MacInTouch Amazon link...

iMacs and related issues

Channels
Apple, Troubleshooting, Products
Hasn't it been more like decades that he's been promising a vaporware Mac version?
A few years ago, Gibson was alpha-testing a new SpinRite disk engine that used available RAM and modern disk controller capabilities to greatly speed up operations. I think it may have also booted with EFI or UEFI instead of requiring a BIOS. Then he got the idea for his SQRL authentication process, so SpinRite was sidelined.

I quit following the progress of SQRL at some point. It now looks like the pages for SQRL at grc.com have not been updated for two years, and there is no mention of the upgraded SpinRite. It is sad to see good (possibly great) ideas wither on the vine this way.
 


The part about using CCC or SuperDuper to clone to an external SSD is good advice even if the main iMac fusion drive is ok.
A few years ago (2010) I had a one year-old iMac, which I backed up daily to another hard drive using Carbon Copy Cloner. One day, CCC started reporting problems with backing up some data. I queried the guys at CCC; they examined what was happening and told me that it looked as if CCC (the program) was encountering occasional bad data, which could mean that the hard drive was failing.

I took my backup hard drive offline, started using another one, so as to preserve what was on the previous one, and also looked at Smart Reporter, which had been running all the time. Well, Smart Reporter showed no errors, and Apple's Diagnostics also found no problems.

However, in about a week, the internal hard drive failed. Smart Reporter finally discovered the failure. I took the iMac to Apple, who agreed and replaced the hard drive. But if I hadn't been backing up using CCC, I could have been in serious trouble.
 


I quit following the progress of SQRL at some point. It now looks like the pages for SQRL at grc.com have not been updated for two years, and there is no mention of the upgraded SpinRite. It is sad to see good (possibly great) ideas wither on the vine this way.
I regularly listen to Steve Gibson's Security Now podcast, where he's often been giving updates on SQRL. From my last listen, he is in the final throws of wrapping up the SQRL project. Steve is a perfectionist; he's wanted to make sure that everything about SQRL is right, from its implementation right down to the complete user experience.

Lately he's been setting up a SQRL forum system (complete with working SRQL authentication for login, with the help of other developers) so that new/novice users have a place to get instructions and discuss/ask questions about SQRL so they don't end up in any sort of "Linux RTFM" world. There are also several clients (iOS/Android/Windows-Wine) that are getting their final touches, and he's working on updating documentation on the final specs.

From what I hear, it'll be quite a neat thing to use, and I'm looking forward to it. I know I probably sound like I'm defending him (and, really, I am), and I, too, have been (im)patiently waiting on all the neat improvements to SpinRite he's mentioned over time (especially now that Apple has soldered-in storage), but he is only one man, working on a really great project. I'm frankly amazed he's been able to get it done, especially considering he's doing it with no intention of getting any monetary gains from it whatsoever.

He's also indicated that once his focus is back on SpinRite, he's planning on releasing incremental point-release updates, ASAP, for each new feature he wants to add, which will be freely available to all SpinRite 6 license holders.
 


A caveat: bizarrely, running a drive certification may not show drive problems, but, conversely, correct them (only temporarily). I had this happen recently, to my shock, with a completely dysfunctional hard drive. After a certification revived it, I finally wiped all data off the drive and trashed it.
Ric, did the drive show any I/O errors during the certify even though the certify passed? I always think of the certify as a way to exercise the drive but rely on the I/O error count and what I see in the SMART data to diagnose if the drive is actually failing. Certify will probably pass the drive if it thinks the errors are recoverable, but once I see any errors, I always presume the drive to be on its way out.
 


Ric Ford

MacInTouch
Ric, did the drive show any I/O errors during the certify even though the certify passed?
No, that is what was bizarre. Here was an ancient drive with a great number of hours on it that showed huge numbers of errors in its SMART data and was so dysfunctional that I couldn't even install a new system and/or keep it running - just constant failures. Yet, at the very end, it passed SoftRAID certification with no errors. (This was so bizarre, I emailed SoftRAID about it.)

I think what happened was this: At one point during this adventure, I was desperately trying to securely wipe the drive to eliminate the client's data before recycling the iMac, and the only thing that ultimately worked was a command-line operation. (I think this was the diskutil command with zeroDisk, randomDisk or secureErase subcomand, but I don't recall exactly.) What apparently happened was that this command somehow magically remapped all the bad blocks, so the subsequent Certify operation succeeded without error, despite the fact that the drive was in horrendous shape and unlikely to work for any extended period of time.
 


I called Apple Support, expecting to be setting up a drive replacement under AppleCare, but the Apple tech suggested running the diagnostics reached through holding down the ‘D’ key while starting up. That reported no issues with the computer. She said Apple does not recommend “third-party” applications for this purpose and that the Apple diagnostic test is reliable.
Apple Diagnostics (formerly Apple Hardware Test or AHT) does a quick basic check to make sure the drive is detected. I would make an appointment and bring it to your local Apple Store. They can run MRI to check the SMART status. There is also a Storage Diagnostic you can have run at the Genius Bar that is a more thorough test of the drive(s). If both of those tests pass, ask to have the computer checked in for extended Apple Service Diagnostics (ASD) at no charge. Make sure you have a Time Machine backup of your data, in case it fails - if not, you can provide an external drive and ask Apple to perform a free Data Transfer before starting extended diagnostics.
 



Apple Diagnostics (formerly Apple Hardware Test or AHT) does a quick basic check to make sure the drive is detected. I would make an appointment and bring it to your local Apple Store. They can run MRI to check the SMART status. There is also a Storage Diagnostic you can have run at the Genius Bar that is a more thorough test of the drive(s). If both of those tests pass, ask to have the computer checked in for extended Apple Service Diagnostics (ASD) at no charge. Make sure you have a Time Machine backup of your data, in case it fails - if not, you can provide an external drive and ask Apple to perform a free Data Transfer before starting extended diagnostics.
That sounds like a good plan to me, Will. I made an appointment for this Wednesday. Thanks.
 


That sounds like a good plan to me, Will. I made an appointment for this Wednesday. Thanks.
A month ago I promised to report back on my experiences since my initial post here regarding diagnostic results from DriveDx. Those results indicated that 8 sectors of my iMac’s boot disk had been reallocated and it was, therefore, “Failing.” Apple’s Disk Utility had reported “No Issues.” I’m finally here with a report.

As suggested here, I made an appointment at the local Apple Store and brought the iMac to the Genius Bar there. I requested they run the “MRI” test and their “Storage Diagnostic” on the computer. Both diagnostics reported no issues with the drive. I then discussed checking the computer in for the extended Apple Service Diagnostics, which they agreed to do, but I decided not to leave it there then, since the testing would take several days, and I needed it particularly for a project I was working on. I could bring it back when I’m ready (or sooner, if the reallocated sector count changes!).

In the meantime, I contacted the producers of DriveDx, BinaryFruit, and asked what they considered to be the threshold for failure in terms of number of reallocated sectors. They responded as follows :
A rise of the bad sector count usually indicates physical defects of the disk surface and/or problems in the mechanical subsystem.

According to the worldwide statistics - after the first offline reallocation, drives have over 21 times higher chances of failure within 60 days than drives without offline reallocations. That is why DriveDx warns you and sets status of this HDDs as "failing".

So, this is what you should do in your case: if your drive is "mission critical" or you are very sensitive to possible unexpected forced outages - replace your drive.

Otherwise, backup your (important) data and keep an eye on following drive health indicators:
* #5 Reallocated Block Count​
* #197 Current Pending Sectors Count​
* #198 Offline Uncorrectable Sector Count​

If the numbers will remain the same and is not big, and there are no any side effects (system freezes, kernel panics, etc) - all is OK. If the numbers will increase over time - replace your drive.
I then got knocked out of commission by a virus (me, not the computer!) for a couple of weeks. Since then, in addition to running Time Machine, I have been doing a daily SuperDuper backup and running DriveDx daily. I also run TechTool Pro 11 periodically and note its SMART report. The reallocated sector count has remained at 8, and no other indicators have changed. I’ve decided to continue with this approach for the time being, unless and until I see movement in the reallocated sector count or in other indicators.

In the meantime I have a couple of items I hope someone will clarify, since I’m not knowledgeable in this area:
1. How many sectors are there on a 2TB rotating disc? I suspect there are very many and that, for example, the 8 reallocated sectors my disk is reporting represent an extremely small percentage of the sectors available.​
2. Is it possible that the 8 sectors reported as reallocated have been there since the computer was assembled? (I can’t remember when I first ran DriveDx.)​

Thanks
 


1. How many sectors are there on a 2TB rotating disc? I suspect there are very many and that, for example, the 8 reallocated sectors my disk is reporting represent an extremely small percentage of the sectors available.
If the sectors are 512 bytes per sector, then there should be roughly 3,906,250,000 sectors, depending on how the manufacturer defines a terabyte.
 


A month ago I promised to report back on my experiences since my initial post here regarding diagnostic results from DriveDx. Those results indicated that 8 sectors of my iMac’s boot disk had been reallocated and it was, therefore, “Failing.” Apple’s Disk Utility had reported “No Issues.” I’m finally here with a report.
... In the meantime I have a couple of items I hope someone will clarify, since I’m not knowledgeable in this area:
1. How many sectors are there on a 2TB rotating disc? I suspect there are very many and that, for example, the 8 reallocated sectors my disk is reporting represent an extremely small percentage of the sectors available.​
2. Is it possible that the 8 sectors reported as reallocated have been there since the computer was assembled? (I can’t remember when I first ran DriveDx.)​
In my home iMac, I have a 3TB internal rotational drive. So, as Richard Murray reported, there are billions of sectors. My drive, which is very typical for standard rotational drives, was manufactured to have a whopping 36 "spare" sectors. Currently, none of the sectors has been put into service. So I have zero reallocated sectors. But once those have all been reallocated, because an area larger than 36 sectors goes bad, then my disk will be in a state of having permanent "bad blocks." At that point, utility software would/should report I/O errors. Until then, all blocks "in service" are functioning properly. No I/O errors.

Normally, when an area of a disk with billions of sectors goes bad, it's a lot more than 36, and they will all be used up in a fairly short period of time, leaving the files at risk.

I have seen, though, where some disks had a small manufacturing defect, and several sectors were reallocated very early after being putting into service. After that, the disk has been reliable for years. Obviously, no promises can be made, but Mike, you seem to fall into that category. You've been stuck at 8 for some period of time and never witnessed reallocations being zero. My suggestion would be to keep an eye on it - if reallocations rise above 8, replace the drive immediately and restore from backup. If you prefer not to have this issue hanging around, you can replace the drive now, for peace of mind.

Additional information: ATA/IDE and all its derivatives have spare blocks. 36 and 100 are the most common numbers of provided spare blocks. Once they are all used up, the drive is considered unreliable. Reformatting the drive will not solve the problem. (That was the case with SCSI but has never been the case with ATA/IDE/SATA/eSATA, etc.)

Reallocation generally causes data loss, because if a sector has file data and goes bad, that sector is replaced with a spare block that contains no data. So a chunk of file data has now been replaced with zeros. In normal circumstances, which means excluding some high-end RAID software and a couple of other circumstances, reallocations happen silently. There is no notification process, and who knows which file was affected.
 



In my home iMac, I have a 3TB internal rotational drive. So, as Richard Murray reported, there are billions of sectors. My drive, which is very typical for standard rotational drives, was manufactured to have a whopping 36 "spare" sectors. Currently, none of the sectors has been put into service. So I have zero reallocated sectors. But once those have all been reallocated, because an area larger than 36 sectors goes bad, then my disk will be in a state of having permanent "bad blocks." At that point, utility software would/should report I/O errors. Until then, all blocks "in service" are functioning properly. No I/O errors.

Normally, when an area of a disk with billions of sectors goes bad, it's a lot more than 36, and they will all be used up in a fairly short period of time, leaving the files at risk.

I have seen, though, where some disks had a small manufacturing defect, and several sectors were reallocated very early after being putting into service. After that, the disk has been reliable for years. Obviously, no promises can be made, but Mike, you seem to fall into that category. You've been stuck at 8 for some period of time and never witnessed reallocations being zero. My suggestion would be to keep an eye on it - if reallocations rise above 8, replace the drive immediately and restore from backup. If you prefer not to have this issue hanging around, you can replace the drive now, for peace of mind.

Additional information: ATA/IDE and all its derivatives have spare blocks. 36 and 100 are the most common numbers of provided spare blocks. Once they are all used up, the drive is considered unreliable. Reformatting the drive will not solve the problem. (That was the case with SCSI but has never been the case with ATA/IDE/SATA/eSATA, etc.)

Reallocation generally causes data loss, because if a sector has file data and goes bad, that sector is replaced with a spare block that contains no data. So a chunk of file data has now been replaced with zeros. In normal circumstances, which means excluding some high-end RAID software and a couple of other circumstances, reallocations happen silently. There is no notification process, and who knows which file was affected.
Thank you, Rusty, for that detailed review of sector availability and the implications of sector reallocation (loss of the data that was there). Most of what resides on my hard drive is certainly not what I would categorize as 'mission critical', and loss of the data on a few sectors would likely never be noticed. However, since there is no notification of what was lost, an error in an important result could lie hidden. I'm re-thinking my approach to this.
 


I just purchased a built-to-order 2019 iMac 27". It won't arrive for almost 2 weeks. I ordered the minimum RAM (8 GB) as it is far cheaper to buy memory from a third party. Does anybody know if this minimal configuration comes with two 4GB DIMM's, or with a single 8 GB DIMM? Thank you.
 


Ric Ford

MacInTouch
I just purchased a built-to-order 2019 iMac 27". It won't arrive for almost 2 weeks. I ordered the minimum RAM (8 GB) as it is far cheaper to buy memory from a third party. Does anybody know if this minimal configuration comes with two 4GB DIMM's, or with a single 8 GB DIMM? Thank you.
I don’t know, but I suspect it’s two DIMMs. You might check with OWC for a better answer - they probably know.
 


I just purchased a built-to-order 2019 iMac 27". It won't arrive for almost 2 weeks. I ordered the minimum RAM (8 GB) as it is far cheaper to buy memory from a third party. Does anybody know if this minimal configuration comes with two 4GB DIMM's, or with a single 8 GB DIMM? Thank you.
The Apple specs page shows it as two 4GB modules.
 



I need some advice, please, with a new 2019 iMac and challenges with new RAM. I'm not sure if it has bad memory slots on the logic board, bad Apple RAM (two x 4GB), or bad OWC RAM (two x 16GB). RAM tests seem to be failing sometimes. The new RAM was installed correctly and seated properly in Banks 1 and 3, with Apple's in 0 and 2.

Apple's 2-minute self-diagnostics pass and TechTool's 5-minute diagnostics pass. OWC's preferred tester, Rember, using Memtest 4.22, sometimes passes (after 100 minutes) and sometimes aborts after 5 minutes, testing Random Value, but doesn't display a message. That's a failure, right?

Testing with only Apple's RAM installed, or only OWC's, produces the same results — i.e. it passes 5 of 5 times, then the test stops after passing 2 of 5 (which seems like a failure), then maybe passing 0 of 5, usually stopping on the Random Value test. Testing has been done from the factory OS and also a bootable macOS 10.4.6 flash drive, with multiple power cycles, countless times.

With the first set of OWC RAM, which has the correct specs, the computer froze a few times, then the computer would only show a solid black screen through all power cycles. Removing that RAM let the computer boot, and these tests have been with the replacement set.

My only conclusion is the memory slots themselves are bad, and the RAM is fine, since Rember/Memtest randomly stops regardless of what RAM is installed. Is it possible for RAM (the correct type) to damage memory slots, if it is damaged?

I'm thinking of taking the iMac to Apple and hope their diagnostics will show the same issues, or that they will trust the Memtest results. Unfortunately, it's past the two-week return window, but hopefully they'll replace something.
 


I need some advice, please, with a new 2019 iMac and challenges with new RAM. I'm not sure if it has bad memory slots on the logic board, bad Apple RAM (two x 4GB), or bad OWC RAM (two x 16GB). RAM tests seem to be failing sometimes. ...
Based on your very thorough testing it looks like the RAM slots are bad. I would take the machine to Apple if one of their stores is nearby. You might want to remove the OWC RAM, Apple might blame everything on OWC if their RAM is installed. Hopefully, they will either replace the Logic Board or just give you a new machine. I don't believe you can replace the RAM slots individually, but I might be wrong about that.

I own a 2019 iMac 27" running a bunch of OWC RAM. Their RAM is usually pretty reliable, but you never know.
 


You should take the iMac in to Apple. But first remove the OWC RAM and leave it out. Then test the Apple RAM until you can get a documented failure. I'm not sure about the program(s) you're using to test the RAM, but even if they won't write a printable report, capture the information about the failure in a photo....

Good luck. I've had two experiences with taking Macs to the Apple store with non-Apple RAM and suggest it's best to leave that RAM at home.
 


Based on your very thorough testing it looks like the RAM slots are bad. I would take the machine to Apple if one of their stores is nearby. You might want to remove the OWC RAM, Apple might blame everything on OWC if their RAM is installed. Hopefully, they will either replace the Logic Board or just give you a new machine. I don't believe you can replace the RAM slots individually, but I might be wrong about that.
Or the RAM controller. If memory serves, Apple does a much more intensive memory test in-store, though it can take a while. If it fails the test, Apple has to repair or replace the machine under warranty (90 days form the date of purchase — and if you didn't already, you can always purchase AppleCare anytime during that period).
 


Not sure this is applicable, but my 2019 iMac would not boot with the two 4GB strips that it came with in Banks 0 & 2 and two OWC 16GB strips in Banks 1 & 3. OWC told me the two OWC strips should go in the two banks closest to the hinge (2 & 3?) and the Apple strips in the other two banks. It has booted just fine since. They added that if there are only two strips they go in banks 0 & 2, which is how it was configured when I got it.
 


Ric Ford

MacInTouch
I need some advice, please, with a new 2019 iMac and challenges with new RAM. I'm not sure if it has bad memory slots on the logic board, bad Apple RAM (two x 4GB), or bad OWC RAM (two x 16GB).
Are you sure you got the right RAM for the new model? It uses faster RAM than the previous version:
Apple Support said:
Install memory in an iMac

iMac (Retina 5K, 27-inch, 2019)
  • PC4-21333
  • Unbuffered
  • Nonparity
  • 260-pin
  • 2666MHz DDR4 SDRAM
iMac (Retina 5K, 27-inch, 2017)
  • PC4-2400 (19200)
  • Unbuffered
  • Nonparity
  • 260-pin
  • 2400MHz DDR4 SDRAM
 


Thanks for the advice on this iMac's RAM. The OWC RAM, which I will leave out when I take it in, is for the 2019 model. Since it was the correct type, it's hard to imagine how, or if, it could have damaged anything, although that first set did produce black screens until it was removed. Maybe it's just coincidence.

If others are interested, MemTest has been around for a long time and is open source. The original website is offline, but the final 4.23 it is available on Github. I don't know how to compile it, but someone should and make it available.

Rember is a nice free GUI for it, which includes MemTest 4.22, and that's available from Kelley Computing. MemTest 4.22 is also listed there as a separate download. There's also MemTest86, which I haven't tried, and I don't know if it's related.

I may have to return the OWC RAM before I get the iMac back for further testing. In that case, is there a good brand or brands of RAM to consider, or should I stick with OWC? There's Kingston and Crucial.
 


Ric Ford

MacInTouch
There's also MemTest86, which I haven't tried, and I don't know if it's related. I may have to return the OWC RAM before I get the iMac back for further testing. In that case, is there a good brand or brands of RAM to consider, or should I stick with OWC? There's Kingston and Crucial.
MemTest86 has worked well for me in the past with Macs (after creating a bootable, DOS-style system on CD or USB stick).

Another RAM vendor is Ramjet, which has long provided Mac memory. I haven't bought RAM from them for quite a while, but they were very good in the past.
 


In regards to Charles's comment, two OWC tech reps confirmed that when all four banks are populated on the 2019 iMac, the original RAM remains in slots 0 and 2, and the new RAM goes in 1 and 3. I haven't found any official documentation to verify.
 


Ric Ford

MacInTouch
In regards to Charles's comment, two OWC tech reps confirmed that when all four banks are populated on the 2019 iMac, the original RAM remains in slots 0 and 2, and the new RAM goes in 1 and 3. I haven't found any official documentation to verify.
I searched Apple Support (and elsewhere) for that documentation yesterday and also couldn't find it.

However, Apple changed RAM slot configurations in the classic Mac Pro, and one year's model required the opposite configuration vs. another year's model, which led to a lot of confusion and problems. I don't know if anything similar has happened with different iMac model years.
 


In regards to Charles's comment, two OWC tech reps confirmed that when all four banks are populated on the 2019 iMac, the original RAM remains in slots 0 and 2, and the new RAM goes in 1 and 3. I haven't found any official documentation to verify.
Yeah, I thought they would go in alternate slots, like they are in my early 2008 Mac Pro. But, as I said, that didn't work. Methinks OWC techs need to have a meeting, because they are telling two different stories.
 


I searched Apple Support (and elsewhere) for that documentation yesterday and also couldn't find it. However, Apple changed RAM slot configurations in the classic Mac Pro, and one year's model required the opposite configuration vs. another year's model, which led to a lot of confusion and problems. I don't know if anything similar has happened with different iMac model years.
Web search turns up How to upgrade the 27-inch 5K iMac (2019) with a massive 128GB of RAM. which indicates pairs are 0-2 and 1-3, though it doesn't specifically address different sizes.

Aside: surprised to trip over PDF repair guides for 2019 iMacs at Apple Support Manuals.
Apple Support said:
 


Ric Ford

MacInTouch
For what it's worth, I checked the 2017 iMac 5K with its OWC RAM upgrade, for what it's worth. This model alternates RAM slots for pairing, so the configuration is:
4GB (Apple)​
16GB (OWC)​
4GB (Apple)​
16GB (OWC)​

That's working well, so far, providing faster RAM performance than the Apple stock configuration. (I haven't run MemTest86 or any other memory testers, however, apart from the Geekbench performance tests.)
 


For what it's worth, I checked the 2017 iMac 5K with its OWC RAM upgrade, for what it's worth. This model alternates RAM slots for pairing, so the configuration is:
4GB (Apple)​
16GB (OWC)​
4GB (Apple)​
16GB (OWC)​
That's working well, so far, providing faster RAM performance than the Apple stock configuration. (I haven't run MemTest86 or any other memory testers, however, apart from the Geekbench performance tests.)
Ric, is that alternating RAM installation due to the different-size modules? I ask, because I, too, have a 2017 iMac 5K and have 8GB in each of the four slots, but two of the modules are from Apple and two from OWC. (I have been purchasing OWC RAM modules for many, many years and have yet to encounter any problems.)
 


Ric Ford

MacInTouch
Ric, is that alternating RAM installation due to the different-size modules?
Yes. However, it may make sense, as a precaution, to do the same thing with same-sized modules from different sources (e.g. OWC vs. Apple), just in case there might be any differences.
 


Ric, is that alternating RAM installation due to the different-size modules? I ask, because I, too, have a 2017 iMac 5K and have 8GB in each of the four slots, but two of the modules are from Apple and two from OWC. (I have been purchasing OWC RAM modules for many, many years and have yet to encounter any problems.)
I have been told that it is due to dual-channel memory which accesses two DIMMs at one time and where discrepancies in performance between DIMMs can cause problems. The alternating layout in this case matches the access pairing.
 


Memory on my iMac in About This Mac shows 40 GB installed:

4 GB 16 GB
4 GB 16 GB

Not sure how the slot numbers go in this screen, or for that matter how they are numbered in the computer. All I know for sure is that the two 16 GB strips are in the two slots closest to the hinge, as I've already mentioned.
 


Ric Ford

MacInTouch
Memory on my iMac in About This Mac shows 40 GB installed:
4 GB 16 GB​
4 GB 16 GB​
Not sure how the slot numbers go in this screen, or for that matter how they are numbered in the computer. All I know for sure is that the two 16 GB strips are in the two slots closest to the hinge, as I've already mentioned.
On this 2017 iMac 5K, 4GB and 16GB cards are installed in alternating slots (not in adjacent pairs). Running macOS Sierra, About This Mac > Memory shows:

4GB4GB
16GB16GB

System Information > Hardware > Memory shows this (with OWC 16GB modules and Apple 4GB modules from the factory):
Memory Slots:

ECC: Disabled
Upgradeable Memory: Yes

BANK 0/DIMM0:

Size: 4 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x802C
Part Number: 0x344154463531323634485A2D3247334532202020

BANK 0/DIMM1:

Size: 16 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x00
Part Number: -

BANK 1/DIMM0:

Size: 4 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x802C
Part Number: 0x344154463531323634485A2D3247334532202020

BANK 1/DIMM1:

Size: 16 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x00
Part Number: -
As mentioned, I tested RAM performance with Geekbench before and after the upgrade and saw improved memory bandwidth (with slightly worse latency) afterwards.

 


On this 2017 iMac 5K, 4GB and 16GB cards are installed in alternating slots (not in adjacent pairs). Running macOS Sierra, About This Mac > Memory shows:
4GB4GB
16GB16GB
System Information > Hardware > Memory shows this (with OWC 16GB modules and Apple 4GB modules from the factory):
Memory Slots:

ECC: Disabled
Upgradeable Memory: Yes

BANK 0/DIMM0:

Size: 4 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x802C
Part Number: 0x344154463531323634485A2D3247334532202020

BANK 0/DIMM1:

Size: 16 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x00
Part Number: -

BANK 1/DIMM0:

Size: 4 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x802C
Part Number: 0x344154463531323634485A2D3247334532202020

BANK 1/DIMM1:

Size: 16 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x00
Part Number: -
As mentioned, I tested RAM performance with Geekbench before and after the upgrade and saw improved memory bandwidth (with slightly worse latency) afterwards.
That's interesting. On my 2017 5K iMac, with the two 4Gbyte slot-protectors from Apple and two 16Gbyte cards from OWC, System Information shows pairs in adjacent slots.
BANK 0/DIMM0:
Size: 16 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x859B
Part Number: -
Serial Number: -

BANK 0/DIMM1:
Size: 16 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x859B
Part Number: -
Serial Number: -

BANK 1/DIMM0:
Size: 4 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x80AD
Part Number: -
Serial Number: -

BANK 1/DIMM1:
Size: 4 GB
Type: DDR4
Speed: 2400 MHz
Status: OK
Manufacturer: 0x80AD
Part Number: -
Serial Number: -
I have a vague memory that I simply filled the two empty slots with the OWC memory.
 


Ric Ford

MacInTouch
On my 2017 5K iMac, with the two 4Gbyte slot-protectors from Apple and two 16Gbyte cards from OWC, System Information shows pairs in adjacent slots.
I'd be curious to see Geekbench 4 results from your system to see if the alternate configuration performs any worse (or better) vs. the alternating configuration for the same RAM cards.

It looks like Geekbench 5 dropped memory tests (copy, latency and bandwidth) from its results data, which is disappointing:

 


My iMac 2015 has 1-terabyte fusion drive containing a small 24-gig SSD. Can I upgrade the internal SSD to a much larger SSD, making the internal hard drive for storage only as a separate partition?
 


I'd be curious to see Geekbench 4 results from your system to see if the alternate configuration performs any worse (or better) vs. the alternating configuration for the same RAM cards.
My results were:

Memory copy: 4800 (13.3 GB/s)​
Memory Latency: 6131 (70.6 ns)​
Memory Bandwidth: 3764 (20.1 GB/s)​
 


Ric Ford

MacInTouch
Memory copy: 4800 (13.3 GB/s)​
Memory Latency: 6131 (70.6 ns)​
Memory Bandwidth: 3764 (20.1 GB/s)​
So, for these two 2017 iMac 5K machines, both with 40 GB of RAM, we've got

ConfigurationAlternating:
4GB 16GB 4GB 16GB
Paired:
16GB 16GB 4GB 4GB
Memory Copy15.4 GB/s13.3 GB/s
Memory Latency72.3 ns70.6 ns
Memory Bandwidth26.5 GB/s20.1 GB/s

So, the big question is what specs does your iMac 5K have? As listed in the Geekbench results linked earlier, the one I tested has:
Intel Core i7-7700K @ 4.20 GHz​
1 processor, 4 cores, 8 threads​
 


Amazon disclaimer:
As an Amazon Associate I earn from qualifying purchases.

Latest posts