We are currently designing the thermal solution for our chassis, so it’s crucial for us to understand the system behavior under full loading.
However, we have not been able to reach the official stated 40W in our tests.
We tried running only the GPU workload in a chamber set to -40°C to 20°C, and the power consumption reached around 35W to 38W, but it could not remain stable at 38W.
When we added a burn-in test for the CPU, the system immediately dropped to around 32W.
Since we are testing under low temperatures in the chamber, we believe it’s unlikely that thermal throttling due to GPU or CPU temperature is the issue. You previously mentioned that exceeding 40W would trigger throttling, but in our case, the power never even reaches 39W before dropping.
Could you please share how you test the 40W full loading? Or could you clarify what kind of testing conditions are considered to meet your definition of full loading?
It seems that the key issue is indeed related to the current limit.
We are using MAXN_SUPER mode, and just running the GPU test alone is already hitting 40W.
We’ve observed that the OC3 counter keeps increasing.
After OC throttling is triggered, is it possible for the system to return to its original state?
Because during GPU testing, we noticed that the power drops from 38W to 33W and never goes back up.
Additionally, it seems that OC3 is not a fixed value?
We tried using different input voltages, and it appears that the OC limit changes accordingly, with the total still capping at 40W.