Sandy Bridge reported VIDs don't match specs

[cljsf]

Hi there,

HWiNFO32 is a great product and been very useful as I overclock my i5 2500k. Thanks for providing it as freeware.

I have noticed that the VIDs reported by HWinfo32 (and a number of other monitoring apps) give figures like 1.3411v, 1.3461v, 1.3511v, 1.3561v. However, the VID table in the Sandy Bridge technical doc shows values like 1.340v, 1.345v, 1.350v, 1.355v, i.e. the reported values seem to be offset from the technical doc values by 0.011v. This seems to be consistent across different clock/load levels and different monitoring apps.

Any idea why this is?

 

[Martin]

Thanks for the report. I'll need to investigate this a bit more and thus need more details. Please enable Debug Mode in HWiNFO32, then run it as usual, try to force different VID values, then close HWiNFO32. Then please attach or send me the HWiNFO32.DBG file it created.

Also please check if you haven't the CPU VID offset applied (Over Clocking Extra Voltage).

 

[cljsf]

Have enabled Debug mode and run the rig three times - twice at 4.5Ghz and once at 4.4Ghz. I saved and renamed the DBG file after each run. I'm not 100% on the CPU VID offset applied that you mention, but I have an Asus P8P67 Pro and there is a setting for "Additional Turbo Voltage" and on the two 4.5ghz runs it was set to Auto on the first run and 0.004v on the second (you can't set it to zero). In both cases, HWiNFO32 reported the same VID request from the CPU of 1.3611v, so I don't think the reported value is being distorted by the settings in BIOS/UEFI; or at least not the additional voltage at turbo setting. I've also seen screen grabs of other people's overclocks and they show the same 11mv offset from the Intel VID table values (mostly shown in Real Temp), which suggests this anomaly is not unique to my rig or HWiNFO. Will email you the DBG files direct.

Thanks

Charlie