- Dec 21, 2010
I did this today and it was fairly easy. I will check how the video came out.
Thanks for the temp info ... I get those temp with case top off and a Noctua sitting on the CPU heatsink.Oh yeah... I knew that.. TBH I was just distracted with getting FreeNAS 10 installed and testing all the new features....
For now I did this..
sysctl -a | grep temperature
I'm seeing from 40 deg Celsius to 48 deg. under medium load. (Not prime 95 or anything super heavy.. just general load.. FreeNAS w/ 2 Bhyve VM's (1 BSD & a 2012 R2 Running Atto and Crystal at the time)
Here is a photo of the 2 heatsinks used to plot the charts above, side by side, note the original SM is cooler is sitting on sata connectors and other items, so it is higher than when on CPU.Update 2: with results !!
Note this is to use the boards in 2u to 4u chassis.
Summary: the passive Cooljag BUF-E offers a considerable improvement in heat transfer and cooling of the Xeon D over the supermicro (SM) passive 1u heat sink as delivered with these boards. The test was run system at idle and then PRIME95 program with a given workload, on each heat sink. It was not required to run a 100% workload, to indicate if the heat sink was an improvement.
The improvement was a reduction in fan speed and a 21 C drop in temp. Note the fan speed in plot below is the average of the 3 fans, there was basically a spread of +/- 100 rpm. Note: the goal was to not try and keep the cpu temp at ambient, that was a consequence of the test conditions.
View attachment 3435
Choose this cooler to test first because:
1. it has more space between fins, which requires less air pressure to cool effectively, and potentially lower power fans.Thanks to Patrick for the heads up on removing the original heat sink. Ran it for 20 minutes, removed mounting screws, and twisted, it did not come easily..... The heat sink backplane was harder to remove as it was stuck to the motherboard, it required a screw to be installed and move from mounting hole to mounting hole, and several taps to loosen it.
2. 2u not 1u
3. copper generally has a superior thermal conductivity than steel, however impurities and the elements in the alloys can significantly change thermal conductivity.
4. to change the coolers was easy, undo 4 screws, remove back plate and install new cooler and backplate
5. easily available at antares pro and frozencpu online retailers, then via shipito to Australia.
Used a SM Xeon d-1518 in X10SDV-4c-7tp4f board in a SM cse-835 chassis with 3 of the 6 fans installed. ( back 2 fans and 1 chassis fan removed). The fans were set to optimal (lowest setting in the original bios), this provided a lowest spin of 2900-3000 rpm which was relatively quiet. The air baffle was in place and will accommodate 2U cooler, but probably not a 3U heat sink.
The only issue was the heat sink back plate on the cooljag had a thinner insulating layer, which required the placement of 4 fibre washer as often used in computer mounts, to provide the same standoff to stop the 4 mounting screws from bottoming out on the threaded section.
Now I have the original SM back plate, will look at some 3 or 4U heat sinks which require a very simple mod to drill 4 holes to mount them. Will try to get a photo of the 2 heat sinks side by side, but he data below provides a good representation..
I am just going to thank you now!I needed active cooling on my SuperMicro Avoton platform and looked at many different options. I settled on the active version of their RPGA-988B HSF: SNK-C0054A4L
I'd be absolutely shocked if that heatsink didn't mount perfectly over this Xeon-D proc on their boards...
SNK-C0054A4L Supermicro Heatsink for X9SCV-QV4/X9SCV-Q Motherboard - Desktop Fans & Heatsinks - SuperBiiz.com
I am just going to thank you now!
I suspect this will fit and I won't have to change out the back plate as the screws will likely match
the current OEM back plate screw holes.
Not to mention it is a 4-pin PWM fan.
Same price for me shipped (~$30) too.
I just ordered one.
Will report back once I get it.
EDIT: I just saw a post/video on TinkerTry mentioning SNK-C0057A4L.
- Wondering what differences are...going find out the expensive way I guess
One other difference is the height of the heatsink as it sits atop the processor. The 54A4L version looks to be fitted for a much shorter processor/socket combination, whereas the 57A4L looks like its designed with taller procs with IHS installed, and maybe less overall heat to dissipate.
Image shows two Supermicro (left / center) and one CoolJag (right) 1U cooling options
And now I know the answer to my question:
1. Height differences - the SupermicroC0057A4L and CoolJag BUF-A are very similar = lowest profile
(Note: SM 54A4L has the same height profile as the stock heatsink that came with my X10SDV-TP8F)
2. Backplate differences - BKT-0043L-F and BKT-0043L-Q
(Note: "Q" version has more threads ...ie ... deeper screw holes than the "F" version)