Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
[New motherboard] MSI X99S XPOWER AC, debug inside/
#1
Posting here to get new information stuff added to hwinfo. In addition to GTX 980's, which I've already submitted, I am now running the following NEW hardware.

Samsung XP941 512GB PCI-E x4 M.2 SSD
MSI X99S XPOWER AC Motherboard
Intel Core i7-5960X Processor
G.SKILL 4 x 8GB DDR4-3000 RAM (F4-3000C15Q-32GRK)

Here is report: https://dl.dropboxusercontent.com/u/9768...ETH-PC.HTM
Here is debug: https://dl.dropboxusercontent.com/u/9768...iNFO64.DBG

Everything in BIOS is on auto, except XMP profile is loaded, which caused these things to change automatically:

CPU Multiplier to 28x instead of 30X
BCLK to 125.03 MHz
RAM speed is 3000 MHz
RAM voltage is 1.35v

System Agent (VCCSA) was also manually set by me to 1.25v
I took a screenshot before and after of setting this voltage from AUTO to 1.25v and did not see any changes in the voltage sensors area, so either this sensor is not monitored yet, or this sensor does not exist, or it was already set to 1.25V in auto mode.

Estimated CPU power seems to be incorrect, but I have not found another program that accurately measures it yet either. Reloading default BIOS settings shows assumingly correct wattage.

Auxiliary temperature sensor on Nuvoton NCT6792D does not appear to be linked to anything. Did not change from 13C even after applier hair dryer heat all over.

VIN6 or VTT could be the PCH +1.05v sensor.

I will be back later to test more things. Right now, trying to get memory stable.
Reply
#2
Thanks for the feedback.
According to MSI, VCCSA cannot be monitored on those boards.
If you're sure, that VIN6 is PCH +1.05V then I'll add this.
Let me know if any changes/updates are required...
Reply
#3
+5, +3.3, and +12 are presumably correct
VCCIN (CPU Input voltage): Value is correct.
VCore (CPU Core voltage): Value is correct on NCT6792D sensor. VCore on NCT7511Y is unknown. Does not respond to actual changes to VCore.
Ring Voltage: No sensors change in response to changing Ring voltage inside OS or BIOS. Presumably not monitorable.
System Agent Voltage: No sensors change in response to changing Ring voltage inside OS or BIOS. Presumably not monitorable.
CPU VCCIO Voltage: This changes on both VTT and VIN6. They both match. The value is correct.

DRAM Channel A/B voltage: This changes on DRAM CH_C/D. Value is otherwise correct.
DRAM Channel C/D voltage: This changes on CPU I/O. The value appears to he half the actual voltage. Changing this voltage caused an "External 1" sensor to appear under the Nuvoton NCT7511Y/NCT7802Y field. Appears as a 0C temperature sensor. Reloading HWInfo caused the sensor to disappear and changing the voltage again did not make the "External" sensor appear again. The "DRAM CH_A/B" sensor inside HWInfo does not respond to changes in either DRAM channel voltages. Does not appear to be monitored in HWInfo but the sensor probably exists.

DRAM VPP Channel A/B voltage: This changes on DRAM CH_A/B. Value is otherwise correct.
DRAM VPP Channel C/D voltage: This changes on VIN5. The value appears to he half the actual voltage.

DRAM VREF Channel A/B voltage: This voltage changes on VSEN2 of Nuvoton NCT7511Y/NCT7802Y. The value is correct. Changed this voltaged and it caused Nuvoton NCT7511Y/NCT7802Y to freak out. A few more external sensors were added, many temperatures went to 255C, some voltages went to 3x to what they were. I closed out HWInfo and re-opened it. It detected sensors very slowly. Upon finishing, the Nuvoton NCT7511Y/NCT7802Y was gone. Does not appear to be monitored in HWInfo, but is monitorable in the BIOS. Also nuked my voltage setting program, so will be rebooting.
DRAM VREF Channel C/D voltage: This voltage changes on VSEN3 of Nuvoton NCT7511Y/NCT7802Y. The value is correct. Did not see any voltages change. Monitorable in BIOS.
Fun fact: MSI Command Center's voltage changing program picks up the values 0.075 lower than it actually is. Setting any of these voltages will set them 0.075v higher than listed. Clearly a bug in MSI's software department.

PCH 1.05 Voltage: This changes on VSEN1 of the Nuvoton NCT7511Y/NCT7802Y. The value appears to he half the actual voltage.
PCH 1.5 Voltage: This changes on the Vcore sensor of the Nuvoton NCT7511Y/NCT7802Y. The value appears to he half the actual voltage.

Those are about all the voltages I can change.

No idea what AVCC is (probably a clone of +3.3v) or 3VSB or VIN14 is.
Reply
#4
Thanks for the additional info.
Unfortunately neither the Debug File nor Report File you attached contain sensor data, so I can't verify this in more detail Sad
I'll ask MSI about that NCT7511Y..
Reply
#5
Nice to see you are still here monitoring the thread Tongue I'm still working on hotswapping voltages with MSI's "Command Center" program. Got a few voltages left.

BTW, that NCT7511Y does have a motherboard temperature sensor, which does seem to have a valid value.

edit: I've finished changing all possible voltages. Hope you got some good feedback from it.

edit2: I think I got the debug to capture sensors this time. I don't think I opened the sensors window in debug mode last time. Here you go. I'm also all done changing all possible voltages to elicit a change in the sensors window. Refer to my larger post above.

Report: https://dl.dropboxusercontent.com/u/9768...ETH-PC.HTM
Debug: https://dl.dropboxusercontent.com/u/9768...iNFO64.DBG
Reply
#6
Thanks for the great analysis ! it's very much appreciated.
Yes, the NCT7511Y is probably monitoring its own temperature which can be used as the mainboard temperature. Similar for the Vcc.

I'll adjust those values, however I'm not sure if this is valid for your model only, or other MSI X99 series. I have sent another request for clarification to MSI and waiting for their response...

For those dumps to contain sensor data you need to open the sensors window while running in Debug Mode. Then close sensors and create the report file.
Reply
#7
(09-29-2014, 08:52 AM)Martin Wrote: Thanks for the great analysis ! it's very much appreciated.
Yes, the NCT7511Y is probably monitoring its own temperature which can be used as the mainboard temperature. Similar for the Vcc.

I'll adjust those values, however I'm not sure if this is valid for your model only, or other MSI X99 series. I have sent another request for clarification to MSI and waiting for their response...

For those dumps to contain sensor data you need to open the sensors window while running in Debug Mode. Then close sensors and create the report file.

Ok, I will do that. I opened HWInfo in debug mode, opened the sensors window and let it finish scanning, closed it, then created the report and saved it, then closed out hwinfo entirely. New debug and report files have been uploaded, just click the links again for new versions

I sure like editing Tongue

Wanted to add hoping for some sort of fix for the "CPU Package Power" sensor. I find it important in that it lets me figure out just how hard I'm hitting the CPU. Seems to be bugged though right now. May have been accurate with "default" bios settings, but now with altered CPU and memory settings/voltages, it shows 0.73 W idling, 4.21 W under prime95 load, 0.79 W under linpack load when it's allocating memory, and 2.91 W under linpack full load. According to my Kill-a-watt AC power measurement device, it uses much more power under linpack's full load, so something is really wonky. I'd expect something like ~180-200W under load of linpack.

Worked perfectly on my x79 platform i7-3930k so I miss it already.

Editing again: The bug is not related to HWinfo apparently. I started up an "Intel Power Gadget" program and it reports the same bug, so it's likely a BIOS or driver issue. Will have to wait for it to get fixed elsewhere.
Reply
#8
Thanks. Building a new version for test, will post in a few minutes...
Reply
#9
OK, please try this build: http://www.hwinfo.com/beta/hw64_445_2324.zip
and let me know...
Reply
#10
(09-29-2014, 10:00 AM)Martin Wrote: OK, please try this build: http://www.hwinfo.com/beta/hw64_445_2324.zip
and let me know...

Looks good to me. Just need to figure out what the mystery sensors are now..

"Memory Ambient"
"AVCC"
"VTT"
"VIN14"
"3VSB"

But will leave that for another day. 4AM over here, time for a nap I think.
Reply
#11
Some of those sensors might not be connected, the AVCC could be the chip's supply voltage (+3.3V usually).
"Memory Ambient" might just be a virtual (not real) temperature used for DIMM power management.

4AM that's really time for bed! Have a good night Wink
Reply
#12
I spent the last 15 minutes moving/sorting sensors to being 'nicer'. Man, moving sensors is really cumbersome. Hope that's on your 'todo' lost. Would be nice to be able to drag and drop the sensors to the place you want it to go instead of clicking the sensor, clicking the up arrow, clicking the sensor, clicking the up arrow, clicking the same sensor, clicking the up arrow.. It probably wouldn't be so bad if the same sensor remained selected instead of having the particular position selected.

Anyways, off to actual bed. Thanks for being in contact so easy. You're always easy to work with.
Reply
#13
Yes, such feature is on my list...
You're welcome Wink
Reply
#14
Have a look at the new Beta build 2327 and drag and drop in the sensors window Wink
Reply
#15
(10-08-2014, 02:37 PM)Martin Wrote: Have a look at the new Beta build 2327 and drag and drop in the sensors window Wink

Hopefully you're still monitoring this thread. Yeah I've checked it out. It's nice, though there is a bug with it.

"Added Drag and Drop support in sensors window as a more convenient way to change order of items or hide (drop outside)."

When it switches to non-fixed position mode, lots of empty lines are inserted for where there are already hidden sensors (this is fine). When you said you can hide lines by dropping them outside, it does not work for empty lines.
Reply
#16
Ah yes, I'll have to check that.. Meanwhile you can remove those empty lines via the config screen.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)