Power Supply Performance
Personally, I have never heard of the brand HEC. One could consider it a fair assumption that even though the unit itself doesn’t seem to be 80PLUS certified, Mesh Computers aren’t the sort of people that would use substandard equipment in their machines, however just to be sure we cracked open the Multimeter and had a look for ourselves.
The Voltage rails remained solid throughout idle and load with the system’s current configuration without any significant fluctuations. All voltages were well within ATX specification and as such I feel that the HEC 550W PSU is more than adequate for the task.
Our temperature testing consists of a 60 minute OCCT Linpack CPU stress test followed by the monitoring of overall system load in 3DMark Vantage. While these applications may induce a load that exceeds what conventional games and real world programs use, it gives us a better insight of how well the computer may handle higher ambient temperatures or how well it may perform after months and months of use with increasing dust build up.
Our OCCT Linpack test revealed that the AMD Phenom II X4 955 CPU was hitting temperatures as high as 65 celcius under full load. While we appreciate that this one of AMD's fastest CPU's and runs at a nominal voltage of 1.3500V, 0.025V higher than most of the product family and that the Linpack test is particularly intensive, we still felt that the temperatures were higher than expected. Furthermore, the computer took just short of 6 minutes to return to the idle temperatures previously reported before the test had begun.
Upon opening the chassis and looking around the CPU area we observed that some thermal paste was oozing around the base of the heatsink. One shouldn't discount the possibility that too much thermal paste was applied so the heatsink was removed and we cleaned the base of it and the CPU’s heatspreader, then applied a pea size portion of Arctic Ceramique on it’s centre. After leaving the system to warm up and idle for an hour, the Linpack test was rerun and this time load temperatures had dropped slightly to 63c. This was still higher than expected. Taking a hint from the slow cooldown time and the 7c increase in GPU temperatures from Idle to Load, we figured that the quality of airflow in the case itself may be the main factor. Upon removing the side panel of the case, temperatures decreased by 4-7c across the board. It was also noted that the Akasa
CPU Heatsink's fan as well as the Power Supply Unit's fan had decreased notably in RPM. Dropping a high end set of components into a relatively small mid tower case aimed towards silent use may result in higher ambient temperatures and that the only way to keep things silent are to opt for larger heatpipe coolers on major heat sources with larger diameter fans (that can then be run at lower speeds) bolted on. In fairness, credit goes to Mesh for implementing a three heatpipe tower cooler with 92mm fan for the CPU but the graphics card cooler uses a smaller high RPM fan but much more importantly does not have a ducting system that exhausts hot air out of the case, unlike the reference HD 4870 cooler design by Ati.
All in all, even though the temperatures were somewhat high, there was no reason for concern. As mentioned, our testing process will inevitably stimulate higher load temperatures than in real world applications. It was however disappointing that the NZXT Hush Chassis wasn’t able to cope particularly well with the heat output of the components inside.