2133 MHz DDR4 memory module." - shouldn't it say the full speed? Super weird.
I just did a GeekBench test and it says my RAM is at 1066MHz. System Report lists each module at 2133MHz.
The setup I have is:
Slot 1: 16GB (OWC)
Slot 2: 4GB (Apple)
Slot 3: 16GB (OWC)
Slot 4: 4GB (Apple)
The Apple RAM is in the slots that it originally shipped in.
Going to try another test with just the OWC RAM.
I just did a GeekBench test and it says my RAM is at 1066MHz. System Report lists each module at 2133MHz.
The setup I have is:
Slot 1: 16GB (OWC)
Slot 2: 4GB (Apple)
Slot 3: 16GB (OWC)
Slot 4: 4GB (Apple)
The Apple RAM is in the slots that it originally shipped in.
Going to try another test with just the OWC RAM.
Last edited:
It would be really interesting running the same tests by someone with more than 2 slots used by Apple's factory memory and see if they had the same results.
My theory is there's something about RAM that Apple is not fully open about ,probably about timings and architecture, or memory addressing , so when different sodimms are used they may have to (somehow) synchronise and lower speeds or something like that.
That might also explain why it also happens with same brand modules ( that's why they usually recommend buying them in pairs, so they belong to the same batch and there are no weird timing related issues).
Does any of you know if anyone tried contacting support about this?
I'm thinking about it, as i would like to be able to use the original modules for those extra 8Gb.
When i have time i'll try and check for the memory speed in Geekbench just with the original stock memory and see what it shows. I cannot remember if the 1333Mhz speed also appeared in my first run and i've tried checking on youtube benchmarks but they seem to have the "right window size" so you cannot see it.
My theory is there's something about RAM that Apple is not fully open about ,probably about timings and architecture, or memory addressing , so when different sodimms are used they may have to (somehow) synchronise and lower speeds or something like that.
That might also explain why it also happens with same brand modules ( that's why they usually recommend buying them in pairs, so they belong to the same batch and there are no weird timing related issues).
Does any of you know if anyone tried contacting support about this?
I'm thinking about it, as i would like to be able to use the original modules for those extra 8Gb.
When i have time i'll try and check for the memory speed in Geekbench just with the original stock memory and see what it shows. I cannot remember if the 1333Mhz speed also appeared in my first run and i've tried checking on youtube benchmarks but they seem to have the "right window size" so you cannot see it.
Even with the Apple RAM only, Geekbench reports 1333 (half).
As mentioned earlier I think the problem is with mixing different sizes more than different brands.
The result of mine is somehow different.
Normal setup is like
Slot 1: 8GB (Kingston)
Slot 2: 4GB (Apple)
Slot 3: 8GB (Kingston)
Slot 4: 4GB (Apple)
And yes, it gives me 24GB at 2133MHz
When I put the sticks like this,
Slot 1: 4GB (Apple)
Slot 2: 4GB (Apple)
Slot 3: 8GB (Kingston)
Slot 4: 8GB (Kingston)
And yes, it gives me 24GB at 2667MHz
The strange thing is when running Geenkbench, the second setting scores higher than the first one, which is different from the OP.
BTW, I am using 24GB at 2133MHz right now. Maybe someday later I would get another 8GBx2 if I am out of control.?
Normal setup is like
Slot 1: 8GB (Kingston)
Slot 2: 4GB (Apple)
Slot 3: 8GB (Kingston)
Slot 4: 4GB (Apple)
And yes, it gives me 24GB at 2133MHz
When I put the sticks like this,
Slot 1: 4GB (Apple)
Slot 2: 4GB (Apple)
Slot 3: 8GB (Kingston)
Slot 4: 8GB (Kingston)
And yes, it gives me 24GB at 2667MHz
The strange thing is when running Geenkbench, the second setting scores higher than the first one, which is different from the OP.
BTW, I am using 24GB at 2133MHz right now. Maybe someday later I would get another 8GBx2 if I am out of control.?
Yes, I've been thinking about this a bit and I seem to recall all of my Geekbench scores saying 1333MHz regardless of my RAM configuration. The technical reason for this is beyond my knowledge.
When I search for other system, the frequency is often 1xxx MHz, it is clear that Geekbench shows the single frequency but not the doubled one.
DDR = double data rate.
1333 MHz is clock frequency
2666 MT/s is transfers/second
At least that is how I understand it.
1333 MHz is clock frequency
2666 MT/s is transfers/second
At least that is how I understand it.
Have been running 2x16GB OWC with the apple ram removed and the system reports 2666. I just got my second set of OWC and now have 64GB in, all the same sticks and the system reports it as 2133 and not 2666. So not sure if it’s worth returning all of the 16gb sticks and just going with 2x32gb? I’m not sure how much of a real world difference I’ll see in the MHZ speed increase?
Could you screenshot the ram details in system info app, to check any difference in part number of the four sticks
You could just place the same set in the same channel to check whether it returns to 2667MHZ
Hi guys, long time listener first time caller here.
I bought the new i7 2020 iMac on Day One from an Apple Store here in Sydney, Australia and picked up 2 x 16GB Crucial RAM a couple of days later and have encountered the same issue as above.
I reported this to Apple Support who grabbed logs from my machine, escalated it to a senior product specialist who escalated it in turn to Apple Engineering.
The result finally came back a couple of days ago that in order to achieve full memory performance, all 4 slots must be occupied with exactly the same size and spec RAM.
I pointed out to them that this is not documented on their web site and was advised I could either return my iMac and order a model with more RAM or go out and buy additional RAM to get more than the 32GB I was expecting to have in total.
I thought it was a pretty irresponsible reply as I, like the rest of you, was simply following Apple's own requirements to upgrade RAM and there is not a single mention of the 4 x identical sticks requirement. To say that my only remedy for fixing Apple's mistake is to spend more money isn't very customer-oriented.
I bought the new i7 2020 iMac on Day One from an Apple Store here in Sydney, Australia and picked up 2 x 16GB Crucial RAM a couple of days later and have encountered the same issue as above.
I reported this to Apple Support who grabbed logs from my machine, escalated it to a senior product specialist who escalated it in turn to Apple Engineering.
The result finally came back a couple of days ago that in order to achieve full memory performance, all 4 slots must be occupied with exactly the same size and spec RAM.
I pointed out to them that this is not documented on their web site and was advised I could either return my iMac and order a model with more RAM or go out and buy additional RAM to get more than the 32GB I was expecting to have in total.
I thought it was a pretty irresponsible reply as I, like the rest of you, was simply following Apple's own requirements to upgrade RAM and there is not a single mention of the 4 x identical sticks requirement. To say that my only remedy for fixing Apple's mistake is to spend more money isn't very customer-oriented.
This is concerning. First time I’ve heard if you buy 2 pairs from the same vendor you can have the same issues as people are having with the original Apple memory.
Adding one more point of data. 32 gigs of RAM from OWC in the open slots. 40 gigs total of RAM but running at 2133mhz. I'll pull the 8 gigs out to get 2667 going.
Anyone who stayed with the standard 8 gigs want another 8 gigs? I'll let them go for cheap. PM me. (I'll edit this out if not allowed)
Anyone who stayed with the standard 8 gigs want another 8 gigs? I'll let them go for cheap. PM me. (I'll edit this out if not allowed)