Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

RAM upgrade question -- new iMac 27"

bizzwriter

bizzwriter

macrumors member
Original poster
Aug 4, 2010
73
24
Left coast...
Just bought a new iMac 27" for my wife -- bought the stock 8GB RAM, and will receive a Crucial 32GB (2x16) upgrade kit today.

The question: Should I go ahead and leave the old RAM in there when I upgrade, giving me 40GB RAM, or dump the old RAM? I know Apple sez that it's best for all the RAM sticks to be identical, but...

MANY thanks!
 
omeletpants said:
I had the same issue and dumped the 4gb modules leaving 32gb. In all my research, matching ram was the answer. Plus i never even use the 32gb
Click to expand...
Thanks for the superfast response om! Yeah -- makes sense to just dump the original 4GB sticks. 32GB should be plenty for what my wife does. THX
 
If it is the 2020 iMac (the current model), only put in the 2x16GB sticks.
If it is any other model year iMac, put in both 2x16 and 2x4.

Reason is that the 2020 iMac has a quirk where it will reduce the memory speed if sticks aren't equal size.
 
Last edited:
  • Like
Reactions: tonysabbath
Others have already answered you but I recommend removing Apple ram and only using the 2x 16GB sticks. Make sure to match them properly either slot 1 and 3 or 2 and 4
 
  • Like
Reactions: wilberforce
Save the origional ram and reinstall it if you sell the computer, then sell the 16mb ram
 
bizzwriter said:
Thanks for the superfast response om! Yeah -- makes sense to just dump the original 4GB sticks. 32GB should be plenty for what my wife does. THX
Click to expand...
Learn about "Swap Memory" then check it in the activity monitor. I'm betting you will never exceed the 32mb ram and go into "swap mode". That should make you feel better
 
wilberforce said:
If it is the 2020 iMac (the current model), only put in the 2x16GB sticks.
If it is any other model year iMac, put in both 2x16 and 2x4.

Reason is that the 2020 iMac has a quirk where it will reduce the memory speed if sticks aren't equal size.
Click to expand...
Oh wow -- thanks for the heads up on the quirk! It's the current model -- just bought brand-new from Apple (with my Apple card!)
 
omeletpants said:
Learn about "Swap Memory" then check it in the activity monitor. I'm betting you will never exceed the 32mb ram and go into "swap mode". That should make you feel better
Click to expand...
Yeah -- I used to watch that all the time, but realized I wasn't doing any swapping! So I don't bother anymore.
 
OP wrote:
"Thanks for the superfast response om! Yeah -- makes sense to just dump the original 4GB sticks"

Don't throw them out.
Keep the original RAM DIMMs in a safe place, and REMEMBER WHERE you keep them.

If you run into some "unexplainable crashes" with the upgraded RAM -- even after a while -- the "first order of diagnostics" would be to TAKE OUT the 3rd-party RAM and restore the factory-original RAM, to see what happens.
 
While it's correct that the 2020 iMac for some stupid reason runs mixed capacity memory at the lowest specified speed (2,133 MHz) I dare say you won't notice the difference. It barely shows in benchmarks and even if it does it's usually a single percentage decrease. In return, you get an extra 8 GB of RAM that the operating system and applications have at their disposal, which is worth much more than a single percentage performance increase.

Case in point: I ran the Unigine Heaven benchmark on Extreme with 8 GB of 2667 DDR4 and 24 GB of 2133 DDR4. Gues which one is which:

1.png
2.png
 
mj_ said:
While it's correct that the 2020 iMac for some stupid reason runs mixed capacity memory at the lowest specified speed (2,133 MHz) I dare say you won't notice the difference. It barely shows in benchmarks and even if it does it's usually a single percentage decrease. In return, you get an extra 8 GB of RAM that the operating system and applications have at their disposal, which is worth much more than a single percentage performance increase.

Case in point: I ran the Unigine Heaven benchmark on Extreme with 8 GB of 2667 DDR4 and 24 GB of 2133 DDR4. Gues which one is which:

View attachment 1925642View attachment 1925643
Click to expand...
iMac display is limited to 60Hz refresh rates so the fact both results are roughly the same is unsurprising. You need to use a memory benchmark, not a benchmark that’s gated by display refresh rates.
 
Of course memory throughput is going to be much higher but how is that even remotely relevant if the effect on real-world performance is negligible? Yes, there may be some highly memory-intensive applications (scientific calculations or data analysis to name two) that benefit from higher memory clock speeds and will run ever so slightly faster but even those will more often than not benefit much more from more rather than faster RAM.

That's the whole point I'm trying to make ;)
 
mj_ said:
While it's correct that the 2020 iMac for some stupid reason runs mixed capacity memory at the lowest specified speed (2,133 MHz) I dare say you won't notice the difference. It barely shows in benchmarks and even if it does it's usually a single percentage decrease. In return, you get an extra 8 GB of RAM that the operating system and applications have at their disposal, which is worth much more than a single percentage performance increase.

Case in point: I ran the Unigine Heaven benchmark on Extreme with 8 GB of 2667 DDR4 and 24 GB of 2133 DDR4. Gues which one is which:

View attachment 1925642View attachment 1925643
Click to expand...
Well, seeing Unigine Heaven is only using about 5% of the CPU and less than 1GB of DRAM (on my machine, extreme preset), it seems not surprising that DRAM speed and quantity make no difference, when it is governed by GPU and VRAM.

But yes, memory transfer speed does not make as much difference to actual performance as it sounds
 
Last edited:
Thanks for the continued discussion guys -- I'm learning lots! I put the old 4GB sticks in the Crucial package for the new RAM -- marked and in a safe place.
 
mj_ said:
Of course memory throughput is going to be much higher but how is that even remotely relevant if the effect on real-world performance is negligible? Yes, there may be some highly memory-intensive applications (scientific calculations or data analysis to name two) that benefit from higher memory clock speeds and will run ever so slightly faster but even those will more often than not benefit much more from more rather than faster RAM.

That's the whole point I'm trying to make ;)
Click to expand...
You’re using a benchmark that doesn’t stress ram to say that ram speeds don’t matter. It doesn’t make sense. Might as well run a disk benchmark and show how disk read/write speeds aren’t affected by slower RAM either.
 
Last edited:
OK, fine. What would be an application then that benefits from higher RAM clock speeds sufficiently for me to not only measure but also feel a difference?
 
mj_ said:
OK, fine. What would be an application then that benefits from higher RAM clock speeds sufficiently for me to not only measure but also feel a difference?
Click to expand...
Video editing, working in Logic. Basically anything where latency is an issue. There is a huge RAM thread where this was benchmarked and detailed last August through October. I’ll see if I can find a link.
 
@tonysabbath Sorry for offtopic in this old thread, but i can't send you a message. I'd like to know if you still have your iMac2020 and if you wold share the acpi dumps of it. You would help me a lot with this and I would be very grateful for it.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back