SuperMicro X10SL7-F

[AleQQ]

Ok, so after a suggestion on another of my threads to get this board and reading this thread, I ordered the Retail packaging from NewEgg (part number N82E16813182821). I also ordered the Samsung 1.35v memory off SuperBiiz (part number D38GRE160S). Both showed up the next day (yay california!) I ripped the old board/memory out of my case and installed the new ones. Ran MemTest for about 3 days straight, and (with the help of the post on Page 6 from @jyavenard) reflashed my LSI chip with the IT drivers. (correct link for those, as mentioned by someone else, was ftp://ftp.supermicro.com/Driver/SAS/LSI/2308/Firmware/IT/Previous%20Releases

So, some quick notes:
1. The new board comes pre-loaded with IPMI v1.42 and BIOS v2.00
2. XHCI definitely has to be turned off. My FreeNAS boot drive is usb 3.0, and i had the tunable installed and it worked fine on my old board, but I got a mountroot error on boot with the new board (plugged into the internal USB port). The error stopped the second I turned off XHCI support and XHCI handoff
3. My NIC wasn't responding at first. Then i had to look up how to configure my static IP again from the console (yes, I'm a newb -.- ).
4. The retail kit DOES come with 6 SATA cords, but I like having the clips on mine, so I decided to buy some from Fry's. They finally carry a cheap and simple SATAIII cable that works on AHCI. UPC is 0837281102235.

I think that's about it.

The 8GB 1.35V DIMMs recommended by Crucial for Supermicro X10-series motherboards exactly match the Micron part number tested by Supermicro, as listed in their memory compatibility list.

That makes me sad. I would rather have Crucial memory than Samsung, though I suppose I rely on Samsung for their SSDs...

Last note:

Thank you to everyone who has tried, tested, and troubleshooted this board over the last year or so. This thread made the purchasing decision and installation of this board go smoother than anything has yet with FreeNAS :P
I love you all!

 

[Sir.Robin]

As much as i love Crucial and their support, other brands as HP etc. also delivers Samsung memory for their servers. So i would not feel sad having Samsung chips in my NAS.

I do however feel a bit sad over my friggin Kingston memory.

 

[Pooh]

X10SL7-F-B plus Intel Xeon E3-1220v3 with x2 Samsung 8GB DDR3L ECC 1600MHz 1.35V.
got it 13mins ago and its gone :D

 

[punk2g]

Hi All,

After doing a fair bit of reading, I'm fairly sold on this board for the FreeNAS build I'm currently planning. It's quite possibly overkill, but it seems to be performing well for a large number of people around here. I was hoping any Australian users could suggest the vendor they used to purchase one? I can only seem to find two local retailers that are stocking it for in excess of $400 AUD. Any info would be appreciated.

 

[jstrom]

Hi,

great thread, lots of info on the X10SL7-F!

I'm looking for some power consumption figures, mainly for idle usage without any disks, but any idle numbers are interesting.
My current plan is to either get this board with a Xeon E3 v3, or an Avoton based board... But I'd like to find out where the X10SL7-F idles at first.

If you have any numbers, please also add the CPU + number of memory sticks used, and any additional stuff added!

(Oh, and in case I missed this info in any of the previous 18 pages, my apologies)

Another question while I'm at it: the idle consumption of the L variants of the E3-v3 CPUs will similar to that of the non-L versions, but the L versions will have a lower max-rating. I.e. if aiming for low idle power usage, but non-idle power usage is not that important, it's no use to go for a L version. Is this correct?

Thanks!

 

[Ericloewe]

Hi,

great thread, lots of info on the X10SL7-F!

I'm looking for some power consumption figures, mainly for idle usage without any disks, but any idle numbers are interesting.
My current plan is to either get this board with a Xeon E3 v3, or an Avoton based board... But I'd like to find out where the X10SL7-F idles at first.

If you have any numbers, please also add the CPU + number of memory sticks used, and any additional stuff added!

(Oh, and in case I missed this info in any of the previous 18 pages, my apologies)

Another question while I'm at it: the idle consumption of the L variants of the E3-v3 CPUs will similar to that of the non-L versions, but the L versions will have a lower max-rating. I.e. if aiming for low idle power usage, but non-idle power usage is not that important, it's no use to go for a L version. Is this correct?

Thanks!

Low-power CPUs are only useful if you're thermally constrained. Power usage at idle is at best trivially better and at worst the exact same - but you miss out on the performance when you do need it. So yes, the L versions are as good as useless for the vast majority of us.

 

[Z300M]

Hi,

great thread, lots of info on the X10SL7-F!

I'm looking for some power consumption figures, mainly for idle usage without any disks, but any idle numbers are interesting.
My current plan is to either get this board with a Xeon E3 v3, or an Avoton based board... But I'd like to find out where the X10SL7-F idles at first.

If you have any numbers, please also add the CPU + number of memory sticks used, and any additional stuff added!

(Oh, and in case I missed this info in any of the previous 18 pages, my apologies)

Another question while I'm at it: the idle consumption of the L variants of the E3-v3 CPUs will similar to that of the non-L versions, but the L versions will have a lower max-rating. I.e. if aiming for low idle power usage, but non-idle power usage is not that important, it's no use to go for a L version. Is this correct?

Thanks!

With E3-1230v3 CPU, 2 x 8GB 1.5V RAM modules and 10 x 2TB desktop drives spinning, my UPS output indicator ranges from 83W to 90W.

 

[jstrom]

Low-power CPUs are only useful if you're thermally constrained. Power usage at idle is at best trivially better and at worst the exact same - but you miss out on the performance when you do need it. So yes, the L versions are as good as useless for the vast majority of us.

Exactly what I though then, good with another source confirming this. Thanks!

Edit:
A follow up question, does anyone know if there are any difference in power consumption between the "refreshed" 2014 chips? I.e. 1231 vs 1230?

 

[jstrom]

With E3-1230v3 CPU, 2 x 8GB 1.5V RAM modules and 10 x 2TB desktop drives spinning, my UPS output indicator ranges from 83W to 90W.

Lets see, the ST2000DM001 uses ~5.8W in idle (~8W operating), and the ST32000641AS ~6.39W idle (~9.23W operating). Just averaging those values over 10 disks gives around 60W, so the actual system would be somewhere 23-30W then I guess. Thanks!

 

[Ericloewe]

Exactly what I though then, good with another source confirming this. Thanks!

Edit:
A follow up question, does anyone know if there are any difference in power consumption between the "refreshed" 2014 chips? I.e. 1231 vs 1230?

Trivial improvement at best due to improved manufacturing. Silicon is the exact same stepping. Only Devil's Canyon has physical differences and even then the silicon is untouched. Motherboards have been having trouble with the new ones, though, before they're upgraded to newer BIOSes, so the old ones are probably the easiest solution.

 

[Pooh]

If i can find my wattage meter ill update with numbers later.


Collected it today and had a nerdgasm when i opened the package

 

[Pooh]

To flash the LSI 2308 into IT mode I used the following firmware:
ftp://ftp.supermicro.com/Driver/SAS/LSI/2308/Firmware/IT/

Do some one happen to have a working link for the firmware?
All guides and discussions lead back to same location.

 

[indy]

Code:

ftp://ftp.supermicro.com/Driver/SAS/LSI/2308/Firmware/IT/

Just copy and paste that, the forum seems to mess with ftp links?

Anyway since you are using a 6 disk setup, did you think about using the Intel sata-ports?
I would save you a little more than 10W if (I dont know that) you could completely disable the SAS2308 controller.

 

[jstrom]

Trivial improvement at best due to improved manufacturing. Silicon is the exact same stepping. Only Devil's Canyon has physical differences and even then the silicon is untouched. Motherboards have been having trouble with the new ones, though, before they're upgraded to newer BIOSes, so the old ones are probably the easiest solution.

Okay, weird that a BIOS upgrade would be required when there are no changes other than improved manufacturing.. But well, it seems my supplier only has the refreshed ones, so I guess I'll have to go with a BIOS upgrade anyway (unless mobo is shipped with a fresh one :)).

 

[SirMaster]

Does anyone have any experience using this board's SAS controller with a SAS Expander?

I'm trying to use this board with this expander: http://www.newegg.com/Product/Product.aspx?Item=N82E16816212021

Everything appears to be working perfectly except I sometimes get some messages in my kernel log during heavy writes to my zpool when using the expander.

Code:

mpt2sas0: log_info(0x31110d01): originator(PL), code(0x11), sub_code(0x0d01)

 

[krikboh]

That Chenbro card is SAS-1 and your LSI chip is SAS-2. Not sure if that could cause issue.

 

[cyberjock]

A few years ago when I was looking for a SAS expander for my Windows Server the Chenbro stuff was not exactly "high quality". For that reason I avoided them despite them being very pocketbook-friendly. Sadly my hunch is that you are being bitten by the same problems I was hoping to avoid.

As for how serious the error is I have no idea. It could be nothing or it could be a serious problem. If your server is important to you I'd seriously look at replacing it though. I use an Intel SAS Expander on my M1015 and I've never had a problem with it.

 

[SirMaster]

That Chenbro card is SAS-1 and your LSI chip is SAS-2. Not sure if that could cause issue.

Shouldn't be.

I will also add that when I connect the Chenbro expander to a LSI 9211-8i (also a SAS2 card) then everything works just as well as the onboard SAS, except I never see the mpt2sas messages.

 

[krikboh]

Are both flashed to the same IT firmware?

 

[SirMaster]

A few years ago when I was looking for a SAS expander for my Windows Server the Chenbro stuff was not exactly "high quality". For that reason I avoided them despite them being very pocketbook-friendly. Sadly my hunch is that you are being bitten by the same problems I was hoping to avoid.

As for how serious the error is I have no idea. It could be nothing or it could be a serious problem. If your server is important to you I'd seriously look at replacing it though. I use an Intel SAS Expander on my M1015 and I've never had a problem with it.

The Chenbro itself should be fine. It came from a buddy who has been using it for years in production with an LSI controller and I also just added that it produces no kernel messages with connected instead to an LSI 9211-8i that I am borrowing.

From my digging the best I can come up with is the messages I'm seeing mean some disk has timed out to a command, and the command queue has to be flushed and re-queued. So it doesn't *seem* too severe so far.

I have also replaced the cable connecting the onboard SAS to the Chenbro with no change in the messages.

My server and data is obviously important to me. But since it is I also keep a complete second array in a backup system which is a mirror of the array in my primary server so I can tolerate catastrophic failures and re-create the zpool in my primary server whenever I need/want to.