Supermicro X7SPA L/H/HF ATOM serverboards (Level 1 Tested)


Recommended Posts

  • Replies 397
  • Created
  • Last Reply

Top Posters In This Topic

Mine reason for looking at the Zotac is that Supermicro boards are almost impossible to get in Denmark.

 

While the information on the Zotac board is helpful and worthy of discussion, please start a new thread on the Zotac boards, reference each others thread with hyperlinks.

This thread was started for information regarding the Supermicro Atom ITX boards.

I would rather see the pros listed in each respective thread then a debate of one vs the other in a thread dedicated to a specific board.

 

If y'all think a debate is worth of discussion, I would suggest a Supermicro vs Zotac vs whatever thread and have each reference the specific motherboard thread for details along that line.

 

It's all good information, I'm only suggesting keeping the topics a little more separate.

 

 

Link to comment

I was trying to steer the thread without being so obvious.  Your way is more adult.  ;D

 

I figure a direct approach is easier. Please do not mistake a direct approach for a judgment on friendliness. This is a good community.

I really want to encourage the discussion, just keep it all positive and on topic.

As I said, it's all good information, the Zotac is worthy of mention and discussion.

Just as the Supermicro was within the other thread and then forked into a clear path topic.

Link to comment

All I can say is IPMI is awesome.  It works flawlessly on this board, it's intuitive, and easy to set up.  You don't even have to install the IPMIview software to see it, as you can enter the IP address in a browser and get to it that way, assuming the box is running.  I checked out the few threads about switching over to eth1, and I have to say, I'm just completely confused.  The IPMI setup in the BIOS lets you specify a static IP address to the BMC, which is only available on eth0.  So I want to switch unRAID's load over to eth1.  ???

Link to comment

Got an AmWatt tester to play with over the weekend.  I'll be testing power consumption with no drives, then I will add the drives and test again.  Since both drives are identical in make and model, it should be easy math to get a figure on just how much power the system will use when fully expanded.  Did I mention IPMI is awesome?

Link to comment

Just got a pm question, and not sure how to answer it:

 

Can you add port multiplers to that motherboard directly to the onboard SATA?  I was just wondering if you use that motherboard and you want 20 + 1 drives, will it be possible?

 

Anyone here have any experience adding port multipliers on Supermicro SATA ports?

I know you could use a 16 port PCI-E card on the x4 slot and still get respectable throughput (~62.5 MB/s), but that may not be a sexy (or cheap) solution.

 

 

Link to comment

Just got a pm question, and not sure how to answer it:

 

Can you add port multiplers to that motherboard directly to the onboard SATA?  I was just wondering if you use that motherboard and you want 20 + 1 drives, will it be possible?

 

Anyone here have any experience adding port multipliers on Supermicro SATA ports?

I know you could use a 16 port PCI-E card on the x4 slot and still get respectable throughput (~62.5 MB/s), but that may not be a sexy (or cheap) solution.

 

 

Supposedly the ICH9 and ICH10 can support PMP.  However I think the support is not up to par yet.

The SIL3132 and SATA_MV drivers support PMP with the silicon image controllers being reliable.

 

Has there been 16 port PCIe cards which are validated to work in unRAID?

 

I'm not so sure this mITX board is the best choice for a large 20 drive system.

A small power wise environment yes. But a larger array should probably go with a different board with more options.

Link to comment

Whats the X factor that makes the Supermicro X7SPA  less attractive for a 20 disk unRAID?

 

Probably the lack of expandability options versus the next sized board.

 

Exactly. It has an x4 slot that can fit up to a x16 sized card, but it's still x4.

Supermicro's other flex atx seems more suitable and may fit better in a large case capable of handling 20 drives.

Link to comment

Hm... I'm not sure where you're getting that.  The 6 on board SATA ports have a nice bit of bandwidth available to them, and the x4 slot with an 8 port card has plenty of overhead as well.  I can tell you the parity rebuild I ran averaged 58 MB/s.

 

I agree here. I think the upper limit of 12-14 drives is good.

I'm sure you can jam more drives in with the right card & right kernel modules, but then it becomes an issue of time & cost. What you save in one place you spend in another I.E. a more dense controller which may not be as cost effective.

 

In comparison the Supermicro Flex ATX card could handle a 4x card and a x8 card.

With the on board ports you can have a pretty dense server without too much of a performance penalty.

 

With port multipliers, you "may" be able to fit more, but I think there will be bandwidth issues downstream.  If you stagger (Interleave) the PMP drives, that bandwidth issue may be diminished.

 

It's really about putting too many eggs in one basket (controller card). If you do find a compatible 16 port card.

What do you think the cost will be?

Then consider what will the cost be to replace it?

 

Link to comment

This post has been modified to remove incorrect power measurements done by me with a faulty meter.  I have instead posted Limetech's measurements, which are of course much more accurate.

 

Here are some preliminary numbers for the X7SPA/HF with 4GB DDR2-667 RAM (2 sticks of 2GB).

 

All measurements taken with Kill A Watt EZ.

 

Power supply is PC Power & Cooling Silencer 470 ATX - this is only a 73% efficient PSU, better numbers should be obtained using an 80+ PSU.

 

With only the motherboard connected to the PSU:

27W - During linux boot.

25W - At linux console

28-30W - Running memtest (reading bounces around depending on test).

 

With motherboard and Supermicro AOC-SASLP-MV8 controller plugged in the PCI-E x4 slot:

36W - During linux boot.

34W - At linux console.

 

With above configuration & 12 hard drives (see below for hdd types):

137W - Idle (drives spinning but no I/O taking place)

160W - Parity sync in process

53W - All drives spun down

304W - Highest observed reading during spinup of all drives

Link to comment

After boot, both NICs active, unRAID installed, drives spun down

40 W

 

I would have expected less.. *sigh*

Actually my Chenbro with Mobile 2.4Ghz duo core 2 with 4 drives is pretty much the same power usage.

 

The machine with the mobile celeron is around 35w idle.

 

I wonder if one of the via boards would get lower.

 

 

Still this is a really nice board and I'll probably buy one for my proxy server.

Link to comment

Well, I'm sure the BMC chip is doing some background hogging.  Oh, and the video card is enabled.  I wonder if I can disable it and still use IPMI?  All I can say is _I_ am pleased and I *LOVE* IPMI.  

 

Does it matter that I have it plugged into a UPS, even though the watt meter is sitting between the tower and the UPS?

 

EDIT: I just checked, and even with the system completely turned off and unplugged, the meter still shows 6 W?  Perhaps it's not well calibrated, or there is some other background thing going on.  Should I readjust the numbers to reflect that background reading?  Please forgive my ignorance, as I have never actually done power metering like this before.  :-[

Link to comment

The UPS pulls some power for it's own power supply, conversion and to charge the battery.

 

Put the power meter after the UPS to get an accurate measurement on the system alone.

Put the power meter before the UPS to get an accurate assessment of total power usage.

 

It's a great board. I'll get one as soon as I get rid of a few spare boards I have laying around.

Link to comment

The UPS pulls some power for it's own power supply, conversion and to charge the battery.

 

That's what I assumed, so I...

Put the power meter after the UPS to get an accurate measurement on the system alone.

 

But even with the meter just hanging on the wall by itself, it measures 6 W.  So is it safe to assume it is measuring 6 W above actual?

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.