MSI PM8M3-V rev. 1 - wrong voltages :)

[trodas]

Using an old yet still working MSI PM8M3-V rev. 1 mainboard with Celeron D 336 2.8GHz (only 256k L2 - suxx) and I wanted to know, what temperature under Windows the CPU and chipset (possible even the mobo itself, it it have a temperature sensor) and HDD have.

So I turned out to trusty old SpeedFan.

Much to my surprise, the SpeedFan crash on "unknown software exception (0x0eedfade) on app into pace 0x7964a4e1." After that, it give a "Instruction at 0x00402780 trying to acces ram at 0x0119623c and that ram cannot be read." And after that it says "Runtime error 216 at 00402780."

And that it is. I tried the causual option for the SpeedFan executable /NOSMBSCAN and also tried the /SMBDEBUG, but it just crashing the very same way it did before. So, no change, no help.

So before I change caps on the mobo I turned to MSI forum, asking if anyone have a sucess using SpeedFan on the MSI PM8M3-V rev. 1 mobo and they pointed me out to the HWiNFO utility.

So I downloaded the portable version (hate installing anything) v4.22-1970 and start it. Hoooray, it works!

HWiNFO really works on MSI PM8M3-V rev. 1 mobo:

It is a bit confused, however. First at all, there is the +3.3V voltages twice, yet +12V is missing. Then the value of -12V is clearly wrong 1.7V and the value of -5V is also wrong, stating a 2.1V.

I have no idea what is the Vccp2 voltage and if this is correct... however I finally get the temperatures of the mainboard (29°C) and CPU (35°C at iddle, using Prime95 it shoot to 42°C quite fast).
I have no idea what is the Auxiliary temp and I hope someone tell me... But it works on Win2k SP4 w/o a glitch, it did not crash and it also not need to be installed when I use the portable version.

Hoooray! At least I have something to compare, when I change the mainboard caps tomorrow.

Since the bugreport require the report and debug files, there they are included in zip archive (the html report grow to 178k, lol).
So what can I do to get these values fixed and HWiNFO reporting the right values?

/me preparing to take a photo from the HW info page in bios...


PS. Jsem sice čech a myslím, že bych mohl udělat bugreport česky, ale zase by to nebylo čitelné pro zbytek světa, takže jsem se rozhodl pro angličtinu. Jinak smekám, Martine, je to krásný program a vím, že čtení HW registrů, ke kterým bude minimum informací není jen tak.
Např. na Sapphire A9RX480 desce SpeedFan spadne bez parametru /NOSMMSCAN tak natvrdo, že bez resetu biosu už mobo nenaběhne (!) ...
HWiNFO jsem ještě nezkoušel, obávám se, že může dopadnou podobně, neb problém není v HWiNFu, ale v tom, že to mobo nerado aby se mu chytalo na střeva... je prostě citlivější. Že SpeedFan padne i na téhle staré MSI desce, to jsem nečakal... Komplet informace o mém HW jsou zde:
http://trodas.wz.cz/c.htm
...když už to fórum nepodporuje podpisy Ještě jednou smekám úctou před tak krásným programem. Ještě kdyby si v ini pamatoval, že mně stačí aby zobrazil aktuální teploty/napětí v tom seznamu a byl byl naprosto spokojen jak želva

 

[Martin]

You're right, some of the voltages will need adjusting in HWiNFO, since it doesn't support this mainboard model yet. I'll wait for the BIOS photo, so maybe I can try to do it based on that information. Another alternative might be some MSI tool for Windows which can report correct sensor values.

PS: Vdaka, ze si zvolil anglicky jazyk. Ide presne o to, aby co najviac uzivatelov rozumelo diskusiam a mohlo vyuzit tieto poznatky. Inak, dnes som bol kusok od vas (Hodonin)

 

[trodas]

So, there are the promised bios screens, since the mobo apparently survived the recapping ( I was worried that it run out like crazy, after replacing failing G-Luxon cap with Nichicon LE polymer cap ). Yet IMHO important are only this one about the voltages:

The rest seems uninportant, yet I can provide more, if they can help:



And thanks for the speedy reply, Martin! Yep, the mobo is pretty nonstandard and weird one. It even supports DDR2 rams in the very same slots as DDR rams - the bios settings have at one value two options - one think it meant when DDR (1) rams are present, another when DDR2 rams are there.
Also it did not like setting the Interleave to 8 banks. 4 seems to be maximum, but that might have changed after the recap? I did not dare yet to try it, bios reset is necessary when fail... so...

MSI tool will be probably overbloated with graphic and stuff and maybe not even working on Win2k - but when necessary, I can try anything to help HWiNFO better recognize this beast.
(it is quite unusual that I cannot make any version of SpeedFan work, lol, that must be something pretty strange there)

The MSI PM8M3-V seems to report reasonable values in the BIOS. Even using better thermal compound reduced now the temperature by one degree, lol. I intend to cool the mobo with much better heatsink soon, because the poor thing suffered from the start with bad caps and stealing marketing, lol. Just look - MSI show on it's website this image, of a MSI PM8M3-V mainbaord:
http://s3.postimage.org/hnt6xziib/MSI_PM8_M3_V_advert.jpg

As you can see, the Vcore output is populated with 9 solid quality Sanyo Os-con polymer caps. In reality, however, you get this:
http://i47.tinypic.com/34hx6jl.jpg
A strip down version, from 9 caps to 5 (!) caps and mainly instead of quality polymer caps you get crappy known bad caps - Ost...!

Poor mobo. Amazing it could hold up to today, when the caps get replaced



PS: No, to je za rohem - copak tě k nám přivedlo? Většinou do Hodonína se jezdí na nákupy komponentů do počítačů... Jinak já mně dnes napilno, na balkóně jsem pájel jak vzteklý... takže bych moc sdílný nebyl, ale příště je klidně možno se zastavit. Rád poznám autora tak hezkého prográmku jako je HWiNFO
A angličtina mi naštěstí nedělá problémy, takže snad to někomu pomůže...
Mimochodem, 10 obrázků na post? Ale no tak...!

 

[Martin]

This board is a bit tricky and the problem is that sensor values between BIOS and Windows can be slightly different, so it makes it more difficult to find proper adjustments. Maybe a side-by-side screenshot with the MSI tool showing sensor values might be more helpful for me...

 

[trodas]

Hmmm, that would be a bit of problem, because I realized that nothing like that can be found on the MSI page for the PM8M3-V mobo:
http://www.msi.com/product/mb/PM8M3-V.html

So I got to ask on MSI forum about it, maybe we got some answers. Or should I ask MSI about the sensors...?

BTW, I find how to configure the sensors page on HWiNFO to what I like to see - and make these changes "stick":

So, hoooray! Yet the fan speeds and +12V values are missing.

If we did not get any support from MSI tool, is possible to do some advanced scan and report more values out of the mobo? I did not trouched the "advanced" parts of HWiNFO config at all.


PS. WTF!!! Since I started Everest, the HWiNFO started registering the fan's! W/o any of my action, just... like that. Glad I posted the screenshot w/o the fans before, lol. Now I have a proof

...but I also have something much better! As I mentioned earlier, there is no MSI tool (or I did not know about one), but Everest working like a charm:



...of course, it should be reporting also -12V and -5V values, but that is a minor detail (and the -5V is not even reported in the bios and the +5V is obviously wrong one, correct one reporting HWiNFO ). The most important thing is, that Everest (unlike some pesky MSI wanna-be-tool) allow to generate a compete report! So now there you have it, included

If some experiments are need, then I'm more that willing to do it and test some alfa version (email?) - just let me save my work first, before the experiemental reading will freeze/reboot or at least crash my Winblows
(Hell, I'm scared to even run HWiNFO on the Sapphire A9RX480 "Grouper" mobo, lol... I never did that yet...)

 

[Martin]

I'll try to adjust most of the voltages, so that they're close to BIOS values and check why fans are not displayed. I believe Everest doesn't show proper values for voltages, especially +5V and +12V.
The change in HWiNFO will be available in the next Beta build released. I see that you have hidden some values in HWiNFO sensors, so when using the next build make sure to show all voltages again, since some of the hidden ones might show correct values then.

 

[trodas]

Thanks for every try. Now I can confirm, that unless I start before the Everest, then HWiNFO ignore the fans. Interesting...

Replacing the GFX card from Radeon 9100 to GeForce 6800 GT - and much to my surprise, even HWiNFO detect correctly the clocks of the card, it doest not detect the temperature sensors on the card and there is even two ones - one for core and one for the card itself, so we know how hot the card is:

 

[Martin]

I plan to release a new Beta build within 1-2 days and that should fix the fan issue. I will also improve reporting of most voltages (though not all of them) for this mainboard.

As for the GeForce 6800 GT, that's correct. NV4x and G7x GPUs require a very complex method to read temperatures and this is not supported by HWiNFO.

 

[trodas]

Spunds great Can I just replace the exe or wait for the standalone (portable) version? :huh:

And indeed the temperature from 6800GT is interesting, because Everest report even THERE temps of the card - core, VRM and "ambient" (or, more precisely, the temp of PCB of the card, witch is far different from ambient temp, lol).

Interesting - too much sensors, lol. Even GPU-Z can read two of them - time to ask the autors how do they do it?

 

[Martin]

You can replace the EXE, it's the same sort of as in the portable version.

As for the 6800GT, HWiNFO should be able to recognize those additional sensors (GPU, GPU Ambient and GPU VRM). Do you have the "GPU I2C Support" option enabled in HWiNFO? Also, please attach a new debug file when using this GPU and make sure it contains sensor information (make sure you open the Sensors windows when making the debug file). I can then analyze if HWiNFO is picking up those GPU sensors.

 

[trodas]

Thanks for the heads up about the portable & exacutable

To me it looks like perfect right now:

Of course I run Everest before, so it might be affected, and I will it run again w/o previously starting Everest. Yet all voltages and fans looks great and fine - and there are ALL THERE temps from the GFX card, so this is terrific!



Great work, Martin!

Including the debug file too, yet more interesting it will be tomorrow without the Everest Yes, as by default, I do have the GPU I2C Support enabled and it should use the NVAPI for the job.

 

[trodas]

Of course, nothing is perfect I have to report two bugs. First, the GPU temps are way off, like 10°C:



It would be cool, if they are at 34°C, but Everest claim 44°C and same will the nVidia control panel, so, even that I would like the 34°C more, it is not happening. The card itself got 37, 38°C, so... Sadly not possible to have the GPU at 34°C, tough I would like that

Second problem is GUI problem. When I started to organize the GUI, then when I click too fast on the most low sensor to get it up in the table, it disappear! LOL
Just annoying how many times it took me to just get the table to the way I like it, lol. Shit happens, but it should not in the GUI. I also on the 6800GT VRM temp noticed that sometimes it just flick there with WEIRD value, like 9999 or such... So definitively a GUI quirk.

On the another hand, your voltages are much more BIOS correspondent AND you got right the CPU fan - Everest swap the CPU and chasis (now used for the GPU 120mm Noiseblocker fan connection) fans, while HWiNFO got it right

Easy to check, who is right and who is wrong - started Prime95 run torture test for the poor CPU and watched witch fan go faster. Because the CPU one is regulated, then it is this one, who change the speed - while the "chasis" fan run on constant speed around 1000 rpm - and since it now spin at 1513 rpm and the other one is still 1000 rpm (1016-1035), then Everest is proven wrong, lol.
HWiNFO wins, hands down

 

[Martin]

Yes, it seems the GPU temperature requires an offset, but I currently have no fix for this.
As for the problems with sensor GUI, I'll have at look at that. It seems there are some issues with the new features I introduced and I need to polish it a bit...

 

[trodas]

Yep, the GPU core temp is 10°C off the mark. Dunno why, but it is.

And indeed, there are some troubles. Todays at boot, I run HWiNFO and all the there temps of the GFX card are gone But on the other hand, the fans are sticking and kicking now - hoooray!
(before they requre Everest to be run first, lol)

Weird is, that suddently and abruptly the machine shut down. This is not happening, the machine has replaced all caps (okay, one 10uF SMD suxxka near AGP slot is not yet replaced with quality caps), used a high quality Nichicon caps and even polymer caps for the CPU VRM and IO/NIC powering, so it is really stable.
(except for some AGP things, when x8 speed the mobo clearly did not like much, but that is separate issue - tought it might be related, because the GFX card sensors are gone...)

Let's hope this is not a new HWiNFO feature, but it happen for me for about first time. Even with bad caps in the mainboard is the mainboard working relatively properly. So... some polishing is really in order At least the fans now work, hooray!

 

[Martin]

Hard to say if that shutdown might be related to HWiNFO. It that would happen again, I might need to check the Debug File produced during such crash.

 

[trodas]

Sure thing. I must admit that altrough I working hard to make the mobo stable, it is not stable in 3D games. In the end of making it stable it is not stable even in 2D Windows... lol.

To cut long story short - at first, it was just a mobo full of known bad caps that sometimes refuse to boot and losing clock AND bios settings when power is cut off. Sometimes it also crash during SoF2 gaming, but that it is.
So I replaced all the caps, put even these that aren't there back and used hi-grade caps. For example the best polymers Nichicon LE for Vcore and for voltage supply there are Nichicon HZ caps - the best elyte caps even made - king in terms of ripple current, only Samxon GA is "par to par" with them, but that it is.

Result - still not stable. So I blamed the Radeon 9100 with questionable caps and added 120mm fan blowing on it.
Still no help.
So I replaced the Radeon 9100 with GeFroce 6800GT, that have only 3 caps (Chemi-con polymers) and worked well for me before.

Problems get worser. Even exiting SoF2 or Quake3 is impossible now. It get reasonably stable for playing, yet try save screenshot and it crash ASAP. So I try lower the AGP rate (allow me to choose only 8x or 4x), disable Fast Writes, disable AGP Master 1 WS Write/Read and even disable AGP 3.0 Calibration Cycle. No help.
So I toyed with the voltage, raising from 1.55 to 1.70 and no help either.
Another thing that I added a huge nice heatsink on the NB, because the original one was pretty small and useless. And I discovered beneath it, that there was just a dip of white paste widely off center and that was it. Most of the heatsink did not even have contact with the chip...!
So I tought - I got this! Laped a BIG heatsink, Arctic past used and screwed it tight on the NB... and again no help.

So the last thing to toy with is the AGP Driving Value. It can be changed from 00 to FF, defaut on Auto show DA.
Dunno what will happen, but I run out of ideas right now.

So that is the story of this mainboard AND GFX card stability. So in this view it is quite possible, that accessing GFX card could cause crash.
If there is a interest, I can post some pics of the recapping and other stuff around the old mobo.


PS. starting to play with the AGP Driving Values and it seems to help, being able to play for some time at AGP 8x speed with 1.55V only and all on, except Fast Writes. Bumped also the AGP Aperture size to 256MB... But there is one good thing:
Since the heatsink is HUGE and absolutely uncomparable to what was there before, then it is obvious, that the 47°C temperature is the CPU VRM temp, not the mainboard NB temp. Because mainboard NB temp _MUST_ be lower, when such huge heatsink is added on the NB!