|
Review Arctic-Cooling Silencer NV5 English |
|
martes, 20 de marzo de 2007 |
|
Página 8 de 10
We have never questioned the probes' readings, since they've been properly verified. But, as we mentioned before, the difference with the internal measurement made the margin of error too high to take it as something accurate.
If we did, the temperature distribution in the different components could be considered as a tendency and it would coincide approximately with what we expected. Of course, taking into account the components location to that of the heat sink, and the copper base in charge of transmitting the heat.
To simplify the chart, we've averaged out the data of the two memory chips of each group, since the values between them were very similar.
We observed that the memory group B was the best refrigerated, as it was located near the fins, with very close air circulation. As these conditions worsened, the performance decreased in groups A and C and even more in D, which is situated further from the center, practically under the fan.
Now -and by way of speculation- if we have a look at the temperatures that the memory probes produced, we will see that all of them are within a range that does not exceede 4°C between the maximum and minimum in a same moment. Then we could assume that on the other side of the PCB the behavior is similar, and extrapolate this data, taking as a starting point the difference there is between the GPU's probe and the reading produced by the driver. As it is also possible to see that the measurement of the GPU's maximum temperature -which was measured with a probe- is higher to that of the memory chips, we can then say that the real temperature of the memory chips cannot be higher to that of the GPU.
|
