<p>Thermal management in multicore microprocessors has become increasingly prominent due to the generation of highly dynamic and non-uniform heat loads, which lead to the formation of localized hotspots. Inefficient cooling of these hotspots degrades the performance and reliability. The present study evaluates the performance of two micro heat sinks (MCH-1 and MCH-2) and compares them to mitigate the effects of non-uniform heat generation. The proposed heatsinks are integrated with three configurations of flow (I-, U-, and Z-type) for optimizing the flow distribution between the microchannels. To mimic the low computational heat loads of the microprocessors, the uniform (115&#xa0;W,155&#xa0;W,185&#xa0;W,230&#xa0;W) and single hotspot (40&#xa0;W) heating conditions are investigated. The flow rates of 0.25 LPM, 0.50 LPM, and 0.75 LPM were used for all sets of experimentation. The experimental results indicate that the MCH-2 design with Z-flow configuration is more suitable for uniform heat-load conditions, as it recorded the minimum thermal resistance and maximum heat transfer coefficient, namely 0.12&#xa0;K/W and 4173&#xa0;W/m<sup>2</sup> K, at a flow rate of 0.75 LPM, respectively. On the contrary, under non-uniform heat-load conditions, MCH-1 also showed a suitable design compared with MCH-2, although with a penalty in pumping power. Particularly, MCH-1 is suitable for better performance in the hotspots near the outlet header (AH1, AH2, AH3, AH4), and MCH-2 is suitable for better performance in the hotspots near the inlet header (AH5, AH6, AH7, AH8, AH9). This study emphasizes the practical design guidelines for developing efficient cooling systems for the thermal management of next-gen multicore processors.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Experimental evaluation of the thermal performance of a parallel microchannel cooling system for thermal management of multicore processors under low computational loads

  • Siva Krishna Karnati,
  • Manikanta Bandlamudi,
  • Mathiyazhagan Shanmugam,
  • Lakshmi Sirisha Maganti

摘要

Thermal management in multicore microprocessors has become increasingly prominent due to the generation of highly dynamic and non-uniform heat loads, which lead to the formation of localized hotspots. Inefficient cooling of these hotspots degrades the performance and reliability. The present study evaluates the performance of two micro heat sinks (MCH-1 and MCH-2) and compares them to mitigate the effects of non-uniform heat generation. The proposed heatsinks are integrated with three configurations of flow (I-, U-, and Z-type) for optimizing the flow distribution between the microchannels. To mimic the low computational heat loads of the microprocessors, the uniform (115 W,155 W,185 W,230 W) and single hotspot (40 W) heating conditions are investigated. The flow rates of 0.25 LPM, 0.50 LPM, and 0.75 LPM were used for all sets of experimentation. The experimental results indicate that the MCH-2 design with Z-flow configuration is more suitable for uniform heat-load conditions, as it recorded the minimum thermal resistance and maximum heat transfer coefficient, namely 0.12 K/W and 4173 W/m2 K, at a flow rate of 0.75 LPM, respectively. On the contrary, under non-uniform heat-load conditions, MCH-1 also showed a suitable design compared with MCH-2, although with a penalty in pumping power. Particularly, MCH-1 is suitable for better performance in the hotspots near the outlet header (AH1, AH2, AH3, AH4), and MCH-2 is suitable for better performance in the hotspots near the inlet header (AH5, AH6, AH7, AH8, AH9). This study emphasizes the practical design guidelines for developing efficient cooling systems for the thermal management of next-gen multicore processors.