Hey everyone, first time poster.

 

 

Quick intro.

Got a bit worried when my external hard drives started heating up and having already lost one, i started looking into putting hard drive into a pc case, cooling them in there and also adding some failure protection.

Stumbled upon unraid as one of the solutions, but really fell in love with the whole rig pimping section. So im trying to make something happen on my end.

 

 

Motherboard socket dilemma.

As i dont have an old computer lying around, at least not one capable of what i need, im gonna have to build it from scratch.

My first logical thought was to cram as many hard drives onto the motherboard as possible, before having to buy a sata controller. This would give me some breathing room and enough time to save up for a controller and other expansion costs.

 

So i started basing my build around a motherboard. On that note, while i would absolutely love an ipmi option, im afraid the cost is a little out of my league and since data accessibility is not critical, i can live with some potential downtime.

 

I ended up looking at Intel 1155 and AMD FM2 motherboards and this is where my dilemma begins.

I initially found only 6(sata2+sata3) connections 1155 motherboars, so i quickly became enthralled with fm2 8(sata3) connection motherboards.

While i later found Intel 1155 motherboards that support 8 sata connections, most of them only support up to 4 sata3 connections, the rest being sata2.

With sata3 drives starting to drop in prices and becoming a logical replacement for sata2, AMDs FM2 motherboards seem like a clear winner, as far as the amount of sata(3) connections go.

 

While i realize that current hard drives can hardly saturate sata2, let alone dip into sata3 (discounting ssd), i want to make sure this build lasts for a while, i.e. is a little futureproof. While sata3 connections and the amount of them on FM2 gets me that piece of mind, i feel a bit unsure when it comes to the cpu the motherboard uses.

 

If i go with 1155 and buy a cheap cpu, i can easily (further down the line, 3 years from now) replace it with a cheap used i5 or i7 cpu. Going from celeron or i3 to i5 or even i7 would mean a big upgrade, futher down the line, wihout extra cost (appart from cpu). The downside means not having 8 sata3 that FM2 provides.

On the other hand, if i go with FM2, A10-5800 is pretty much the limit as far as cpu upgrades go. And since i would be aiming at A8-5600k to begin with, this means that i am already pretty much hitting the ceiling with the cpu. Seeing how quickly FM1 got dumped in favor of FM2, this also raises the question if any better cpus will be available further down the line or will they be introduced on FM3.

 

So the question is, should i go with either Intels 1155 socket and "settle" with 4xsata2 + 4+sata3 while gaining the ability to upgrade the cpu further down the line OR pick AMDs FM2 and get 8xsata3, but loose the ability to upgrade the cpu?

 

 

To consider while trying to decide.

- I would really like to make this a quiet build, since its going to be 3-4 meters away from my bed and running 24/7 (got a small apartment).

- I could sacrifice some silence for a decent performance/airflow with moving the case to the kitchen, but if i could choose, id rather not to. The reason for this is that i am leaning towards virtualization in the long run and would like to have it near my TV (in the bedroom), to serve both as a file server and an htpc.

- Having said that, it needs to be able to run Plex and be able to transcode at least 2 720p streams (web/handheld - for when im on the road). At least 2, but 3 max, wont need more.

- As it is also going to be used for a htpc, i would like a hdmi from igpu. Onboard igpus should be more than enough to play 1080p blurays at home, so no need for a discreet graphics card (which makes am3+ viable).

- On the side note, as far as cpu goes, amd a8-5600k is pretty much the limit i can spend on cpu right now, so i3 is a bit out of my league (having 4 cores would also allow me to do some encoding on handbrake on this machine). Having said that, being able to upgrade to an i5 or i7 further down the road, would mean i can put a bigger load on it later (transcoding, encoding).

 

Help me decide, pretty pretty please. Ive been banging my head with this for almost 2 months now and still cannot decide between 1155 and FM2.

 

Big cheers and a big thanks to all for providing techie porn in the pimping section.

Hi,

IMHO, Sata3 is a big marketing unless you use them for SSD. For UnRAID Sata2 is more than enough.

If you move to virtualization, ESXi way for example, then processor/motherboard with vt-d should be the first choice.

Expansion card, like IBM M1015, is not very expensive, you could buy them on Ebay. Many of us, including me, are running ESXi fine with M1015 in IT mode passed to unRAID.   

 

I wouldn't sweat the Sata2/3 delema.  The drives/controllers are backwards compatible so Sata3 drive will work on Sata2 controller and Sata2 drive will work on Sata3 controller.  Most drives currently cannot saturate Sata1 speeds (newest Seagate drives are exception) so the difference between Sata3 and Sata2 does not matter unless you are planing on an array of SSDs.

I agree with what's been said about SATA II vs SATA III.  I have both on my X9SCM-F and the only time SATA III comes into 'measurable' effect is with SSD.

While there's a slight increase in speed with SATA III drives on a SATA III controller. It's not worth the time going back and forth.

 

However, if you are going to virtualize, I would suggest going with what's tried and true on the forum.

the X9SCM-F a compatible CPU.. Then later expand with the M1015 and/or a more appropriate CPU for virtualization.  The super micro boards are not cheap, but they are very reliable. Sometimes you can get a X9SCM-F on special at newegg.  you can always get a cheaper CPU now and buy a more powerful used one later on.

 

Another vote for SATA speed not mattering unless you're going for SSDs. It's going to be a looooong time before a spinner can saturate even SATA II.

Cheers to all for the input. Looking at things from this perspective, it does make more sense to go with Intel for features and for cpu per core performance. Guess i have been putting too much emphasis on sata speed. Was under the impression that hard drives were beginning to saturate sata2 (instead of sata1) and will spill onto sata3 soon. Makes my job to continue this build a lot easier.

 

p.s. got the info i was looking for, problem solved, now if i can only figure out how to tag it as such

 

Again, cheers guys

I would suggest going with what's tried and true on the forum. the X9SCM-F a compatible CPU.

Does the 3770k work with this board? I've found mixed answers

I would suggest going with what's tried and true on the forum. the X9SCM-F a compatible CPU.

Does the 3770k work with this board? I've found mixed answers

 

 

Do you already have that chip? If not the Xeon E3-1230 v2 is the better choice even if the i7 was supported. It's basically the same chip without the GPU for $100 less.

 

Anyway from Supermicro's website:

 

* Per Intel's spec, for platforms using C202/C204 PCH:

  1) Intel i5/i7 CPUs are not supported

  2) Intel E3-12x5 series processors which have integrated graphics support are not recommended.

 

 

The C204 requires ECC memory and i5/i7 do not support it.

I am not sure I agree with the SATA 2 vs 3 arguments.

 

Don't get me wrong every word said is 100% correct but let me put a counter argument.

 

When i purchased my first unRAID system I am not even sure SSD existed. If they did they were the stuff of pipe dreams. Now I have 2 in that system.

 

It is not beyond the realm of possibility that unRAID will get some features that would take advantage of SSD or even SSD get bigger and cheaper.

 

I would suggest that a sensible design goal for many (not all) unRAID builders would be high SATA3 count. Could this number be sensibly set at 4 and negate this debate?

 

Otherwise i suspect some systems will become EOL just that little bit earlier.

 

I also suggest that since most of the world cant source Supermicro or Tyan server motherboards without serious hoops/delays/costs/import tax maybe none of this matter. Buy cheap and replace often perhaps?

 

Food for thought

I am not sure I agree with the SATA 2 vs 3 arguments.

 

Don't get me wrong every word said is 100% correct but let me put a counter argument.

 

When i purchased my first unRAID system I am not even sure SSD existed. If they did they were the stuff of pipe dreams. Now I have 2 in that system.

 

It is not beyond the realm of possibility that unRAID will get some features that would take advantage of SSD or even SSD get bigger and cheaper.

 

I would suggest that a sensible design goal for many (not all) unRAID builders would be high SATA3 count. Could this number be sensibly set at 4 and negate this debate?

 

Otherwise i suspect some systems will become EOL just that little bit earlier.

 

I also suggest that since most of the world cant source Supermicro or Tyan server motherboards without serious hoops/delays/costs/import tax maybe none of this matter. Buy cheap and replace often perhaps?

 

Food for thought

 

I definitely do NOT agree with the "Buy cheap and replace often" philosophy      Quality components are always a better choice.

 

As for the SATA-2 vs. SATA-3 thoughts ... I agree that a FEW SATA-3 ports would be a good idea, if only for an SSD cache drive (or 2 ... Tom has indicated multiple cache drives are likely to be supported in the not-so-distant future).

 

Other than SSD, the only time a SATA-3 drive actually uses more bandwidth than a SATA-2 port provides is in transfers to/from the drive's buffer -- a VERY tiny percentage of the activity (especially with typical UnRAID usage for streaming media).    It's not beyond the realm of possibility that spinners will hit the SATA-2 barrier ... but certainly not likely in the relatively near future.    Even if platter densities were to double ... from 1TB/platter to 2TB/platter ... the sustained transfer rates of the drives wouldn't cross the 300MB/s SATA-2 rate UNLESS the drives were spinning at 10k or better => VERY unlikely with such a high density platter.

 

It is, of course, true that 8TB drives will eventually happen -- and this will almost certainly mean 2TB platter densities.    But as I noted, that will effectively mean SATA-2 is maxed out on outer-cylinder transfers ... they still won't really need SATA-3 rates.    ... of course, when the NEXT iteration beyond that eventually gets here [16TB drives  ], THEN SATA-2 will indeed be a bottleneck.

 

Of course if you're concerned about the "limitations" of SATA-2 speeds - and actually want to take advantage of the SATA-3 speeds for SSDs and a few 16TB drives -- then you'll also need to be sure your system has 10Gb Ethernet ports 

great thread and thank you to the OP - this is exactly the dilemma I have been pondering over for some time.