GPU temperature curve
The two diagrams show the warm-up behavior of the GPU temperatures of the MSI GeForce RTX 5090 SUPRIM LIQUID SOC in gaming and silent mode under real gaming load as well as under synthetic full load (“Torture”), each at 22 °C room temperature. Gaming mode shows clearly controlled behavior with stable temperature values: The GPU reaches a plateau of around 53 to 54 °C after a few minutes in the gaming scenario, while it remains just under 60 °C in the Torture test at maximum load. Both curves are very flat and without major fluctuations, which speaks for a powerful, well-regulated cooling system. It is noticeable that the temperatures in the gaming scenario are significantly lower than in the synthetic stress test despite the high power consumption.
In silent mode, the temperature curve is slightly higher across the board, which was to be expected. The GPU stabilizes at around 56 to 57 °C in gaming mode and rises to around 64 °C in the Torture test. Here, too, the temperature rise is controlled and flat, although the temperature level is around 3 to 5 Kelvin higher than in gaming mode. This is mainly due to the lower fan speeds and the more conservative temperature target used in silent mode.
It is interesting to note that the difference between Gaming and Torture remains similar in both modes, which indicates good thermal scaling and even heat distribution. The temperature jump at the beginning of both curves can be explained by the loop ramp-up and the thermal inertia of the system, but stabilizes after a few minutes. Gaming mode offers a visibly cooler GPU temperature thanks to the more aggressive fan control, but also remains quiet. Silent Mode focuses on acoustic comfort and accepts a slightly higher temperature, but remains far below critical values even in the Torture scenario. Both profiles demonstrate the high efficiency of the 360 mm AiO system and confirm the thermal reserve strength of the card. Critical throttling temperatures are never reached, neither in everyday use nor under synthetic full load.
Memory temperature curve
The two diagrams show the temperature development of the memory of the MSI GeForce RTX 5090 SUPRIM LIQUID SOC in the gaming and torture scenario, each in gaming mode and silent mode at an ambient temperature of 22 °C. While GPU temperatures are strongly influenced by the active airflow and the cooling surface distribution, the GDDR7 memory is subject to different thermal dynamics, as it is connected to the copper base via thermal pads and receives less direct cooling through airflow.
In gaming mode, the memory temperatures rise rapidly in both load scenarios within the first few minutes and then reach a stable plateau. The maximum temperature in the gaming scenario is around 67 °C, in the torture test slightly lower at around 65 °C. The slight fluctuations in the Torture curve indicate more dynamic load changes, but do not show any thermally critical swings. Both curves are close together, which indicates even heat distribution and a reliable memory connection to the cooling block.
In silent mode, the behavior is comparable, but with a slightly higher thermal base. The memory again reaches around 67 °C in gaming mode, while the Torture test stabilizes at around 66 °C. Both curves are largely flat and parallel, without any noticeable fluctuations. The memory modules thus also exhibit thermally stable behavior in silent mode, albeit with a slightly higher temperature than in the actively cooled gaming mode.
It is noticeable that there are hardly any relevant differences between gaming and torture scenarios in both modes. This speaks for an effective, passive memory connection that is not dependent on dynamic air cooling. Critical values do not occur in either the quiet or the performance-oriented operating profile. The storage cooling remains constant and predictable, even under maximum continuous load. The temperature reserve is therefore sufficient to reliably cover the thermal requirements of the GDDR7, regardless of the mode.
Clock rates
The next two diagrams show the behavior of the GPU clock rates as a function of temperature when running a gaming scenario for the MSI GeForce RTX 5090 SUPRIM LIQUID SOC. Tests were conducted in Gaming Mode and Silent Mode at a room temperature of 22 °C. In gaming mode, after a short initial boost with clock rates above 2900 MHz, a rapid transition to a stable plateau at around 2760 MHz can be observed. At the same time, the temperature rises evenly after a short drop and stabilizes at just over 53 °C. The slight drop in the clock rate following the short peak behavior at the beginning can be attributed to the interplay between the power target and thermal headroom. The temperature then remains constant, without any major outliers, and the clock rate shows high stability over the entire test duration.
In silent mode, a comparable picture emerges with only marginally different progression. Here too, the short-term boost value is above 2900 MHz, followed by a stable operating frequency of around 2730 MHz. The GPU temperature reaches an average of around 57 °C. The difference of around 30 MHz to the active cooling variant is small and is unlikely to have a measurable effect on gaming performance in practice. It is interesting that despite the quieter fan profile and slightly higher temperatures, there is no significant clock reduction. The card maintains its frequency confidently even in silent mode, which indicates a well-designed thermal and electrical design.
It should be noted that both modes enable very stable boost clock rates. The gaming mode offers a little more thermal leeway and slightly higher clock rates, while the silent mode is only a few MHz behind, but works with a quieter cooling profile. The differences are functionally minor and show that the SUPRIM LIQUID SOC can almost reach its full potential even without maximum fan performance. This once again confirms the quality of the AiO cooling, which enables consistently high performance even in whisper-quiet operation thanks to the 360 mm radiator.
- 1 - Introduction, overview and technical data
- 2 - Test system and equipment
- 3 - Teardown: PC, components and cooler
- 4 - Material analysis and TIMs
- 5 - Gaming performance
- 6 - Power consumption, transients and PSU recommendation
- 7 - Temperatures und clock rates
- 8 - Thermal imaging and modding
- 9 - Fan curves and noise
- 10 - Summary and conclusion
30 Antworten
Kommentar
Lade neue Kommentare
Veteran
1
Veteran
Moderator
1
Urgestein
Urgestein
Urgestein
Urgestein
Urgestein
1
Urgestein
Veteran
Urgestein
1
Urgestein
Urgestein
Urgestein
Mitglied
Alle Kommentare lesen unter igor´sLAB Community →