Anybody else have any experience with low voltage Xeons?

[Drudenhaus]

Hey guys, I'm doing testing on a platform at work and the customer pretty stuck on using the following config:

Chassis: Supermicro 513L-420B
Motherboard: Supermicro X7DCL-i
CPUs: 2x Xeon E5410 (QC 2.33ghz)
Heatsinks: 2x SNK-P0017

The problem here is that on full load the CPUs runs 67-68C with Intel's thermal spec being 67C (and personally, I'm not fond of running over 60C). I tried out a pair of L5420 CPUs (same specs, but 50W draw instead of 80W) figuring that with lower power consumption, they would run cooler. Instead, a pair of L5420s at 100% load run up to about 80C with Intel's thermal spec being 57C on these. I'm not expert in thermodynamics or anything related, but I would have though that lower power consumption = lower heat output.

Does anybody else here have any experience with the L series Xeon CPUs and seen something like this before?

Reference:
E5420 specs
L5420 specs

 

[FlyinBrian]

Temps dont look to bad to me for an Xeon. Depending on Amient temp of course. Double check air flow within the case and possibly reseat the heat-sinks with some better thermal grease. Personally considering those are passive heatsinks id consider puting a couple of fans on them.

 

[Drudenhaus]

Temps don't look to bad to me for an Xeon. Depending on Ambient temp of course. Double check air flow within the case and possibly reseat the heat-sinks with some better thermal grease. Personally considering those are passive heatsinks id consider putting a couple of fans on them.

The vast majority of 1U cases use passive heatsinks with chassis fans (that way you get a ton more surface area for heat dissipation as opposed to active 1U heatsinks). Personally, I don't like the fact it's a suck rather than push fan config. It's sitting in ambient room temp of about 23C and heatsink contact and airflow are good. I expected the system to run hot with the crappy cooling design, but what got me was that the L5420 supposedly consumes 30W less than the E5420, but it runs quite a bit hotter. Have you seen that particular characteristic before?

 

[FlyinBrian]

Yeh I agree man I would of came to the same conclusion as you and by all specs it should be runing cooler. You could take off the heat sink lay a strait edge razer over it and see if the metal plate has a concave to it if its beveled that would ruin the heat Transfer. Also Intel makes a copper heatsink with a fan for those chips. I bet you 100 dollars they would cool a lot better than your solution. The thing is passive is great and all but active almost always out performs and with a 23C ambient temp you should see a nice drop using an active cooling solution. You probably have more experience with Xeons than me I havent built a lot of servers maybe 4 in my whole life but 100s of desktops and entry level servers and thats my best advice.

 

[ilkhan]

passive without airflow doesnt really help. If its the same heatsink on both, perhaps it switched to *F instead of *C?

 

[Drudenhaus]

passive without airflow doesnt really help. If its the same heatsink on both, perhaps it switched to *F instead of *C?

The heatsinks are passive. but the chassis design has the rear fans drawing air through the heatsinks. Additionally, there is an airshroud that goes over the fans and heatsinks.