I just got my hands on a pair of 32GB SODIMM memory modules (64GB total) which I had been waiting to evaluate since last Fall. Apparently, it has taken some time for these high capacity memory modules to be readily available in the consumer market. Even after the announcement of the new 2018 Apple Mac Mini last year, which officially supports 32GB SODIMMS, I was not aware of any vendors who were selling these modules direct to consumers.
My primary interests in these memory modules was whether they would work on the latest Intel NUCs, specifically the Hades Canyon (NUC8i7HNK) which are the prosumer versions of the standard Intel NUCs that many folks use for vSphere Home Labs. Both the standard and Skull/Hades Canyon NUCs all officially support a maximum of 32GB of memory (2x16GB SODIMM), however it been hypothesized by the community that they *should* in theory be able to go up to 64GB, especially as some of the newer CPUs technically state support for it.
With the lack of 32GB SODIMM availability and the cost, there has not been anyone that I am aware of that has confirmed whether the Intel NUCs can see the full 64GB of memory and more importantly, does ESXi also see the full memory capacity? With help from MITXPC on providing the evaluation units, I can finally answer this question which I actually shared on Twitter earlier this afternoon 🙂
YES! It works 😂
— William Lam (@lamw) March 13, 2019
Here are some screenshots of ESXi 6.5 Update 2 running on the Hades Canyon NUC with the 64GB memory modules:
I figured that if this would work, the Hades Canyon had the best chance as it was the latest generation of the Intel NUCs. What came next was a complete and unexpected surprise and hence the tweet. I still own and frequently use my Intel 6th Gen NUC (NUC6i3SYH), which was actually my first experience with the NUCs. I was curious if the memory modules would even be detected with such an old NUC, especially given this platform had already been discontinued.
UPDATE (04/20/19) - Here are additional testimonials from folks in the community who have also verified 64GB memory support with their respective NUCs.
- Intel 6th Gen NUC - Me! (see section below)
- Intel 7th Gen NUC - Fellow colleague Christian Loerner has shared he's had success as well!
- Intel 8th Gen NUC - See here, here, here, here
- Intel Skull Canyon NUC - See here, here
- Intel Hades Canyon NUC - Me! (see section above), See here
As you can see from the screenshots below, my old 6th Gen Intel NUC (purchased in 2016) recognized all 64GB memory running the latest ESXi 6.7 Update 1 release! This means anyone with at least a 6th Gen Intel NUC or newer can definitely benefit from these new 32GB SODIMM modules. From a setup standpoint, there was nothing special I needed to do. The Hades Canyon was running a fairly new BIOS but my old NUC was running a pretty old version 044 vs. 068 and it did not have any issues. Once I confirmed that it successfully posted, I ended up flashing the BIOS to the latest version which you can find on Intel's website and its a good practice, especially if you have taken the downtime.
This is a pretty freaking amazing if you think about the tiny footprint of a classic Intel NUC and combining that with M.2 NVMe for storage and now the ability to go up to 64GB of memory! Best of all, you can have these benefits today without having to purchase a brand new system, which many assumed would be required if/when Intel officially supports 64GB for their NUC platform.
For the few still in doubt, yes I can actually consume all the available memory. I was able to deploy 17 x Windows 10 VMs, each configured with 1 vCPU and 4GB of memory to demonstrate memory overcommit as not all memory is active. Obviously, there will be some amount of memory that will not be consumable, for example if you are using vSAN, ESXi will require a portion of the memory to run the service. This is no different than any other system you are managing today 🙂
In addition, a number of you have also asked if you *could* mix a 16GB and 32GB module, not recommended for obvious reasons, but it looks like the system will also recognize that configuration for those wanting to iteratively update their existing Intel NUCs.
Finally, I am sure folks are wondering about what this will cost them. Currently, on Amazon a single Samsung DDR4 32GB SODIMM is going for $298. Interestingly, I also recently came across a Slickdeal (one of my favorite sites to visit) and they have this exact module for just $213! For a little over $1K USD, you can have a pretty nice vSphere/vSAN/NSX Home Lab kit that can run next to you with very little no noise at all.
Great news, trank you for testing! Any idea if it will work on a Bean Canyon Nuc (i.e. Nuc8i5beh), too?
He said “anyone with at least a 6th Gen Intel NUC or newer can definitely benefit”. You have an 8th gen.
There’s been a number of folks who’ve verified on 8th Gen NUCs (Bean Canyon/etc). Take a look at the updated blog post w/references
Thanks for testing. I have 8x 6th gen. Will upgrade them as needed. Nice to know.
My Skull Canyon NUC cluster is very happy you discovered this. THANK YOU!
Thanks for the great test. Whats with temperature of the NuC?
I would definatly go for 64 gb, but Im wondering, why Intel dont official supprt 64 gb, what are the unknwn negative sideeffects?
Does anyone will made some 24/7 tests and repot back in a few days?
On a German memory shop, the spec for the system is written with 64 gb ram:
https://translate.google.de/translate?sl=auto&tl=en&u=https%3A%2F%2Fwww.speicher.de%2Farbeitsspeicher-16gb-ddr4-intel-nuc-kit-nuc8i7hnk2-ram-so-dimm-sp282859.html
The RAM of the Intel can be expanded to a maximum of 64GB . * 2 memory sockets are available for this purpose.
One question: must be both memory sockets setup with the same size, or are 16 gb + 32 (total 48 gb) supported, so that the upgrade can be done in steps?
May or may not work. The benefit of using the same size memory DIMM or SO-DIMM is that you get better performance because of memory interleaving.
My Skull Canyon Locks up with 64GB mushkin RAM
This is a great share. I just upgraded my 3 NUCs with 64GB after reading this last week. Thanks for the great work you do.
Hi Norman,
can you please share which NUC Models you have and how the new RAM modules are working?
Thanks,
Michele
Hi, I am looking to buy this RAM kit in UK. Can you share where did you buy this from please?
Hi Matheen found them on “scan computers uk”. Here you are the link: https://www.scan.co.uk/products/32gb-1x32gb-samsung-ddr4-so-dimm-laptop-sff-memory-pc4-21300-2666-non-ecc-unbuffered-cas-19-19-19-12
Thank you so much for testing this and sharing your results! I was wondering once they were generally available if they would work in the Hades Canyon NUC, and now I know.
William , thank you very much. This was the info, I´m locking for a long time.
Now our ESXi Server with 64Gb(mandatory) is set.
Kind regards Pascal
Great stuff William! good surprise for sure 🙂
@all
Has anyone tested these with NUC 7i7DNHE? In theory they should be working.
Only glitch would be these are 2666 MHz and NUC officially support 1866, 2133 and 2400 MHz. So these will not run at full speed but at 2400 max. Still an impressive 64GB of “space” to use. Very tempted to buy them for my NUCs, 8 of them now..
I’ve upgraded my Hades Canyon NUC HVK to 64GB ram. Unfortunately, I’m experiencing stability issues with apps like Adobe Photoshop. I work with 3-4 GB psb files… Swapped back to 32GB of Corsair 2666MHz and no issues. I do overclock my NUC though so I was wondering if anyone has had any issues?
NUC6i5SYH with latest GK0066 BIOS. Works! Memory is Samsung 32GB DDR4 2666MHz RAM Memory Module for Laptop Computers (260 Pin SODIMM, 1.2V) M471A4G43MB1. Setting up a new NUC8i7BEH with the same memory next week.
Update.. Also running 64GB in my newer NUC8i7BEH. Ok with Windows 10 and vSphere 6.7.
Thanks for the tip! The prices have gone down quite a bit, at the moment I see these modules at around $170.
bought two samsung 32gb for $148ea yesterday for a new primeday nuk8i7HVK, big price drop since this was posted
Hi !
Many people talking about NUC. I’m existing to talking about others closing hardware choice as Shuttle device. You can define CPU and memories as well. In my opinion cheaper than Intel.
See my post about it, It’s works on Proxmox or VMware 6.7.
https://translate.google.fr/translate?hl=fr&tab=wT0&sl=fr&tl=en&u=https%3A%2F%2Fwww.extein.fr%2Fsujet%2Fshuttle%2F
If this can’t help somebody… 🙂
Confirming the Samsung 32GB works,and with latest BiOS (July 2019), the NUC automatically supported 2666 speed
confirming Shuttle DV310v2 i5-9400 is working as an esxi hypervisor with samsung dual 32gb = total of 64gb ram
also nuc6i7kyk, nu8i5beh, nuc6i5syk
Installed 2 x 32 GB in a NUC8i3BEH. Works fine !
@Michele: did you test that actually? I’m thinking of upgrading mine to 64GB
Greets
@Michele: did you test that actually (upgrading the NUC7i7DNHE to 64GB)? I’m thinking of upgrading mine to 64GB
Greets