LGA 1700 Alder Lake "Servers"

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Alex_T0000

Member
Aug 7, 2021
35
11
8
This is interesting since the same command only gives me “5” on 2 other systems that should be 100% ECC enabled (X10SLM with Xeon and C246 board with Xeon). Also the C246 system shows ECC “enabled” according to Memtest.

I was under the impression all ECC was only capable of correcting single bit errors and that multi bit referred to OS reporting of uncorrectable errors (extraordinarily rare).
Hmm, this is a question for some of the other home server and pro server members on here - I don't know the answer - but I did always have an expectation when building my X10SDV system that the core reason was so I could detect and correct errors, however, I always did have the impression that the OS and processor required this multi directional ECC support to perform some kind of "magic" and ensure that bits didn't flip as they were corrected and also reported to the OS.

You've got me curious too, but from other things I have read, I should have been expecting a "6" from WMIC, at least from my own computing experience and my rationale for ensuring I ECC this system to maintain what I have with my X10SDV system.

Anyone else reading the thread have insights on the Multi-bit aspect? I am going to go off and do some more reading, because I am curious.

Edit: are those Xeons that fit into X10SLMs supporting the same type of ECC as the Xeon D would (My Xeon D system uses ECC RDIMMs - is that the difference? I am thinking it is now) I guess I ask myself a question - are all Xeon's the same when it comes to ECC support? now off to do some reading, because I'm curious about this ECC stuff ..

Edit2: I also tried out another command on my X10SDV - which I have read elsewhere that needs to report back the following values as a validation that ECC is in fact running:

wmic MEMORYCHIP get DataWidth,TotalWidth
DataWidth TotalWidth
64 72
64 72

Interestingly enough, this is similar to what Kingston reports as the case for their ServerPremier (WS680 ACE/IPMI compatible modules) in their spec sheet.


DESCRIPTION SPECIFICATIONS KSM48E40BD8KM-32HM 32GB 2Rx8 4G x 72-Bit PC5-4800 CL40 288-Pin DIMM

Continued >> FEATURES Kingston's KSM48E40BD8KM-32HM is a 4G x 72-bit (32GB) DDR5-4800 CL40 SDRAM (Synchronous DRAM), 2Rx8, ECC, memory module, based on twenty 2G x 8-bit FBGA components.

On-Die ECC • x72 ECC (x36, 2 independent I/O sub channels) • 32 internal banks • Hard/Soft Post Package Repair • Sideband access with I3C/I2C • PCB: Height 1.23” (31.25mm) • RoHS Compliant and Halogen-Free

Unlike last time I did check Kingston - as I think stated by someone else here earlier it is now listed as compatible with the board - Memory for a ASUS - Pro WS W680-ACE / IPMI Motherboard - Kingston Technology
 
Last edited:

RolloZ170

Well-Known Member
Apr 24, 2016
5,483
1,663
113
I get the sense that the T series is silicon that is not only binned to consume lower power, but is actually very efficient silicon
thats definitely wrong. was shown with a 'K' model and a 'Z' motherboard adjusted to same base clock etc than the 'T' model.
its just the lower clocks because the frequency/power consumption chart/courve is not linear.
the 'T' models are more likely processors that can not get higher frequencies within the normal model specification.
furthermore all models may start init powered at full baseclock and this will overload the VRMs in a 35W system.
 
  • Like
Reactions: Alex_T0000

heromode

Active Member
May 25, 2020
380
203
43
Hello, i don't own any W680 mobo, ECC DDR5, or LGA 1700 CPU, but i run a Asus WS PRO C246-ACE mobo with a Xeon E-2186G CPU, and 2x16GB Mushkin Proline ECC udimm's, model MPL4E266KF16G28 .

So the hardware definitely should be ECC. Have not tested with memtest (because it requires reboot). Anyways i wanted to post my dmidecode output, just in case it would help anyone here that wants to compare outputs:

Code:
# dmidecode -t memory
# dmidecode 3.3
Getting SMBIOS data from sysfs.
SMBIOS 3.2.1 present.

Handle 0x0042, DMI type 16, 23 bytes
Physical Memory Array
        Location: System Board Or Motherboard
        Use: System Memory
        Error Correction Type: Single-bit ECC
        Maximum Capacity: 64 GB
        Error Information Handle: Not Provided
        Number Of Devices: 4

Handle 0x0043, DMI type 17, 40 bytes
Memory Device
        Array Handle: 0x0042
        Error Information Handle: Not Provided
        Total Width: 72 bits
        Data Width: 64 bits
        Size: 16 GB
        Form Factor: DIMM
        Set: None
        Locator: ChannelA-DIMM1
        Bank Locator: BANK 0
        Type: DDR4
        Type Detail: Synchronous
        Speed: 2666 MT/s
        Manufacturer: Avant
        Serial Number: 00000000
        Asset Tag: 9876543210
        Part Number: W722GU42J9266NA     
        Rank: 2
        Configured Memory Speed: 2666 MT/s
        Minimum Voltage: Unknown
        Maximum Voltage: Unknown
        Configured Voltage: 1.2 V

Handle 0x0044, DMI type 17, 40 bytes
Memory Device
        Array Handle: 0x0042
        Error Information Handle: Not Provided
        Total Width: Unknown
        Data Width: Unknown
        Size: No Module Installed
        Form Factor: Unknown
        Set: None
        Locator: ChannelA-DIMM2
        Bank Locator: BANK 1
        Type: Unknown
        Type Detail: None
        Speed: Unknown
        Manufacturer: Not Specified
        Serial Number: Not Specified
        Asset Tag: Not Specified
        Part Number: Not Specified
        Rank: Unknown
        Configured Memory Speed: Unknown
        Minimum Voltage: Unknown
        Maximum Voltage: Unknown
        Configured Voltage: Unknown

Handle 0x0045, DMI type 17, 40 bytes
Memory Device
        Array Handle: 0x0042
        Error Information Handle: Not Provided
        Total Width: 72 bits
        Data Width: 64 bits
        Size: 16 GB
        Form Factor: DIMM
        Set: None
        Locator: ChannelB-DIMM1
        Bank Locator: BANK 2
        Type: DDR4
        Type Detail: Synchronous
        Speed: 2666 MT/s
        Manufacturer: Avant
        Serial Number: 00000000
        Asset Tag: 9876543210
        Part Number: W722GU42J9266NA     
        Rank: 2
        Configured Memory Speed: 2666 MT/s
        Minimum Voltage: Unknown
        Maximum Voltage: Unknown
        Configured Voltage: 1.2 V

Handle 0x0046, DMI type 17, 40 bytes
Memory Device
        Array Handle: 0x0042
        Error Information Handle: Not Provided
        Total Width: Unknown
        Data Width: Unknown
        Size: No Module Installed
        Form Factor: Unknown
        Set: None
        Locator: ChannelB-DIMM2
        Bank Locator: BANK 3
        Type: Unknown
        Type Detail: None
        Speed: Unknown
        Manufacturer: Not Specified
        Serial Number: Not Specified
        Asset Tag: Not Specified
        Part Number: Not Specified
        Rank: Unknown
        Configured Memory Speed: Unknown
        Minimum Voltage: Unknown
        Maximum Voltage: Unknown
        Configured Voltage: Unknown
 
  • Like
Reactions: Alex_T0000

Alex_T0000

Member
Aug 7, 2021
35
11
8
thats definitely wrong. was shown with a 'K' model and a 'Z' motherboard adjusted to same base clock etc than the 'T' model.
its just the lower clocks because the frequency/power consumption chart/courve is not linear.
the 'T' models are more likely processors that can not get higher frequencies within the normal model specification.
furthermore all models may start init powered at full baseclock and this will overload the VRMs in a 35W system.
Interesting - so if I am understanding you right - I would be able to adjust a "K" model but only in a "Z" motherboard to be able run like a T model? do you know if that would work on a WS680 board however?

Appreciate this perspective for sure - I always had the assumption that these T models were being built to a total power consumption target like the mobile / embedded SKUs were - like this one that has very similar specs to the Desktop (and even has Iris Graphics) but uses even less power than the 13500T -> Intel® Core™ i5-1350PE Processor (12M Cache, up to 4.60 GHz) - Product Specifications | Intel too bad I can't put that into my WS680! oh well...
 

Alex_T0000

Member
Aug 7, 2021
35
11
8
Hello, i don't own any W680 mobo, ECC DDR5, or LGA 1700 CPU, but i run a Asus WS PRO C246-ACE mobo with a Xeon E-2186G CPU, and 2x16GB Mushkin Proline ECC udimm's, model MPL4E266KF16G28 .

So the hardware definitely should be ECC. Have not tested with memtest (because it requires reboot). Anyways i wanted to post my dmidecode output, just in case it would help anyone here that wants to compare outputs:

Code:
# dmidecode -t memory
# dmidecode 3.3
Getting SMBIOS data from sysfs.
SMBIOS 3.2.1 present.

Handle 0x0042, DMI type 16, 23 bytes
Physical Memory Array
        Location: System Board Or Motherboard
        Use: System Memory
        Error Correction Type: Single-bit ECC
        Maximum Capacity: 64 GB
        Error Information Handle: Not Provided
        Number Of Devices: 4

Handle 0x0043, DMI type 17, 40 bytes
Memory Device
        Array Handle: 0x0042
        Error Information Handle: Not Provided
        Total Width: 72 bits
        Data Width: 64 bits
        Size: 16 GB
        Form Factor: DIMM
        Set: None
        Locator: ChannelA-DIMM1
        Bank Locator: BANK 0
        Type: DDR4
        Type Detail: Synchronous
        Speed: 2666 MT/s
        Manufacturer: Avant
        Serial Number: 00000000
        Asset Tag: 9876543210
        Part Number: W722GU42J9266NA    
        Rank: 2
        Configured Memory Speed: 2666 MT/s
        Minimum Voltage: Unknown
        Maximum Voltage: Unknown
        Configured Voltage: 1.2 V

Handle 0x0044, DMI type 17, 40 bytes
Memory Device
        Array Handle: 0x0042
        Error Information Handle: Not Provided
        Total Width: Unknown
        Data Width: Unknown
        Size: No Module Installed
        Form Factor: Unknown
        Set: None
        Locator: ChannelA-DIMM2
        Bank Locator: BANK 1
        Type: Unknown
        Type Detail: None
        Speed: Unknown
        Manufacturer: Not Specified
        Serial Number: Not Specified
        Asset Tag: Not Specified
        Part Number: Not Specified
        Rank: Unknown
        Configured Memory Speed: Unknown
        Minimum Voltage: Unknown
        Maximum Voltage: Unknown
        Configured Voltage: Unknown

Handle 0x0045, DMI type 17, 40 bytes
Memory Device
        Array Handle: 0x0042
        Error Information Handle: Not Provided
        Total Width: 72 bits
        Data Width: 64 bits
        Size: 16 GB
        Form Factor: DIMM
        Set: None
        Locator: ChannelB-DIMM1
        Bank Locator: BANK 2
        Type: DDR4
        Type Detail: Synchronous
        Speed: 2666 MT/s
        Manufacturer: Avant
        Serial Number: 00000000
        Asset Tag: 9876543210
        Part Number: W722GU42J9266NA    
        Rank: 2
        Configured Memory Speed: 2666 MT/s
        Minimum Voltage: Unknown
        Maximum Voltage: Unknown
        Configured Voltage: 1.2 V

Handle 0x0046, DMI type 17, 40 bytes
Memory Device
        Array Handle: 0x0042
        Error Information Handle: Not Provided
        Total Width: Unknown
        Data Width: Unknown
        Size: No Module Installed
        Form Factor: Unknown
        Set: None
        Locator: ChannelB-DIMM2
        Bank Locator: BANK 3
        Type: Unknown
        Type Detail: None
        Speed: Unknown
        Manufacturer: Not Specified
        Serial Number: Not Specified
        Asset Tag: Not Specified
        Part Number: Not Specified
        Rank: Unknown
        Configured Memory Speed: Unknown
        Minimum Voltage: Unknown
        Maximum Voltage: Unknown
        Configured Voltage: Unknown
Thanks for sharing! that's interesting how all the DDR-5 ECC is reporting back Single-Bit ECC - which still suggests just the on-die but totally cool to be wrong in this case.

In the thread associated with the X13SAE review another person posted a cool build that was leveraging SuperMicro approved ECC with that board, so I asked the question about whether that person is seeing Multi-Bit ECC or Single-Bit..
 

RolloZ170

Well-Known Member
Apr 24, 2016
5,483
1,663
113
I would be able to adjust a "K" model but only in a "Z" motherboard to be able run like a T model?
yes. if a T model could go to that high frequencies than a K model, it will draw more power, for that reason its out-selected for T model use.
do you know if that would work on a WS680 board however?
no. can you adjust the baseclock ? can you adjust the turbo bins ?
I always had the assumption that these T models were being built to a total power consumption target
thats true. yes, with lower baseclock you set lower TDP.
the all core turbo is lower, the max. turbo is lower (and holds not as long because it must earlier return to 35Watts i.e.)
like the mobile / embedded SKUs were - like this one that has very similar specs to the Desktop (and even has Iris Graphics) but uses even less power than the 13500T -> Intel® Core™ i5-1350PE Processor (12M Cache, up to 4.60 GHz) - Product Specifications | Intel
the TDP is lower. means not uses less power at same operating points.
 
  • Like
Reactions: Alex_T0000

Kiririn

New Member
Apr 7, 2022
18
11
3
Update on this:
  • IPMI sensor/fan control crashes are pretty much 100% repro on S3 wake, but resolved by running 'ipmitool mc reset cold' from task scheduler on s3 wake. I would be amazed if this is not fixed in a future update with how easy it is to reproduce
  • Seeing an occasional failure (<10% repro) to resume from S3 sleep. Windows Hybrid Sleep keeps this from resulting in data loss, but not ideal. Possibly same root cause as the next point
  • Some SATA concerns - seems picky on SATA cable, and even after fitting a bend-free supermicro branded cable, still seeing occasional boot hangs on AHCI init (near 100% repro on previous known-good cable) and/or not recognising the attached drive. Based on another poster having issues with certain hard drives, I suspect some aspect of the SATA system is marginal
  • Fan control works the same as on previous Supermicro X10,X11,X12 boards (see https://forums.servethehome.com/index.php?resources/supermicro-x9-x10-x11-fan-speed-control.20/). I've added support for this IPMI fan control technique to LibreHardwareMonitor, hopefully coming to a public release of that and Fan Control soon
Further update:

The boot hangs I experienced were unrelated to "AHCI Init" / SATA, that is just what it shows on display but has hung on a different post code relating to memory. This is only visible on IPMI, the display post code freezes on a previous one

Resolved by disabling fast boot and MRC Fast Boot. Downside is a minute or so of extra POST time, but small price to pay for reliable booting. I think this is probably a side effect of using non-certified RAM (Kingston KSM48E40BS8KM-16HM)

Also no longer seeing any S3 resume failures. I'm not sure if this is due to the fast boot changes. I also enabled DeepSx Power Policies, upgraded to Windows 11 and enabled HVCI (purging many dodgy old drivers in the process)

Overall very pleased with the board now, no issues remaining, touch wood!
 

Alex_T0000

Member
Aug 7, 2021
35
11
8
yes. if a T model could go to that high frequencies than a K model, it will draw more power, for that reason its out-selected for T model use.

no. can you adjust the baseclock ? can you adjust the turbo bins ?

thats true. yes, with lower baseclock you set lower TDP.
the all core turbo is lower, the max. turbo is lower (and holds not as long because it must earlier return to 35Watts i.e.)

the TDP is lower. means not uses less power at same operating points.
Thanks for sharing and I don't know enough about the WS680 - since I am still trying to solve for ECC and awaiting availability (in Canada) for the Core i5-13500 (which I think I am feeling better about, based on everything you have shared - thank you!)

I can't wait to build and boot up this system - it's fun :) - I am also waiting for my SlimSAS to SATA cable to arrive.

Once the system is running, I'll be happy to share my learnings about the WS680 chipset, since I think it's really key for others going down this path to share their experiences.

Cheers and thanks for your comments and thoughts.
 

saf1

Member
Nov 27, 2022
55
10
8
Further update:

The boot hangs I experienced were unrelated to "AHCI Init" / SATA, that is just what it shows on display but has hung on a different post code relating to memory. This is only visible on IPMI, the display post code freezes on a previous one

Resolved by disabling fast boot and MRC Fast Boot. Downside is a minute or so of extra POST time, but small price to pay for reliable booting. I think this is probably a side effect of using non-certified RAM (Kingston KSM48E40BS8KM-16HM)

Also no longer seeing any S3 resume failures. I'm not sure if this is due to the fast boot changes. I also enabled DeepSx Power Policies, upgraded to Windows 11 and enabled HVCI (purging many dodgy old drivers in the process)

Overall very pleased with the board now, no issues remaining, touch wood!
I may have missed it but if you are using SATA drives what size? I've tried a few different cables after reading the post but still only able to see the 8 TB drives. 10 and 16 TB are non starters for me but yet visible on my older system.

Edit: outside the hba/lsi card and SATA drive issue at least the board used the Intel 13900K correctly :)
 

saf1

Member
Nov 27, 2022
55
10
8
power disable feature ?
I've not heard of that feature. Bios setting? Let me check real quick.

Edit: I didn't see it as an option. I did see spin-up under the respected port, tried that, no luck. Also should say that I had swapped out the drive it did see (8 tb) and replaced it with the 10 and 16 (using same cables) and it didn't see it. So known working cable and port with no luck.
 
Last edited:

saf1

Member
Nov 27, 2022
55
10
8
Crucial mx100 SSD (500gb), appeared intermittently in bios/os with an old blue gigabyte sata cable that worked perfectly on my old x10 motherboard, working reliably with a new supermicro branded sata cable (which honestly looks lower quality / wire gauge... very odd). A few UDMA CRC errors are logged in SMART, unfortunately don't have records of whether they were there before. Could be red herring / coincidence, wouldn't be surprised if it was somehow related to the memory-related boot issues I was having
Thanks for the info.
 
  • Like
Reactions: RedX1

apnar

Member
Mar 5, 2011
115
23
18
Thanks for sharing! that's interesting how all the DDR-5 ECC is reporting back Single-Bit ECC - which still suggests just the on-die but totally cool to be wrong in this case.

In the thread associated with the X13SAE review another person posted a cool build that was leveraging SuperMicro approved ECC with that board, so I asked the question about whether that person is seeing Multi-Bit ECC or Single-Bit..
So as a reference point I have a i9-12900K in an ASUS Prime Z690-P with non-ECC DDR5 memory, but obviously has the standard on-die ECC that all DDR5 has. I ran dmidecode just to see what it would look like:

Code:
dmidecode -t memory
# dmidecode 3.3
Getting SMBIOS data from sysfs.
SMBIOS 3.4.0 present.

Handle 0x0047, DMI type 16, 23 bytes
Physical Memory Array
    Location: System Board Or Motherboard
    Use: System Memory
    Error Correction Type: None
    Maximum Capacity: 256 GB
    Error Information Handle: Not Provided
    Number Of Devices: 4
So it looks like it doesn't recognize the standard on-die ECC at all. Maybe when it's saying error correction type "Single-bit ECC" that means it's working correctly. After all real ECC can only correct a single bit error even if it can detect a two bit error as well.
 

RolloZ170

Well-Known Member
Apr 24, 2016
5,483
1,663
113
I've not heard of that feature. Bios setting? Let me check real quick.
SATA 3.3 Power Disable Feature
some new big drives have a special power connector, the 3.3volt input is used to disable the drive.
if you put a power connector with 3.3v on it it will not spin up. if your old hardware has only 5V & 12V on the Sata power connectors it works, newer PSU have the 3.3v on all connectores and will not work.
 

saf1

Member
Nov 27, 2022
55
10
8
SATA 3.3 Power Disable Feature
some new big drives have a special power connector, the 3.3volt input is used to disable the drive.
if you put a power connector with 3.3v on it it will not spin up. if your old hardware has only 5V & 12V on the Sata power connectors it works, newer PSU have the 3.3v on all connectores and will not work.
Ignorance on my part as I've not heard of it or run into it before. My apologies. The 10 and 16 TB drives are working fine in my older X11SSH-CTF system. Would this explain why?

How could I test - would I need a different power supply or cable to try? I would be thrilled if it was that simple while also feeling pretty stupid I guess. I believe the power supply that I'm using currently is a Corsair RMx Series (2021), RM850x modular.

Thank you btw.
 

RolloZ170

Well-Known Member
Apr 24, 2016
5,483
1,663
113
How could I test - would I need a different power supply or cable to try?
just check if the SATA power includes the orange 3.3V voltage. if yes test with MOLEX->Satapower cable which usualy has no 3.3V because not provided by the 4 Pin molex.
 

RolloZ170

Well-Known Member
Apr 24, 2016
5,483
1,663
113
I would be thrilled if it was that simple while also feeling pretty stupid I guess. I believe the power supply that I'm using currently is a Corsair RMx Series (2021), RM850x modular.
do you use in both cases the SAME PSU and same SATA Power cables ? IF YES your issue is not this !!
 

saf1

Member
Nov 27, 2022
55
10
8
do you use in both cases the SAME PSU and same SATA Power cables ? IF YES your issue is not this !!
No. The X11SSH-CTF uses a Sea Sonic SSR-750RM power supply if I remember correctly. I took a quick look at the current power supply I'm using (Corsair RM850) and the SATA cable has 5 wires. Not sure if that helps. I will look to see if I have a MOLEX to SATA adapter.
 

heromode

Active Member
May 25, 2020
380
203
43
No. The X11SSH-CTF uses a Sea Sonic SSR-750RM power supply if I remember correctly. I took a quick look at the current power supply I'm using (Corsair RM850) and the SATA cable has 5 wires. Not sure if that helps. I will look to see if I have a MOLEX to SATA adapter.
easiest way is to hold your fingers on the hdd as you power on the system, you will feel if they spin up or not. Next easiest way is molex to sata adapter cable.