I am using a 3U server cooler from dynatron, the S6: S6 | Dynatron - it should still be within spec for the socket in vertical-board orientation, I would think. It gets a bit loud under load, but not any louder than the A6000 with its blower cooler already does under load. I used the same cooler for both cpus. Also some of the tests were conducted with the case laying on its side mostly because I had been to lazy to stand it back up after the reseating test for a while (and then again, after installing the 2nd cpu), but I did not notice any differences between vertial and horizontal board orientations.
My thermal paste application on the 2nd chip might leave something to be desired, since under prime95 some cores on the 2nd tile (cores 12-23) get near to 90C (TjMAX for this chip is 95C) on small FFTs, but the 2nd chip is not having problems (outside of the pstate issue under low load that RolloZ170 mentioned, which is definitely noticeable when the computer has been sitting idle, though the lag from that isnt as terrible as the lag the 1st chip had). The first chip was just using the stock dynatron thermalpaste application, and that was the chip that had problems (but it never got as hot, so I think I'm safe in assuming the dynatron thermal application was doing its job adequately).
I did try reseating the cooler w/ original chip still attached, but it didnt change anything.
The reported temps for the DIMMs are at least well within their 0-95C operating spec (Peak temp on hottest DIMM I've recorded under load was 66C). These are 16GB RDIMMs running at 4800MTS.
In addition to the cooler, providing some front-back airflow over the VRM heatsink, I've got a pair of 120mm fans blowing downwards onto the ram and vrm. The PSU exhausts air from the front to the top (case is the Sliger Cerberus), and a small noctua fan along with the airflow from the cooler exhausts air out the rear.
Yes, I installed the ASPEED driver (on windows). Windows recognizes it is there but it turned out that it was only using the A6000 for both 'low power' and 'performance' graphics modes.
My thermal paste application on the 2nd chip might leave something to be desired, since under prime95 some cores on the 2nd tile (cores 12-23) get near to 90C (TjMAX for this chip is 95C) on small FFTs, but the 2nd chip is not having problems (outside of the pstate issue under low load that RolloZ170 mentioned, which is definitely noticeable when the computer has been sitting idle, though the lag from that isnt as terrible as the lag the 1st chip had). The first chip was just using the stock dynatron thermalpaste application, and that was the chip that had problems (but it never got as hot, so I think I'm safe in assuming the dynatron thermal application was doing its job adequately).
I did try reseating the cooler w/ original chip still attached, but it didnt change anything.
The reported temps for the DIMMs are at least well within their 0-95C operating spec (Peak temp on hottest DIMM I've recorded under load was 66C). These are 16GB RDIMMs running at 4800MTS.
In addition to the cooler, providing some front-back airflow over the VRM heatsink, I've got a pair of 120mm fans blowing downwards onto the ram and vrm. The PSU exhausts air from the front to the top (case is the Sliger Cerberus), and a small noctua fan along with the airflow from the cooler exhausts air out the rear.
Yes, I installed the ASPEED driver (on windows). Windows recognizes it is there but it turned out that it was only using the A6000 for both 'low power' and 'performance' graphics modes.
