MSI X99S SLI PLUS On Linux

[phoronix]

Phoronix: MSI X99S SLI PLUS On Linux

For Intel Core i7 5960X Haswell-E Linux testing I originally bought an MSI X99S SLI PLUS motherboard as it was one of the most interesting, lowest-priced boards available at the time of the Intel X99 chipset debut. While I initially ran into some problems, those issues have now been confirmed to be isolated, and with a replacement X99S SLI PLUS motherboard I have been stressing it constantly for the past few weeks on Fedora and Ubuntu. The X99S SLI PLUS has now proven itself to be a reliable motherboard that's still among the least expensive X99 ATX motherboards on the market.

http://www.phoronix.com/vr.php?view=21069

[DanL]

I personally find details like the following important:

Audio: Realtek ALC892 (rather than just saying "7.1 HD audio")
Network: Intel I218 (rather than just saying Gigabit Ethernet)

Super I/O chip looks like a Nuvoton NCT6792D. If that's the case, you will have to build your own kernel module for it right now since the commit just landed a few months ago and it hasn't been pushed to upstream kernel yet: https://github.com/groeck/nct6775/co...084d5a80821fad
It would be nice if Michael tested that module out and see if sensors and fancontrol worked.

[NotMine999]

Sensors Are Always A Challenge

I personally find details like the following important:

Audio: Realtek ALC892 (rather than just saying "7.1 HD audio")
Network: Intel I218 (rather than just saying Gigabit Ethernet)

Super I/O chip looks like a Nuvoton NCT6792D. If that's the case, you will have to build your own kernel module for it right now since the commit just landed a few months ago and it hasn't been pushed to upstream kernel yet: https://github.com/groeck/nct6775/co...084d5a80821fad
It would be nice if Michael tested that module out and see if sensors and fancontrol worked.

Based on my own past experiences working with developers from the LM Sensors project on testing various sensors, being able to actually access the chip comes first, then reading the sensors comes next, and fan control features come later, assuming the vendor's datasheet can clearly and accurately explain how their fan management features work. It's easy to poke a value into an entry in "/proc/sys/..." via Linux CLI provided that entry permits changes; some entries can be marked "read only" by a driver. It's much harder to adjust those values in a programmatic manner based on some external "stimulus", like a temperature reading or threshold limits.

Reading the sensors accurately and correctly identifying which inputs are monitoring which motherboard aspects is critically important as not every motherboard voltage is monitored or even "instrumented" on the board, or monitored on a predictable pin on the SuperIO chip. Temperature readings can be tricky on Intel boards due to the presence (or absence) of a Intel PECI between the CPU and the SuperIO chip; it's cheeper not to include this link (think "budget board"). Some motherboard vendors will monitor CPU temperature via the PECI Agent rather than an expected temperature input.

[dragonn]

I personally find details like the following important:

Audio: Realtek ALC892 (rather than just saying "7.1 HD audio")
Network: Intel I218 (rather than just saying Gigabit Ethernet)

Super I/O chip looks like a Nuvoton NCT6792D. If that's the case, you will have to build your own kernel module for it right now since the commit just landed a few months ago and it hasn't been pushed to upstream kernel yet: https://github.com/groeck/nct6775/co...084d5a80821fad
It would be nice if Michael tested that module out and see if sensors and fancontrol worked.

Hmm thanks, I tested this module on my Z97 Gaming 3 and it works, for fans good but voltages and some temperatures are complete wrong, but still much better then having only coretemp.

[DanL]

Hmm thanks, I tested this module on my Z97 Gaming 3 and it works, for fans good but voltages and some temperatures are complete wrong, but still much better then having only coretemp.

I wonder if you could work with the lm-sensors dev(s) to see if it could be straightened out? The lm-sensors site says that the nct6792 is untested, which probably means they don't have access to the hardware and are just going off datasheets and/or similar chips.

[dragonn]

I wonder if you could work with the lm-sensors dev(s) to see if it could be straightened out? The lm-sensors site says that the nct6792 is untested, which probably means they don't have access to the hardware and are just going off datasheets and/or similar chips.

I am not really an programer but I know many about Linux kernel - I often moding kernel for Android phones. Do you have maybe a link or mail with one should I contact to give them help in testing this driver?

[DanL]

The lm-sensors mailing list is where I would start: http://www.lm-sensors.org/wiki/FeedbackAndSupport