- Joined
- Apr 28, 2018
- Location
- NC
Recap: x299 Dark MB, i9-7940x CPU, EVO 960 pci-e SSD, 32GB G Skill PC 3000 RAM, Noctua NH-D15S Cooler
Over the past three weeks, I've put the machine through more than 20 power cycles to flush out any early-life failures. Soak times during each cycle ranged from 6 hours to 12 hours. Load on the machine was very little with nothing more than Win 10 background processes running. Power plan settings were set at maximum which set the cpu to the clock speed of 3.1GHz. The machine now has over 200 hours runtime.
Using the Eleet software provided by Evga and HWinfo, temps, speed and cpu utilization were monitored. During basic no-load conditions, temps ranged from 26C to 28C across all 14 cores. CPU package temp was mostly constant at 27C. These values were recorded after a 3 - 4 hour soak time.
Over the past week, I have run a few iterations of Prime95. This was run for 3 - 5 hours each time. Temps ranged fro 52C to 55C across all 14 cores and an average temp of 56C for the cpu package. CPU speed was 3.8GHz for the tests. This is the maximum multipler for cores 13 and 14 as indicated by the software while cores 1 and 2 support a 44 multiplier. Utilization for both threads in each core was 100%.
One interesting observation I made was when the test was stopped and utilization dropped to 0%. The core temps dropped, on average, 20C in 20 seconds. I'm certainly no expert on thermal performance, but this seems to indicate very efficient cooling by the Noctua. The very tight core to core temp delta seems to support the conclusion that the cpu/cooler interface is quite uniform.
The last stage of baselining the machine's thermals and overall stability was to run an extended Prime95 iteration. I allowed Prime95 to run 40 hours uninterrupted. During the run, temps occasionally peaked at 58C with a core to core delta of 3C. This appeared to correlate with certain sizes of FFTs used in Prime95. Mostly though, the max temp observed was 56C with a core delta of 2C. Again, cpu speed was 3.8GHz with 100% utilization of all 28 threads.
I may be in error and please inform me if I am, but these results seem pretty good. I plan to try a little overclocking - in small amounts, in stages over the next couple of weeks - monitoring temps and overall stability. Any comments or observations on these results are welcomed.
Over the past three weeks, I've put the machine through more than 20 power cycles to flush out any early-life failures. Soak times during each cycle ranged from 6 hours to 12 hours. Load on the machine was very little with nothing more than Win 10 background processes running. Power plan settings were set at maximum which set the cpu to the clock speed of 3.1GHz. The machine now has over 200 hours runtime.
Using the Eleet software provided by Evga and HWinfo, temps, speed and cpu utilization were monitored. During basic no-load conditions, temps ranged from 26C to 28C across all 14 cores. CPU package temp was mostly constant at 27C. These values were recorded after a 3 - 4 hour soak time.
Over the past week, I have run a few iterations of Prime95. This was run for 3 - 5 hours each time. Temps ranged fro 52C to 55C across all 14 cores and an average temp of 56C for the cpu package. CPU speed was 3.8GHz for the tests. This is the maximum multipler for cores 13 and 14 as indicated by the software while cores 1 and 2 support a 44 multiplier. Utilization for both threads in each core was 100%.
One interesting observation I made was when the test was stopped and utilization dropped to 0%. The core temps dropped, on average, 20C in 20 seconds. I'm certainly no expert on thermal performance, but this seems to indicate very efficient cooling by the Noctua. The very tight core to core temp delta seems to support the conclusion that the cpu/cooler interface is quite uniform.
The last stage of baselining the machine's thermals and overall stability was to run an extended Prime95 iteration. I allowed Prime95 to run 40 hours uninterrupted. During the run, temps occasionally peaked at 58C with a core to core delta of 3C. This appeared to correlate with certain sizes of FFTs used in Prime95. Mostly though, the max temp observed was 56C with a core delta of 2C. Again, cpu speed was 3.8GHz with 100% utilization of all 28 threads.
I may be in error and please inform me if I am, but these results seem pretty good. I plan to try a little overclocking - in small amounts, in stages over the next couple of weeks - monitoring temps and overall stability. Any comments or observations on these results are welcomed.
