Supermicro X11SCA X11SCA-W and X11SCA-F for Xeon E-2100 Workstations

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

SDLeary

Member
Aug 4, 2015
48
9
8
56
Patrick,

In the article, its stated that only the -F has a BMC with IPMI, and this is supported by the pictures. However, in the spec listings, each is listed as having the Aspeed 2500.

Thanks Much
SDLeary
 
  • Like
Reactions: Patrick

Krobar

Member
Aug 25, 2012
54
10
8
Pre-ordered one of these a few weeks ago and will share experiences once it arrives. (Probably going I3 so no need to wait for CPU)
 

jfoor

Member
Feb 4, 2017
81
20
8
36
Why do these boards have a beige legacy PCI slot? I didn't know if there was some reason I'm just not thinking of or if they have a lot of people trying to run older cards in these workstation boards.
 

fsck

Member
Oct 10, 2013
51
12
8
Maybe people running antique hardware where all the documentation has been lost?

PCIE controllers exist for GPIB and it came out in the 1970s. Also USB dongles exist, but that's a different issue.
 

Mam89

Member
Jan 14, 2016
58
11
8
34
SoCal
Believe it or not, I've worked with clients that had pci modems (passed through to vms) just so they could retain their windows fax servers.... When the VoIP systems we were installing gave the same if not better functionality.

Aka, change is scary
 

niekbergboer

Active Member
Jun 21, 2016
154
59
28
46
Switzerland
Nice Patrick! I am particularly interested in the idle power of a small server based on that platform - in particular after my great experience with Skylake in that regard.
 

zir_blazer

Active Member
Dec 5, 2016
356
128
43
Why do these boards have a beige legacy PCI slot? I didn't know if there was some reason I'm just not thinking of or if they have a lot of people trying to run older cards in these workstation boards.
The Supermicro X10SAT (Xeon E3 v3 Haswell) I am using is pure PCIe, and was intended for Workstation. It has an identical PCB to the consumer C7Z87-OCE, which was released like 6 months before the X10SAT (Ironically, my X10SAT has C7Z87-OCE imprinted on the PCB) since for it, Supermicro waited for the C2 Stepping of the 8 Series Chipsets with the fixed USB Sleep errata. The X10SAT has a similar model, the X10SAE, which is regular green PCB with legacy stuff.
After the X10SAT, I don't recall seeing a Xeon E3 Motherboard that is a worthy successor to the legacy-free X10SAT, they are instead similar to the X10SAE. They did a X10SAT-like Motherboard for LGA 2066, the X11SRA-F, which has similar PCB style (But is not the same) than the consumer C9X299-RPGF.

Sadly, if I were to buy a new Xeon E3 I couldn't decie on how to pick a Motherboard simply because I can't stand legacy PCI Slots...
 

Patrick

Administrator
Staff member
Dec 21, 2010
12,512
5,800
113
@Krobar - you can fit M.2 22110 in these boards using the supplied adapter. There are spots for 2x.
 

Krobar

Member
Aug 25, 2012
54
10
8
@Krobar - you can fit M.2 22110 in these boards using the supplied adapter. There are spots for 2x.
Sure the card is actually a 2280 and one I had left over which I will use a boot drive. The little instruction cards that come with the mounting plastic for the M2s are somewhat confusing and its easier to eyeball it; if you are using 2280 size then you only need the little plug as shown in the picture, for all the other sizes you need the bigger lump of plastic.

For those wondering about PCI-E overlap:
M2_1 shares with U.2.
M2_2 shares with the PCI-E X4 slot

My intended use for this is a NAS and some KVM virtualisation. I'm hoping that Supermicro get SRIOV and IOMMU groups right first bios revision these days :)

@Patrick I'm using an I3 with ECC; guessing you will be reviewing with higher end chips (?)
 

Krobar

Member
Aug 25, 2012
54
10
8
Quick FYI. The X11SCA-F seems to suffer from some PCI Passthrough bugs at present in the 1.0 bios (Broken IOMMU groupings). Its sounding like Supermicro support will fix it but right now this board is not a good choice for virtualization.
 
  • Like
Reactions: niekbergboer

zir_blazer

Active Member
Dec 5, 2016
356
128
43
Quick FYI. The X11SCA-F seems to suffer from some PCI Passthrough bugs at present in the 1.0 bios (Broken IOMMU groupings). Its sounding like Supermicro support will fix it but right now this board is not a good choice for virtualization.
Post the IOMMU Groups. Is hard to get them wrong these days...
 

zir_blazer

Active Member
Dec 5, 2016
356
128
43
Attached; look broken to me although I'm no expert.
Group 9 is broken. Intel Chipsets since Skylake have a rather innofensive ACS related errata for the Chipset PCIe Root Ports that requires a Linux Kernel workaround, which is applied on a per-PCIID basis. Since Xeons C246 Chipset is new, chances are than its PCIID is not included when applying the errata workaround. If it is so, it isn't Supermicro related at all and should affect all C246 based Motherboards.
I can batcall the main VFIO developer to check into that, he is usually on Reddit VFIO subreddit. It may be possible that you have to recompile the Linux Kernel with his fix to check if it works.


BTW, just in case, drop also a full lspci -vvvnn. According to the quirks file...

Per the datasheet[5], root port PCI Device IDs for this* chipset include:
This works based on the PCIID of the PCIe Root Ports themselves. Since Intel seems to not have released a Data Sheet for the C246 Chipsets (Intel did released a Data Sheet for most of the other 300 series Chipsets except Z390), you need to get info by yourself about what the PCIID target is. The same series usually have the same PCIID for the Root Ports, so unless the C246 uses different ones than the rest of the 300 series it should already been workarounded.

Also, for good measure, include a lspci -xxxx of the PCIe Root Ports so you get the PCI Configuration Space, check here. That one would actually be Supermicro fault if they are not exposing ACS correctly.

Finally, what distribution and Linux Kernel are you using? If it is too old, it may be possible that they do not include the quirks for the Coffee Point Chipset series.
 
Last edited:

Krobar

Member
Aug 25, 2012
54
10
8
Thanks for the reply. I'm using Arch with a pretty new 4.17 kernel. I will need to get the info you mention tonight (Am at work currently). I'm happy to patch the kernel if I have to (I tend to find this easier with arch as it is so close to Vanilla).

dmesg is complaining of ACPI bugs with various ones similar to this:
[ 0.211889] ACPI: 11 ACPI AML tables successfully acquired and loaded
[ 0.218872] ACPI BIOS Error (bug): Could not resolve [\_SB.PCI0.LPCB.SPMI._STA.OSFL], AE_NOT_FOUND (20180313/psargs-330)
[ 0.218879] ACPI Error: Method parse/execution failed \_SB.PCI0.LPCB.SPMI._STA, AE_NOT_FOUND (20180313/psparse-516)
[ 0.223962] ACPI: Dynamic OEM Table Load:

Do you think its related?
 

Krobar

Member
Aug 25, 2012
54
10
8
@zir_blazer Big thanks for your help on this. Have not found a solution yet but always interesting to learn.

The quirks file includes ids as below for 300 series:
0xa32c ... 0xa343

I think this is the device root port:
lspci -s 00:1d.0 -nn
00:1d.0 PCI bridge [0604]: Intel Corporation Cannon Lake PCH PCI Express Root Port 9 [8086:a330] (rev f0)

So an ID of a330 is already covered by the patch and the patch appears to be present in the 4.17.9 kernel (4.17.9-1-ARCH) I'm using:
quirks.c\pci\drivers - kernel/git/stable/linux.git - Linux kernel stable tree
Here is the PCI Info:
PCI Info - Pastebin.com