Where did you buy the memory modules? I can only find places with crazy markups or they are out of stock.
LGA 1700 Alder Lake "Servers"
- Thread starter tjk
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
Hey there, you can pick them up from this US store (US store name of PC-Canada.com) -> https://www.tech-america.com/item/crucial-32gb-ddr5-sdram-memory-module/mtc20c2085s1ec48ba1r?lan=en
$163.00 USD as of this post.
$226.00 CAD as of this post.
They have stock too.
Yes, I have indications from a few posts here and Windows itself that I am running in true ECC - 80bit with my UDIMMS - the most recent STH YouTube went in depth about ECC and RDIMM ECC, but based on the 80bits, I know I am good.
Posted a couple here indicating what HWInfo and Windows says.
For those curious about power, total system power draw (at the UPS) is 180W with all the hardware in there (HDDs, SSDs) (ignore the Sabrent, it's connected via USB (transferring data from my old server) - (UPS connected directly to my ASUSTOR NAS that is acting as a Network UPS server) - and it's drawing some power on it's own 28-34W or so.
System has been absolutely rock solid with this ram, no BSODs, no random crashes, has been hosting Valheim, Minecraft, 7DTD and CoreKeeper game servers with zero issues.
Cheers
Attachments
Last edited:
Likewise - just keep a note of caution on the UPS import fees / commercial goods invoice - I had a number transposed and almost paid $120 CAD import fees from UPS to import two $8 USD 70MM (XBOX size) fans
, contacted QuietPC and they immediately issued a new commercial invoice and got a refund from UPS.Ya, I should post back to ASUS Support to update their page, as the Micron ECC UDIMMs that I have bought are running like champs on my system, no issues whatsoever, and true 80bit ECC goodness.
Minimal tweaking on my part, not OC my system at all, here is total power draw with a light load - at the UPS level. (UPS connected directly to my ASUSTOR NAS that is acting as a Network UPS server) - and it's drawing some power on it's own 28-34W or so.
Attachments
Last edited:
Do you mean your server load by itself is 180W minus the 28-34W? So closer to 150W with all that hardware? If so, that’s great news for my planned build—seems like it should idle fairly close to where my W480 plus 10100E does, which would be amazing for having much more headroom.
Yes, without a Kill-A-Watt, I would suggest that I am consuming only this much power (146-150W) on average, as a note, when the server drops down to 800mhz, the system drops down to around 130W, so quite ideal for me.
I'm coming from a X10SDC-4C-TLN2F which was drawing around 86-95w on average with all the same equipment with a Xeon D-1521, again at the UPS readout level, so I am pretty pleased given the massive uplift in performance and capability.
I am still debating if I am going to let drives go to sleep or not, not sure if the head load / unload cycles should be as huge concern or not.
What does your W480 + 10100E consume in Watts?
have you tried Neofetch? I just typed in "alternative to HWINFO64 Linux" and found this -> https://alternativeto.net/software/neofetch/about/
GitHub - dylanaraps/neofetch: ️ A command-line system information tool written in bash 3.2+
️ A command-line system information tool written in bash 3.2+ - dylanaraps/neofetch
github.com
Cheers
Idles around 90-100W with 64GB DDR4 ECC, 6x 3.5” SATA, 2.5” SATA SSD, 2x NVME, 10G, and a HBA
Nice! so yes, you will likely see very similar running costs with a newer build.
I'll eventually look into power tuning as well, but didn't want to miss out on the top 5.1ghz clocks that this Core i5-13600K can hit. What I am overall curious about when it comes to this board is when ASUS and/or Intel will release some drivers to better support the W680 chipset, as it seems pretty sparse, so I wonder if I am leaving performance on the table with Server 2022 vs. let's say Win 11 or 10, but that being said, I suspect I'd get even more performance under Linux for certain tasks.
I did not. I will take a look. Thanks!have you tried Neofetch? I just typed in "alternative to HWINFO64 Linux" and found this -> https://alternativeto.net/software/neofetch/about/
GitHub - dylanaraps/neofetch: ️ A command-line system information tool written in bash 3.2+
️ A command-line system information tool written in bash 3.2+ - dylanaraps/neofetch
Cheers
Following up now that I have the W680 and it’s on my bench. With a 13600K, 32gb of DDR5 at 4800, 1 SSD, and a single 10Gb SFP+ NIC it idles between 25W and 27W. The x670 with the same components and a 7900x idled at 48-52W. So I’m really happy with the lower idle power. Load power is something else entirely though, LOL.
That's awesome, I was messing around with the CPU power settings, especially the PL2 to avoid going over the 185W barrier, but I am really liking your numbers for power usage! what are you using to measure power consumption?
Load power is absolutely something else indeed!
A smart plug with energy monitoring, so that’s at the wall. Running prime95 it peaks at 335W(!!!!), and then settles in at mid-to-high 200s after a few minutes. I haven’t tested it for longer than 10 minutes bc the cooler that’s on it isn’t really up to the task. Once I get it under the AIO I can test the longer-term load power.That's awesome, I was messing around with the CPU power settings, especially the PL2 to avoid going over the 185W barrier, but I am really liking your numbers for power usage! what are you using to measure power consumption?
Load power is absolutely something else indeed!
On power consumption... I am running stably at a PL1 of 225W with the Noctua NH-U12A CPU cooler, and the CPU's hottest cores are at 85 to 90C when running at 100% CPU and 100% on my 4080 GPU. Pretty good for air.
My entire workstation, excluding the monitors draws about 550W at full power, and about 100W at idle.
On the ECC memory front. I put in a ticket at Supermicro to get to the bottom of the 72 vs 80 bit width and Single vs Multi-bit ECC reporting in dmidecode -t memory. I will let everyone know what I find out. I WILL point out that with no ECC capability, there is NO "Error Correction Type" line put out in the dmidecode output.
My entire workstation, excluding the monitors draws about 550W at full power, and about 100W at idle.
On the ECC memory front. I put in a ticket at Supermicro to get to the bottom of the 72 vs 80 bit width and Single vs Multi-bit ECC reporting in dmidecode -t memory. I will let everyone know what I find out. I WILL point out that with no ECC capability, there is NO "Error Correction Type" line put out in the dmidecode output.
Last edited:
Here is my case request to Supermicro to ask about 2.0a and also about the ECC DDR5 72 versus 80...
Here is their response...
X13SAE
Hi James,
Thanks for choosing SMC. Appreciate your delightful feedback.
1. For the Release note of each firmware update, it depends on the product series. Some we do post the release note of the FW changes, but not this one. I'll feedback your advise but it is up to the management if the release note is to be released to the public. In this case, for BIOS 2.0a update, the major update was
a. updating Raptor lake microcode.
b. updating Intel RC code
c. updating ME
d. other bug fixes
2. For the DIMM MTC20C2085S1EC48BA1, it is a 72 bit width, you may refer to the link below for its spec and datasheet.
www.micron.com
It didn't specify whether this DIMM is a SEC (single ECC) or MEC (multi ECC) but from its SKU "S1EC48" in its suffix, it seems it is a SEC (S for single). You may check with Micron as we don't have its detail.
Let me know if you have further question.
| Question on difference between BIOS ver 2.0 and 2.0a and also ECC memory enable | |
| Description | This workstation is operating beautifully. Many thanks for a great MB design! I have two questions: 1. A change log was not posted for the difference between BIOS version 2.0 and 2.0a. When I did the original installation, I upgraded to 2.0 to enable the recognition of my i9 13900K processor. l would like to know what changed between 2.0 and 2.0a and whether it is worth flashing the new version. A suggestion... post the change logs for the BIOS so people know whether they should upgrade. In general my position is that BIOS should not be upgraded unless the change log indicates something that makes it worth upgrading. 2. I procured the Micron ECC DDR-5 modules (Micron Technology Part Number: MTC20C2085S1EC48BA1), since the Hynix was not available at the time. The modules operate fine, however there appears to be no indication about ECC status in the BIOS. See the attached log file for the output of dmidecode -t memory in Linux. Note that 72 bits is reported for the data width. Why is this not 80 for DDR5 true ECC? Also error correction is reported as Single-bit ECC rather than Multi-bit ECC. I inserted the Alderlake edac module, igen6_edac and it made no difference. Also ras-mc-ctl reports: jco-monster-ng:/lib/modules/5.14.21-150400.24.46-default/kernel/drivers/edac # ras-mc-ctl --layout ras-mc-ctl: Error: No memories found at via edac. What is going on with ECC support here? Is this a kernel version issue? OpenSUSE is very conservative with kernels. |
Here is their response...
X13SAE
Hi James,
Thanks for choosing SMC. Appreciate your delightful feedback.
1. For the Release note of each firmware update, it depends on the product series. Some we do post the release note of the FW changes, but not this one. I'll feedback your advise but it is up to the management if the release note is to be released to the public. In this case, for BIOS 2.0a update, the major update was
a. updating Raptor lake microcode.
b. updating Intel RC code
c. updating ME
d. other bug fixes
2. For the DIMM MTC20C2085S1EC48BA1, it is a 72 bit width, you may refer to the link below for its spec and datasheet.
MTC20C2085S1EC48B
www.micron.com
It didn't specify whether this DIMM is a SEC (single ECC) or MEC (multi ECC) but from its SKU "S1EC48" in its suffix, it seems it is a SEC (S for single). You may check with Micron as we don't have its detail.
Let me know if you have further question.
Wonder if it is worth updating the bios. I've not noticed anything but then again maybe I'm not really pushing it much. Mine is only serving up data and Plex.Here is my case request to Supermicro to ask about 2.0a and also about the ECC DDR5 72 versus 80...
Question on difference between BIOS ver 2.0 and 2.0a and also ECC memory enable Description This workstation is operating beautifully. Many thanks for a great MB design!
I have two questions:
1. A change log was not posted for the difference between BIOS version 2.0 and 2.0a. When I did the original installation, I upgraded to 2.0 to enable the recognition of my i9 13900K processor. l would like to know what changed between 2.0 and 2.0a and whether it is worth flashing the new version.
A suggestion... post the change logs for the BIOS so people know whether they should upgrade. In general my position is that BIOS should not be upgraded unless the change log indicates something that makes it worth upgrading.
2. I procured the Micron ECC DDR-5 modules (Micron Technology Part Number: MTC20C2085S1EC48BA1), since the Hynix was not available at the time. The modules operate fine, however there appears to be no indication about ECC status in the BIOS. See the attached log file for the output of dmidecode -t memory in Linux. Note that 72 bits is reported for the data width. Why is this not 80 for DDR5 true ECC? Also error correction is reported as Single-bit ECC rather than Multi-bit ECC. I inserted the Alderlake edac module, igen6_edac and it made no difference. Also ras-mc-ctl reports:
jco-monster-ng:/lib/modules/5.14.21-150400.24.46-default/kernel/drivers/edac # ras-mc-ctl --layout
ras-mc-ctl: Error: No memories found at via edac.
What is going on with ECC support here? Is this a kernel version issue? OpenSUSE is very conservative with kernels.
Here is their response...
X13SAE
Hi James,
Thanks for choosing SMC. Appreciate your delightful feedback.
1. For the Release note of each firmware update, it depends on the product series. Some we do post the release note of the FW changes, but not this one. I'll feedback your advise but it is up to the management if the release note is to be released to the public. In this case, for BIOS 2.0a update, the major update was
a. updating Raptor lake microcode.
b. updating Intel RC code
c. updating ME
d. other bug fixes
2. For the DIMM MTC20C2085S1EC48BA1, it is a 72 bit width, you may refer to the link below for its spec and datasheet.
MTC20C2085S1EC48B
It didn't specify whether this DIMM is a SEC (single ECC) or MEC (multi ECC) but from its SKU "S1EC48" in its suffix, it seems it is a SEC (S for single). You may check with Micron as we don't have its detail.
Let me know if you have further question.
Edit: thank you btw for getting the information on what it is, and the memory, and sharing.