Hi, I have a 4-socket system with 256 GB RAM (32 x 8 GB). Socket 0 shows unstable behavior when running the stream benchmark. It runs well for some time and then the performance drops to 50 %. It seems like throttling of some component. The CPU temperatures are very low and therefore I consider it a problem with one of the 8 DIMM modules on Socket 0. Is there any way to read the temperature? I checked the documentation on the DRAM [0] and it says that there is a thermal sensor on the DIMM. There is also some information on thermal sensor information in [1]. It says: Support for integrating RDIMM thermal sensor information into Intel SMI Status Frame. Best regards, Andreas Hollmann System information: 4 x Intel(R) Xeon(R) CPU E7- 4870 @ 2.40GHz [0] 32 x 8 GB Samsung M393B1K70CH0-CH9 [1] Intel 7500 Scalable Memory Buffer [2] Mainboard X8QB6-F [0] http://www.samsung.com/global/business/semiconductor/file/2011/product/2010/2/4/676543ds_ddr3_2gb_c-die_based_1_5v_rdimm_rev10.pdf [1] http://www.intel.la/content/dam/doc/datasheet/7500-7510-7512-scalable-memory-buffer-datasheet.pdf [2] http://www.supermicro.com/products/motherboard/Xeon7000/7500/X8QB6.cfm?SAS=Y http://www.intel.com/content/dam/doc/platform-memory/ddr3-rdimm-nehalem-ex-validation-results.pdf _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
