SuperMicro X10SL7-F

Status
Not open for further replies.

panz

Guru
Joined
May 24, 2013
Messages
556
KVR16E11/8 is listed, but I was quoting jgreco
 

jyavenard

Patron
Joined
Oct 16, 2013
Messages
361
I am reasonably certain it and the 1333's had been listed on the compat table for the X10's.

The KVR16E11/8 is still listed for X9S.


I purchased 8 KVR16E11/8EF Kingston modules from the Australia supermicro distributor, based on their recommendation that this ram was certified by Kingston as compatible with the x10sl7-f

Turned out it's not anymore.

Having said that, I have no issue whatsoever with them.

I'm using 2 KVR16LE11/SE with an ASRock E3C224 with no issue either. Those too were certified by Kingston as being compatible.
 

panz

Guru
Joined
May 24, 2013
Messages
556
I'm running a 180 hours HCI memtest (fingers crossed) ;)
 

DJABE

Contributor
Joined
Jan 28, 2014
Messages
154
Having said that, I have no issue whatsoever with them.
Yes, that's how it goes with official HCL's.... something might work out of the box, but something won't.
However, since I need to order specific hardware from abroad online shop, it would be great if someone could confirm that my combination would work.
SM X10SLM-F && Kingston ValueRAM 2x8 kit KVR1333D3E9SK2/16G...
 

Z300M

Guru
Joined
Sep 9, 2011
Messages
882
Yes, that's how it goes with official HCL's.... something might work out of the box, but something won't.
However, since I need to order specific hardware from abroad online shop, it would be great if someone could confirm that my combination would work.
SM X10SLM-F && Kingston ValueRAM 2x8 kit KVR1333D3E9SK2/16G...
Wouldn't you do better to get RAM that is on Supermicro's own "tested memory list"? -- Hynix HMT41GU7MFR8C-PB or Samsung M391B1G73BH0-CK0.
 

cyberjock

Inactive Account
Joined
Mar 25, 2012
Messages
19,525
DJABE: The next time you jump into another thread with a topic you're going to get a warning. Twice you've done that. Stick to your own thread. There is no reason to jump into another thread and expect someone to review your stuff. If they wanted to review it, they'd look at your thread!
 

indy

Patron
Joined
Dec 28, 2013
Messages
287
Tower coolers are designed with a different approach to solve thermal silicon issues... it does not matter that they direct no airflow to the CPU socket! They're keeping cold heat pipes who are in a direct contact with the CPU surface / die...
I was talking about the SAS-controller getting cooled by the CPU-fan.
 

G Brown

Dabbler
Joined
Jan 2, 2014
Messages
31
For LSI 2308 Firmware updates, the below link has pictures/screen shots! One can get to the UEFI shell by hitting F11--if that key is not mapped to something else (and of course for me it was)--and then note the devices when the shell first starts up. In my case it was fs0:. I took out the freenas boot stick so there would only be one usb stick, because I can get easily confused. So now I am running version 16 IT. I was running version 15 IR, which came with the board. I don't know if the streamlined IT code will make much difference over the slightly bloated IR code. The board so far seems very good. Next, I must virtualize--of course!

http://www.napp-it.org/doc/manuals/f...srh-7tf_it.pdf
 

indy

Patron
Joined
Dec 28, 2013
Messages
287
The Intel cooler works absolutely fine and quiet for the cpu.
I am not as happy regarding the sas-controller though - it gets hot to the touch idling with 1 disk attached.
Honestly I would stay away from tower coolers since they direct even less airflow towards it than the stock cooler.
I have a top-blow-cooler in the mail (Noctua NH-L12), I will post the results.
Ok, that did not work at all ;)
Actually the sas-controller got even hotter than with the stock cooler, I think.
The problem is that the Noctua fans go to super low speeds where they basically do not move any air.
Couple that with the pretty much non-existent fan-speed settings of the Supermicro boards and that concept does not work in itself.
However I used the included 92mm fan to blow some air directly onto the sas-controller which seems to lower temperatures to semi-reasonable regions.

Other notes:
- I really like the cooler in itself, great build quality and very capable for its size and weight.
- The cooler does not get properly pressed onto the cpu in its stock configuration.
In fact it just 'dangled' on there, thank god I noticed that.
In the attached picture you can see a red stripe, where I had to insert a washer.
- Configuring ipmi to accept the low fan speeds was a major pain for a FreeBSD novice like me.
 

Attachments

  • noctua_nh_l12.jpg
    noctua_nh_l12.jpg
    65 KB · Views: 387

cyberjock

Inactive Account
Joined
Mar 25, 2012
Messages
19,525
Not to bash you, or argue with you, but Noctua is very well known for quality products. If it was dangling there then I'd say either a piece is missing(or lost), or user error is playing a part. As we all know(and you obviously know) the heatsink is supposed to be applied with pressure onto the CPU heatspreader. I'd go looking for pictures of the cooler to see if maybe you are missing a piece.
 

indy

Patron
Joined
Dec 28, 2013
Messages
287
User error is of course a likely suspect ;)
However I could not find any installation error, there are not many 'adjustable' parts determining the distance between cpu and cooler anyway.
Just double-check your installation if you choose to install this particular cooler.
 

mstang1988

Contributor
Joined
Aug 20, 2012
Messages
102
With the kingston memory support being removed what modules are recommended? I know Hynix HMT41GU7MFR8C-PB or Samsung M391B1G73BH0-CK0, but where to get them?

I'm looking at upgrading my current configuration to handle more drives and ECC memory while still running ESXi. I would be buying this board, the processor, and memory. Today I have 4x3TB 7200.11 drives and looking to add up to 4x4TB drives. The config would be:
4x3TB in RAIDZ1
4x4TB in RAIDZ2

Total drive space would be 28TB, Total usable space around 20TB or less. I would like to go with a minimum of 16GB of RAM if not 24-32GB. I would like to keep the RAM cost immediately around $200-$300. I've found a few DIMM's of those listed but they seem to be around 200 for 8GB so I must be looking in the wrong place.
 

Z300M

Guru
Joined
Sep 9, 2011
Messages
882
With the kingston memory support being removed what modules are recommended? I know Hynix HMT41GU7MFR8C-PB or Samsung M391B1G73BH0-CK0, but where to get them?
Probably depends on where you are located.
Being in the USA, I would look at amazon.com, newegg.com, or superbiiz.com for a start.
 

cyberjock

Inactive Account
Joined
Mar 25, 2012
Messages
19,525

Radu

Dabbler
Joined
Mar 7, 2014
Messages
45
I am using Hynix memory and no memtest v4.20 errors so far ... 6 passes.
Did anyone used some method of cooling the LSI SAS cipset or tryed to change the heatsink? In my opinion it runs too hot.

I have seen that lot's of people virtualise ... how does ZFS works with virtualization? Wouldn't be better if ZFS has direct disk control?

@indy How does noctua nh l12 fits after all? Have you thought of Thermalright AXP-100?
 

mstang1988

Contributor
Joined
Aug 20, 2012
Messages
102
When using ESXi 5.0 and later you can use a feature called "pass-through" that allows you to pass the physical adapter and the disks to the virtual machine. It runs natively with no performance issues.
 

Z300M

Guru
Joined
Sep 9, 2011
Messages
882
Surprise! looks like the problem may be with the RAM and not the mainboard...

http://forums.freenas.org/index.php?threads/unscientific-findings-with-kingston-ddr3-1600.19114/
But Kingston took both varieties of 8GB modules off the recommended list for the X10-series motherboards -- the "Server Premier" variety and the "vanilla" variety. And the X10-series motherboards are not on Kingston's compatibilty list for their new, replacement 8GB "Server Premier" modules (KVR16EF/8KF); nor is there even a 4GB "Server Premier" recommendation for the X10-series boards.
 

cyberjock

Inactive Account
Joined
Mar 25, 2012
Messages
19,525
But Kingston took both varieties of 8GB modules off the recommended list for the X10-series motherboards -- the "Server Premier" variety and the "vanilla" variety. And the X10-series motherboards are not on Kingston's compatibilty list for their new, replacement 8GB "Server Premier" modules (KVR16EF/8KF); nor is there even a 4GB "Server Premier" recommendation for the X10-series boards.

Okay? But you've just added a ton of other variables compared to the thread I linked to above. And virtually all of those variables can have other variables that we as simple consumers are not privy to know or ever have access to. There's no doubt in my mind that they have some kind of non-disclosure clause specifically for issues like this, and there's no doubt in my mind that there's far more going on "behind the scenes" with this problem than we will ever appreciate. Even if you assume that both of my "assumptions" are wrong, you are still left with the fact that historically situations like this have been far far more complex than you or I can appreciate, and there's no telling when or if that information will be made public.

But, in the here and now, the one thing to keep in mind is that those sticks of RAM should have had a reasonably high chance of working, but didn't. Maybe if we can get others to test their sticks that are in question we can get a better idea of what is going on. But, based on our currently available information it's sure looking like more of a RAM problem than anything else. But time will tell.

Edit: There's also likely good reasons why Supermicro took them off their list. Two I'd consider to be perfectly valid and that I have seen in other situations are:

1. The RAM stick model number is no longer manufactured.
2. Some lots of that component are not what they are labeled to be. For example, maybe they changed manufacturers for the silicon and didn't change the model number(crap like this happens regularly and they will NEVER admit to what they did or why) and the new silicon isn't compatible. Since a small subset of those parts aren't what they claim to be and you have no way of proving if a given stick will or won't work they take the conservative approach of simply removing that model since you can't tell the difference. It's better to remove stuff from the list that *might* not work than to leave stuff that is proven to not work.

And if you look at some people here that have 4 of those "unapproved" sticks, both of the reasons I just listed above would explain why many people have not had a problem with the "unapproved" sticks while others have.

The bottom line, there's just not enough information for us to sit and try to point a finger at the RAM or the motherboard definitively. But, considering that the RAM should have worked on the AsRock it's a good hint of what *may* have gone on behind the scenes.

What I'd really like to see is what other hardware had the Kingstom sticks listed and if those are having problems in other hardware. ;)
 

jyavenard

Patron
Joined
Oct 16, 2013
Messages
361
At the risk of repeating myself, I have 8 of the EF sticks in two x10sl7-f machines, both lots are working just fine.
So your mileage may vary.
 

Dr.T

Dabbler
Joined
Nov 9, 2013
Messages
18
Finaly got first stage of building my home server finished: Supermicro X10SL7, Xeon 1230v3, 8GB Kingston KVR16E11/8, Fractal Design Define R4.
Now testing it with seven 500GB drives in raidz3 mode (had them anyway).
Future second stage - buying HDDs and appropriate amount of RAM for them.
my_freenas2.jpg
 
Status
Not open for further replies.
Top