Dual Xeon Gold 6150 in Corsair Air 540

[NickTdot]

Heya,

Started this build last night...

Build’s Name: Corsair-Gold
Operating System/ Storage Platform: Windows Server 2016
CPU: Dual Xeon Gold 6150 @ 2.7Ghz base
Motherboard: Supermicro X11DAi-N
Chassis: Corsair Carbide Air 540
Drives: Samsung 961 nvme M.2 , WD Re 3.5"
RAM: 128GB 2400Mhz DDR4
Add-in Cards: XL710 40GbE , AOC-SLG3-2E4 nvme PCIe card
Power Supply: Corsair 1000Watt
Other Bits:

Had to drill a couple holes for the E-ATX motherboard standoffs layout. Relocated the 2x 140mm fans on the case front to the top, and replaced with 3x 120mm Arctic fans (it's pre drilled to accomodate). The 140mm fans were too thick and interfered with the DIMMs.

Still have to figure out permanent cooling on the heatsinks. There's no active coolers for the LGA3647 that I can find , so for now, sticking a 120mm fan on top of each heatsink.

 

[Monoman]

Woah. Nice!

 

[NickTdot]

Some initial numbers..

CPU-Z Validator: Intel Xeon @ 1196.91 MHz - CPU-Z VALIDATOR

Cinebench:

Passmark CPUMark:

 

[Boris]

And what do you plan to do with all that Gold?

 

[voodooFX]

Looks nice!
What temperatures you have with this semi-passive cooling?

 

[NickTdot]

As is in the picture, the temps get too high, I needed to put a fan atop each cooler to keep the temps in check.. Seeing peak of 93C under heavy continuous load. Under normal load, 5C above ambient

 

[megabird]

Heya,

Started this build last night...

Build’s Name: Corsair-Gold
Operating System/ Storage Platform: Windows Server 2016
CPU: Dual Xeon Gold 6150 @ 2.7Ghz base
Motherboard: Supermicro X11DAi-N
Chassis: Corsair Carbide Air 540
Drives: Samsung 961 nvme M.2 , WD Re 3.5"
RAM: 128GB 2400Mhz DDR4
Add-in Cards: XL710 40GbE , AOC-SLG3-2E4 nvme PCIe card
Power Supply: Corsair 1000Watt
Other Bits:

Had to drill a couple holes for the E-ATX motherboard standoffs layout. Relocated the 2x 140mm fans on the case front to the top, and replaced with 3x 120mm Arctic fans (it's pre drilled to accomodate). The 140mm fans were too thick and interfered with the DIMMs.

Still have to figure out permanent cooling on the heatsinks. There's no active coolers for the LGA3647 that I can find , so for now, sticking a 120mm fan on top of each heatsink.

You Can use Active heatsink for CPU.

 

[PigLover]

As is in the picture, the temps get too high, I needed to put a fan atop each cooler to keep the temps in check.. Seeing peak of 93C under heavy continuous load. Under normal load, 5C above ambient

Not surprising given the layout. Your primary air goes out the top, but the passive heatsinks are aligned for front-back flow. Old days I'd tell you to turn them 90-degrees and watch your temps fall, but you can't because the package is not square anymore.

Definitely go with active HSF as suggested by @megabird. Just putting fans onto the passive heatsinks won't be very efficient because they are designed for a constant flow front-to-back rather than air blowing down onto them.

 

[NickTdot]

Not surprising given the layout. Your primary air goes out the top, but the passive heatsinks are aligned for front-back flow. Old days I'd tell you to turn them 90-degrees and watch your temps fall, but you can't because the package is not square anymore.

Definitely go with active HSF as suggested by @megabird. Just putting fans onto the passive heatsinks won't be very efficient because they are designed for a constant flow front-to-back rather than air blowing down onto them.

Primary air does not go out the top. 2 posts so far mention this, but that's not the case, and you can tell by the fan blades.

You Can use Active heatsink for CPU.

Point me towards a Active Heatsink for LGA3647.... Pls show me one.

 

[PigLover]

Primary air does not go out the top. 2 posts so far mention this, but that's not the case, and you can tell by the fan blades.

Given that layout it probably should go out through the top. Positive pressure is rarely efficient for cooling and almost never effective using low static pressure fans like those. Besides - you are fighting nature by trying to make hot air go down.

Point me towards a Active Heatsink for LGA3647.... Pls show me one.

Probably need patience on that one. The platform is brand new and the channel needs time to fill. Meantime fans on the passive heatsinks would probably help a little bit.

 

[PigLover]

One more thought, but only if you are in the mood to be especially creative.

You could consider building a duct (more precisely, a pair of ducts) that would force air through the passive HSFs. Something to cover the top, with one side open and the other side ducted to one fan set in the "exhaust" direction. Would be pretty easy to do the "back" CPU using the rear fan. The "front" one would require some tricky folds to hit one of the top mounted fans. Or - just to see how it works out - a single duct over both HSFs using the rear fan (though the air might be too pre-heated before it hits the rear CPU.

Make sure you use a materiel with flash-point over 150c - probably not cardboard for this one, though you could use cardboard or thick paper for quick prototypes. Could be an awesome 3d print project.

 

[megabird]

Primary air does not go out the top. 2 posts so far mention this, but that's not the case, and you can tell by the fan blades.



Point me towards a Active Heatsink for LGA3647.... Pls show me one.

Here is the supermicro part number for heat sink and picture.

 

[NickTdot]

I know of this one, but there's 0 stock from Supermicro . I'll email again to see if there's new stock

 

[Patrick]

I put my request in for that since I could not buy it and we have a few motherboards to test (for @William). Still nothing on my end.

 

[NickTdot]

I was finally able to source 2x of the Supermicro heatsinks, and they've just come in!

 

[Evan]

I was finally able to source 2x of the Supermicro heatsinks, and they've just come in!

Please let us know how loud they are.
That's a very powerful system you have there