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

[GaryMaster]

 

No version of the X7SPA require the aux power connector; the connector is not even present on the board.

 

That's very odd...  The online manual shows its location on page 1-7 (Item 51) and explains you must plug it in on page 2-16:

 

http://www.supermicro.com/manuals/motherboard/ICH9/MNL-1166.pdf

 

Someone at Supermicro must have screwed up the manual.  I am still curious if the Chenbro has the 4 pin CPU connector since my new H55 ITX board most certainly needs it.

 

By the way, I think a small 4 bay mini-itx NAS would make a nice addition to your retail offerings...

 

 

[purko]

A big one, based on this board, that's really exciting news!

 

[tractng]

2) I assumed this board was made for a 1U case? I saw a link at supermicro for a suggested case.

I already ordered the chenbro ES34069 case which isn’t a 1U case but has very limited height from the pictures.  I am doing comparison just to get an idea.

 

Have you attempted to put the Supermicro X7SPA HF together with the Chenbro ES34069 yet?

 

I was reading the documentation for the Supermicro and Chenbro and it appears this Supermicro board requires a 4 pin CPU power connection and there is no mention of the Chenbro power supply having one.  

 

 

I will verify it tonight.  But I did remember seeing it when I first opened the case . Motherboard hasn't been shipped yet from wirezone (they told me it will be drop ship from SM).

 

 

 

tnt

[GaryMaster]

Have you attempted to put the Supermicro X7SPA HF together with the Chenbro ES34069 yet?

 

I was reading the documentation for the Supermicro and Chenbro and it appears this Supermicro board requires a 4 pin CPU power connection and there is no mention of the Chenbro power supply having one.  

 

 

I will verify it tonight.  But I did remember seeing it when I first opened the case . Motherboard hasn't been shipped yet from wirezone (they told me it will be drop ship from SM).

 

 

Great!  Another user stated that they thought it was in there, but I would appreciate your confirmation.  Attached is a picture of the 4 pin CPU atx connector I am looking for.  The manual only shows the existance a 4 pin MOLEX connector (I could buy an adapter, but would rather use a direct connection).

 

[WeeboTech]

Great!  Another user stated that they thought it was in there, but I would appreciate your confirmation.  Attached is a picture of the 4 pin CPU atx connector I am looking for.  The manual only shows the existance a 4 pin MOLEX connector (I could buy an adapter, but would rather use a direct connection).

 

It might be under the tray by the dc converter board.

[limetech]

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

 

Drive types (note: these are not "green" drives):

Hitachi HDS72101 x8

Seagate ST31500341AS x2

Seagate ST31000340AS x2

 

Parity Sync Rate:

44,877 KB/sec (K = 1024) => 46MB/sec (M = 1,000,000) => over 550MB/sec aggregate bandwidth

 

Parity Check Rate:

51,300 KB/sec => 53MB/sec => 630MB/sec aggregate bandwidth

 

So, yeah, I'd say this combination works pretty well 

 

[Chris Pollard]

Hmm.  Need to find a UK supplier for this board.

[GaryMaster]

With only the motherboard connected to the PSU:

27W - During linux boot.

25W - At linux console

 

Thanks for posting the power data! 

[tractng]

Have you attempted to put the Supermicro X7SPA HF together with the Chenbro ES34069 yet?

 

I was reading the documentation for the Supermicro and Chenbro and it appears this Supermicro board requires a 4 pin CPU power connection and there is no mention of the Chenbro power supply having one.  

 

 

I will verify it tonight.  But I did remember seeing it when I first opened the case . Motherboard hasn't been shipped yet from wirezone (they told me it will be drop ship from SM).

 

 

Great!  Another user stated that they thought it was in there, but I would appreciate your confirmation.  Attached is a picture of the 4 pin CPU atx connector I am looking for.  The manual only shows the existance a 4 pin MOLEX connector (I could buy an adapter, but would rather use a direct connection).

 

 

 

Here are the pics:

 

It does have the 4 pin CPU. The main connector has a 20+4 connectors as well.  Btw, great case.  I took out the cd rom holder in the pic. Waiting for the board to arrive.

 

tnt

 

 

 

 

[GaryMaster]

It does have the 4 pin CPU. The main connector has a 20+4 connectors as well.  Btw, great case.  I took out the cd rom holder in the pic. Waiting for the board to arrive.

 

tnt

 

tractng-

 

Thanks so much for verifying and taking the time to post pictures. 

 

I have spent a number of hours looking around at small cases and there is no question that this is the best 4 bay system available right now. 

 

I really look forward to building it up with the new Mini ITX Zotac H55ITX-A-E motherboard I have on the way!  Part of me regrets the 9-10W of idle power and fanless operation I'm going to lose compared to this supermicro, but the performance headroom for running other applications (and wicked-fast startup time) is something I really value.

[WeeboTech]

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

 

...

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

 

Hey any note this controller?  Woo Hoo.... Looks like we have a new cost effective 8 port PCIe controller for unRAID!!!!!

 

 

Parity Sync Rate:

44,877 KB/sec (K = 1024) => 46MB/sec (M = 1,000,000) => over 550MB/sec aggregate bandwidth

 

Parity Check Rate:

51,300 KB/sec => 53MB/sec => 630MB/sec aggregate bandwidth

 

Tom, how were the drives configured or laid out?

I would have expected slightly faster times then this.

 

I wonder if interleaving drives between the controllers would help.

ya know, disk1, motherboard, disk2  PCie controller, etc, etc.

 

[WeeboTech]

It does have the 4 pin CPU. The main connector has a 20+4 connectors as well.  Btw, great case.  I took out the cd rom holder in the pic. Waiting for the board to arrive.

tnt

.....

 

Remember to make a post in the pimp my right thread.

Is it worthwhile to create a thread based on this case alone? It's a great case I've been using for years.

In fact I wrote a blog about it a while back.

[WeeboTech]

I really look forward to building it up with the new Mini ITX Zotac H55ITX-A-E motherboard I have on the way!  Part of me regrets the 9-10W of idle power and fanless operation I'm going to lose compared to this supermicro, but the performance headroom for running other applications (and wicked-fast startup time) is something I really value.

 

I posted a new thread regarding this board

http://lime-technology.com/forum/index.php?topic=5479.0

I would love to know more about the wicked fast startup times.

[NAS]

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

 

Drive types (note: these are not "green" drives):

Hitachi HDS72101 x8

Seagate ST31500341AS x2

Seagate ST31000340AS x2

 

Parity Sync Rate:

44,877 KB/sec (K = 1024) => 46MB/sec (M = 1,000,000) => over 550MB/sec aggregate bandwidth

 

Parity Check Rate:

51,300 KB/sec => 53MB/sec => 630MB/sec aggregate bandwidth

 

So, yeah, I'd say this combination works pretty well 

 

 

Pretty impressive. As you say you can easily drop those numbers by 10% by using a half decent 80+ PSU. Add in using Eco greens and you could easily run 12 + drives on a 350W PSU.

 

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

 

...

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

 

Hey any note this controller?  Woo Hoo.... Looks like we have a new cost effective 8 port PCIe controller for unRAID!!!!!

 

Well spotted ! 14 HDD with no port multiplying and a nice neat layout all Supermicro brand for not a huge sum. Chnaged days indeed.

 

Hmm.  Need to find a UK supplier for this board.

 

This is a real killer. Its as if the board is made of unobtanium which is illegal in most EU countrys cause literally a hand full of places stock it and none in the UK.

 

 

I really look forward to building it up with the new Mini ITX Zotac H55ITX-A-E motherboard I have on the way!  Part of me regrets the 9-10W of idle power and fanless operation I'm going to lose compared to this supermicro, but the performance headroom for running other applications (and wicked-fast startup time) is something I really value.

 

I am concerned at the 6 minute boot time you are seeing for the Atom. can anyone confirm / deny this? Also I am annoyed that the Zotac is easily available to buy here but the Supermicro is not.

[GaryMaster]

I am concerned at the 6 minute boot time you are seeing for the Atom. can anyone confirm / deny this? Also I am annoyed that the Zotac is easily available to buy here but the Supermicro is not.

 

I'm starting to think that this may be something peculiar about my Atom 330 board and unRAID loading from FLASH.  When I run an ubuntu web server on that same machine, it loads in less than 2-3 minutes from the hard disk.  I'm sure someone else here can quickly measure their startup time on this Supermicro board which has a much better I/O controller.

[Joe L.]

It sounds as if the flash drive is being read in USB-1 mode.  Perhaps a BIOS setting might affect the boot time.

[Chris Pollard]

Hmm.  Need to find a UK supplier for this board.

 

This is a real killer. Its as if the board is made of unobtanium which is illegal in most EU countrys cause literally a hand full of places stock it and none in the UK.

 

 

 

Everything supermicro seems to be the equivalent of rocking horse shit here in the UK.  The suppliers listed on the Supermicro website all seem to be systems integrators or wholesale only.  I'm going to email scan later to see if they plan to stock it.

[GaryMaster]

It sounds as if the flash drive is being read in USB-1 mode.  Perhaps a BIOS setting might affect the boot time.

 

It is a brand new board and I'm fairly certain I have confirmed that the USB is set properly in the BIOS, but worth a second look.  

 

If a user of this supermicro can time from pushing the power button to the unRAID double beep (or accessibility of the web console if you are running headless with no speaker), it will put this question to bed.  I am also personally curious how quickly this supermicro gets unraid operational.

[WeeboTech]

It sounds as if the flash drive is being read in USB-1 mode.  Perhaps a BIOS setting might affect the boot time.

 

It is a brand new board and I'm fairly certain I have confirmed that the USB is set properly in the BIOS, but worth a second look. 

 

If a user of this supermicro can time from pushing the power button to the unRAID double beep (or accessibility of the web console if you are running headless with no speaker), it will put this question to bed.

 

You can see the issue by how fast the dots go across the screen when booting.  If it crawls, then it's a bios/usb issue.  sometimes the usb key interface is faster when it is formatted as a usb-hd. Since the bios comes into play with usb-fd, usb-key, usb-zip or usb-hdd, the format of the key could have something to do with this too. It should not take 6 minutes to boot. 

 

Also there is the possibility of using grub or grub4dos with syslinux/memdisk. when using grub or grub4dos you can boot from one the array drives as long as you rsync /boot to /mnt/disk1/boot and set up the menu.lst to point to it, but this is a discussion for another thread.

[NAS]

Hmm.  Need to find a UK supplier for this board.

 

This is a real killer. Its as if the board is made of unobtanium which is illegal in most EU countrys cause literally a hand full of places stock it and none in the UK.

 

 

 

Everything supermicro seems to be the equivalent of rocking horse shit here in the UK.  The suppliers listed on the Supermicro website all seem to be systems integrators or wholesale only.  I'm going to email scan later to see if they plan to stock it.

 

Let me know how you get on if you can please.

[tractng]

I am getting frustrated waiting for this board to arrive. I ordered from ewiz which showed it had some in stock.  Then got an email saying the supplies ran out and they won't have it until March 16.  I didn't want to wait and reordered from wiredzone which they told me they expect some on March 2 (I was willing to pay more).  Then today, they told me it won't have any until March 8.

 

Wiredzone:

"Thank you for your order! Unfortunately an item in your order is out of stock. It is possible we had the item in stock and our inventory ended before we could process your order. "

 

Is this board made of gold or SM just don't make enough ?

 

tnt

[Chris Pollard]

 

Everything supermicro seems to be the equivalent of rocking horse shit here in the UK.  The suppliers listed on the Supermicro website all seem to be systems integrators or wholesale only.  I'm going to email scan later to see if they plan to stock it.

 

Let me know how you get on if you can please.

 

Email from Scan :-

 

Chris Pollard,

 

Thank you for your enquiry

 

We have checked with our supplier and have confirmed this product has not been added to our new products list as yet by the supplier.

 

Our suppliers have direct access to our system enabling them to add new products directly onto this list so I will watch out for these for you and as soon as we have more information of specification and price the web site will be updated

 

Please contact us if you have any further queries

 

Best Regards

 

Scan Computers

 

[GaryMaster]

You can see the issue by how fast the dots go across the screen when booting.  If it crawls, then it's a bios/usb issue.  sometimes the usb key interface is faster when it is formatted as a usb-hd. Since the bios comes into play with usb-fd, usb-key, usb-zip or usb-hdd, the format of the key could have something to do with this too. It should not take 6 minutes to boot. 

 

I verified that the BIOS settings were correct, but it does crawl during the USB load.  It is a very high speed flash drive (Patriot XT Boost).  This isn't my NAS, so I'm not going to waste any more time troubleshooting the USB boot. 

 

Would still like to get feeback on unRAID load time from a cold start by a Supermicro X7SPA-HF user in case I decide to build one with this board.

[limetech]

Would still like to get feeback on unRAID load time from a cold start by a Supermicro X7SPA-HF user in case I decide to build one with this board.

 

For the -L version:

 

- No hard drives connected

- Flash is a Lexar Firefly 2GB (not the speediest flash out there, but typical)

 

(No card plugged into the PCI-E slot)

17 seconds - Reset to the blue unRAID boot menu screen

11 seconds - To load bzimage and bzroot

16 seconds - To get to linux login prompt following bzimage/bzroot being loaded

 

(With Supermicro SAS controller plugged into PCI-E slot)

39 seconds - Reset to unRAID boot menu screen

11 seconds - To load bzimage and bzroot

27 seconds - To get to linux login prompt following bzimage/bzroot loaded

 

For the -HF version:

 

I'll get the card back in a few days, but I do know that time to load 'bzimage' and 'bzroot' from the flash is 38 seconds.

 

The slower USB access on the -HF is probably something that will be addressed in a bios update since this is a new card for Supermicro & the bios is currently at revision 1.0b.

 

Supermicro in general has slightly slower USB boot.  For comparison, I have an old Intel D865GLCLK that loads 'bzimage' and 'bzroot' in exactly 5 seconds from the same Flash.

[GaryMaster]

(No card plugged into the PCI-E slot)

17 seconds - Reset to the blue unRAID boot menu screen

11 seconds - To load bzimage and bzroot

16 seconds - To get to linux login prompt following bzimage/bzroot being loaded

 

(With Supermicro SAS controller plugged into PCI-E slot)

39 seconds - Reset to unRAID boot menu screen

11 seconds - To load bzimage and bzroot

27 seconds - To get to linux login prompt following bzimage/bzroot loaded

 

For the -HF version:

I'll get the card back in a few days, but I do know that time to load 'bzimage' and 'bzroot' from the flash is 38 seconds.

 

Thanks for that, Tom.  If I understand you correctly, that's between 44-77s total boot time from a warm start on the -L version depending on whether an additional controller is used.

 

Based on the 38s bzimage and bzroot segment, it sounds like a minimum time of 71s on the -HF model (with current BIOS).  That wouldn't be hard to live with.  Glad to hear the USB load performance of the Atom board I have here is atypical.

 

I had benchmarked a variety of flash cards with unRAID and (to my surprise) the slowest card (11MB/s) had virtually the same startup time as the fastest card available (35MB/s), so I doubt your results would change much with a faster flash.

 

The details of the comparison are posted here if anyone is interested in USB flash speed impact:

http://lime-technology.com/forum/index.php?topic=5242.msg48963#msg48963