Defective Mobo - What to do?

[parsec]

Re: Defective Mobo - What to do?

Ok and I just checked my bios, it seems there is one fan running and the others all say N/A so I guess I don't have them.

But the cpu temp according to the bios is 50 c, and the M/B temp is 37 c, so the M/B temp is SYSTIN I guess.

50 seems pretty high to me, and its somewhat higher than the CPUTIN, which is only about 45 c or so.

This is after running for about 2 hours. It is the SYSTIN value that psensor doesn't like.

50C doesn't strike me as high -- people get worried when things get up to 80C or so, and 100 is the real danger zone...

SelfGovern, remember MBwoes is using an AMD processor and chipset board. AMD processors do not have the same max temperature ratings that Intel processor do, such as the 80C and 100C you quoted above. Given what I have read, 70C for a FX-8350 is a hard maximum.

If the processor temperature is 50C in the BIOS, that is rather high IMO, and indicates a stock CPU cooler. We also have no idea what the CPU fan speed configuration is in the BIOS, not to mention everything else. Given the recent description of the PC case cooling in this thread (hardly any, which is after adding fans...), the PC's cooling is lacking.

The "Psensor" readings or graphs are virtually impossible to translate into which readings are which without more information. If that software does recognize the 8350, and is creating temperature alarms, then excessive heat is an issue here.

This board has software for reading CPU and other temperatures, ASRock Extreme Tuning Utility has that as part of its features. Free to download on the board's download page, and all readings will be labeled correctly. If the fans being used in the case are not connected to fan headers on the mother board, their speed cannot be displayed by any software.

There are many other good hardware monitoring programs, like HWiNFO: HWiNFO, HWiNFO32/64 - Download

Given the OP's comment about the faulty temperature sensor, etc, his PC building experience is wanting, sorry to say.

[MBwoes]

Re: Defective Mobo - What to do?

Guys thanks again, I am going to explore this warning I am getting, not getting it on the identical one which works great (the third of the batch).

I agree with SelfGovern, think it could be the heatsink fan combo, which is not listed as running on mine but is on the one which is perfect. Mine has no fans shown as running, that one has fan 2 going hard. The fan is running of course, but if it isn't showing up on the monitor, I wonder if it is defective or running really slowly or whatever. The good one hasn't built up any heat at all, heatsink is good to the touch, and its been on for at least 9 hours at this point and possibly days.

No mods or anything, I am too junior for that.

The one at my home is running pretty damn good by now, one crash in 3 days. So its probably just a little excess heat, and the better power supply has reduced the heat which is how it got so much better. There was one night with repeated crashes, but heat would explain that better than anything.

Will report back soon. I think the lesson here is pay attention to the power source and the heatsink/fan. Especially when you pick out a performance mobo and seriously fast processor ;-)

[SelfGovern]

Re: Defective Mobo - What to do?

My apologies -- didn't realize that there was such a difference in heat sensitivity between AMD and Intel processors.
Interesting coincidence, I have known that HP's latest servers can run in much hotter data centers than previous ones,
and now see a similar announcement from Lenovo -- ambient temps up to 40C will certainly help with energy use
as these servers become widespread.

Woes, next time you're in the market, do look at business-class machines, which are often supported running Linux,
and which will save you a bunch of headaches and time -- and the manufacturer stands behind the complete system,
lots fewer problems, and if a problem does come up, there's no finger-pointing by vendor A saying it's vendor B's
component's fault. Workstation-class machines from HP (as an example I am familiar with; I would guess others
have their own offerings) also come with high-performance components such as Xeon processors, SAS SSD or hard
drives, etc.

[MBwoes]

Re: Defective Mobo - What to do?

SelfGovern:

I do know what you mean, it certainly is a lot less risky. In fact that's what I did when this problem came up, because I needed working machines, so I bought a couple Acer comps with the AMD a8 cpu, and advertised for linux too, I was a bit shocked by that! They have been running fine and seem more than fast enough, but I haven't pushed them to the limit yet.

But, I still love to build a custom system, this is the 4th or 5th time I have done it, and first time to have trouble. I researched motherboards for 2 weeks before choosing the ASRock Extreme 3 R2 board, it has everything I wanted, USB 3 support, fast internet and so on. I knew I wanted the AMD 8350 8-core processor because I helped make a comp for my friend last year with it and his is super fast.

Guess what, I think we found the problem, and you were right! The fan on the heatsink has a problem!

The fan on the good one (which I asked my staff, she hasn't been turning it off at night, so its been running for days now) is going at about 6500 rpms, the one one mine is going at half speed (3300 rpm) and doesn't even show up as running in the psensor, but i can see it in bios.

Here is a pic, the left one is the good one with the fan going. ALL of the max temps are HIGHER in the good one, except for AUXTIN, which is crazy higher in the one with problems. 8c vs. 23c!

That means the AUXTIN temp probably is causing the freezes, I feel pretty sure about this. Yes, there could be yet another problem, but now I KNOW my fan is defective, so of course I must fix that first. The other components are not getting seriously stressed because of the shutdowns. BTW the difference in the two charts is probably because one is Mint 17 and the other is Ubuntu 14.04.

So, my guy must not have worried about these little issues, and just threw stock power and heatsink on there, and we got a little unlucky and got a broken fan and maybe a unstable power supply as well.

Looks like parsec was right, it was a cooling issue. Thanks parsec!

[parsec]

Re: Defective Mobo - What to do?

Sorry, not seeing a picture of anything.

The difference in CPU fan speed can be caused by a BIOS setting, in the screen that has the fan speed control options. That should be checked. We don't know yet if the CPU fan is defective or not. You should check the setting of the CPU fan speed control in both of those PCs.

The AUXTIN temperature, whatever that sensor belongs to, is an odd reading. 23C is a normal room temperature, 8C is cold and cannot be a valid component temperature. How can any component temperature in a PC be below room temperature? Sensor readings may need to be adjusted before the true value is determined, that might be the case here. It might also be an invalid, unused sensor reading.

I'm aware that that AMD processors can have different temperature readings than Intel processors, at least when they are not being stressed or heated up.

A mother board's temperature sensor chip normally has a few readings of sensors that are not used, and have random, false values. Most programs that read those sensors have no idea if the values are valid, they simply display them.

Since you are using Ubuntu, your options for monitoring programs is limited. Please post the picture you mentioned above again, it is not in your previous post.

[MBwoes]

Re: Defective Mobo - What to do?

Ah, sorry parsec I forgot to attach. Here is the pic, sorry I can't upload so pls use the imgur link:. imgur: the simple image sharer

Here is a pic, the left one is the good one with the fan going. ALL of the max temps are HIGHER in the good one, except for AUXTIN, which is crazy higher in the one with problems. 8c vs. 23c!

More freezes recently, I left it off all night last night and then had a freeze within about 30-45 minutes of running this morning, and two days ago it froze twice while watching a south park episode.

However yesterday it ran all day through relatively high load for 12-15 hours and it did fine. Still seems to be either video or the browser which does it. Even virtualbox doesn't seem to push it over, at least not that I can remember.

I just put all values back to defaults, and also noted that the setting for the fan was full on, so this fan is defective for sure, it is changing speeds and the setting is full on. The sensor for all of systin cputin and auxtin is lmsensor nct6776-isa-0290. It is systin that is complained about on startup of the monitor, every single time, and never pops up on the good one. So I gotta change that fan for sure which is clearly not working properly, maybe that's the warning, as its not showing up in the monitor and not running consistently.

When I turn on and go into bios, I can see the cpu temp is like 47+ degrees. I think I tried that before from a cold start. Its very comfortable here, about 22 degrees and air is moving. Today I checked after the crash, only running for about an hour or less. But I do remember SelfGovern explained how the heat can build up quick.

[MBwoes]

Re: Defective Mobo - What to do?

I think I have to change this gpu, I have a new one ready. It doesn't even seem to have a fan on it, it could be a hotter gpu...

[parsec]

Re: Defective Mobo - What to do?

Ah, sorry parsec I forgot to attach. Here is the pic, sorry I can't upload so pls use the imgur link:. imgur: the simple image sharer

Here is a pic, the left one is the good one with the fan going. ALL of the max temps are HIGHER in the good one, except for AUXTIN, which is crazy higher in the one with problems. 8c vs. 23c!

More freezes recently, I left it off all night last night and then had a freeze within about 30-45 minutes of running this morning, and two days ago it froze twice while watching a south park episode.

However yesterday it ran all day through relatively high load for 12-15 hours and it did fine. Still seems to be either video or the browser which does it. Even virtualbox doesn't seem to push it over, at least not that I can remember.

I just put all values back to defaults, and also noted that the setting for the fan was full on, so this fan is defective for sure, it is changing speeds and the setting is full on. The sensor for all of systin cputin and auxtin is lmsensor nct6776-isa-0290. It is systin that is complained about on startup of the monitor, every single time, and never pops up on the good one. So I gotta change that fan for sure which is clearly not working properly, maybe that's the warning, as its not showing up in the monitor and not running consistently.

When I turn on and go into bios, I can see the cpu temp is like 47+ degrees. I think I tried that before from a cold start. Its very comfortable here, about 22 degrees and air is moving. Today I checked after the crash, only running for about an hour or less. But I do remember SelfGovern explained how the heat can build up quick.

The lack of decent sensor readings is making it difficult to determine what is going on! No offense, just the lack of programs (I guess...) for that in Linux. But I think we have some data to go on now, given what you wrote.

First a quick clarification, the boards you have are ASRock 970 Extreme3 R2.0. Your first post in this thread says ASRock 970 Extreme R2.0. It's possible that different parts of the world have different products that aren't showing up on the ASRock website for the USA, for example. Just double checking.

Also, the best FX 8350 stock cooler picture I could quickly find is this, for reference purposes, does it look familiar:



We can't see the fan's connector, but we can see four wires, so we know it is a PWM speed controlled fan, that is best used on a four pin PWM fan header on a mother board.

The lack of any fan speed reading for the troublesome PC in your picture of the sensor readings could also indicate a bad fan. But a few other things are possible:

The fan on the CPU cooler is not connected to the CPU fan header, or the appropriate CPU fan header. From the board's specs:

2 x CPU Fan connectors (1 x 4-pin, 1 x 3-pin) 3 x Chassis Fan connectors (1 x 4-pin, 2 x 3-pin) 1 x Power Fan connector (3-pin)

All to easy to connect the fan to the wrong header, particularly since the three pin header is also labeled CPU. The fan header currently being used needs to be checked.

You mentioned the CPU temperature in the BIOS, you should also be seeing fan speed readings in the BIOS, same screen actually. If you don't see a CPU fan speed reading, either the fan is connected incorrectly, or is defective.

As I said earlier, all the sensor readings shown in the program you use are not valid, that is true for any mother board. All the sensor inputs on the sensor chips are not used, so any data they provide is random and false. The AUXTIN readings are as low as -1C, and -2C, how is it possible that a component temperature is below freezing, 0C? If you don't see a corresponding temperature reading in the BIOS, AUXTIN is a false reading IMO.

Why does one PC show a GPU temperature, but the other doesn't?

Your own experience with the freezing indicates it may be caused by something else. The Adobe video player in my browser would crash all the time, until they fixed it.

[MBwoes]

Re: Defective Mobo - What to do?

Hi parsec, you were right! The fan was plugged into the 3 prong cpu2 header rather than the 4 prong, the other two both were done right. So I fixed it and it shows up in the sensors now, running at about the right speed (although actually its still a little lower than the good one tbh).

Still, I think overheating is the real problem, we talked about this before. Psensor is still giving me a warning on startup, but its a different sensor now, its telling me the mobo is too hot at 34c. I just saw another thread with a very similiar AMD cpu and you told him the mobo was too damn hot, my numbers are pretty close to that.

When I fixed it, I went into bios to check the speed, and the cpu is listed at 50c and higher, and this is just on startup at an ambient temp of like 23c.

So, let me ask you, now that the fan seems to be working normally, is it worth it to change the fan, or should I go all the way and change the heatsink too? I guess my question is, in cases like this of overheating, which of the two is more likely to be the real problem?

I still have the backup gpu I bought, which I still plan to try out. I have this new one and two older ones that should work, and so, maybe just maybe its worth trying.

[parsec]

Re: Defective Mobo - What to do?

Glad you have this temperature problem at least understood at this point.

Just for reference purposes, if you check the newest high end CPUs from AMD, like the FX 9000 series, which are basically based on the FX 8350 but running at higher stock clock speeds, what CPU cooler is used? An All In One (AIO) liquid cooler. That is a huge jump from the stock '8350 cooler.

Looking back on the FX 8350, and knowing what we know now about the new FX 9000 processors, the stock '8350 cooler is only adequate in situations where the rest of the PC's configuration and cooling is done very well. That is, the PC case has good ventilation, with a lot of cool air coming in and heat created inside the PC case being removed.

The design of the video card's cooling is another major factor. Some cards exhaust at least some of their heated air out the back of the PC, but many cards simply dump it into the PC case. That warm air is then pulled in by the CPU cooler fan, which does not help cooling the CPU. If the PC case is not ventilated well, the heat builds up and everything is too warm, including the mother board.

My guess is this is why there is a difference between the three PCs you built. Are they all identical, same PC case, ventilation, and video cards? If not, that explains why some run fine but the other one does not. It's worth checking the temperatures of the other PCs, as they may be borderline, unless you've already done that. Also, are they all used for the same thing, do they have identical usage profiles?

Now about your question, which is a good one. Although we don't know the CPU temperatures of your other PCs, it is always better to upgrade the CPU cooler rather than simply change the fan on a stock CPU cooler. Given the size of the fan on that cooler, which cannot be larger than 80mm, finding a better one that is also four pin/PWM speed controlled will be virtually impossible. If the fan is smaller than 80mm, then your choices will be even fewer and really a waste of time to even try.

One thing to consider, is if the other PCs have reasonable CPU temperatures with the stock AMD CPU cooler, then the mounting of the cooler on the over heating PC may be bad. A fan speed difference of ~100RPM on fans that size and maximum speed is normal. So getting an identical replacement fan won't solve the problem IMO. Also the things I described above about the video cards and the PC case ventilation may be the cause of the heat issue in that unit. The various parts of a PC don't operate in complete isolation of each other, and the environment inside the PC case is one they all share.

A good but not to expensive CPU cooler that would definitely be an improvement would be a CoolerMaster Hyper 212+ or Hyper 212 Evo. Of course it is much larger than the stock cooler, and whether or not it will fit must be researched.

[MBwoes]

Re: Defective Mobo - What to do?

Yes parsec I think you are right that heat is the real issue. My video card doesn't even have a fan on it at all! I didn't think about it when I picked it out, didn't skimp on price really, just didn't know.

Thanks for the detailed info about the heat sinks. So far since hooking the fan up correctly everything has gone well, no crashes so far. So I will hold here for now to see if this makes the difference. Recently crashes were happening more often, so this is nice.

If I need to keep going, the gpu is next. But need to wait and see if this is really better or what.

[MBwoes]

Re: Defective Mobo - What to do?

Ok parsec! Its been 6 days, and not one crash. Last night I really tested it out, stressed the system to the max, and it went fine.

I also put those case fans in there to really confirm the heat is going to be removed quickly, those things helped a lot, should have put those in there weeks ago. Two on the side pulling out drags so much fresh air over the board, its great. I kind of want to get another one coming in from the back to push air "across the board", but I don't know if I have another header to connect it to.

Recently, even with the fixed fan, I would see my gpu getting up into the high 50s and even low 60s, and while I haven't had a crash in a week, that's still too high too high for an AMD system. And I have learned not to stress the system after all these problems, so I wasn't seeing the real picture of typical use. So while I was staying under threshold for freezes, the problem really wasn't fixed 100%.

But the case fans really pull the heat out quickly, so while the gpu or cpu will generate a lot of heat under stress, the heat doesn't build up at all.

To be honest, I probably should replace the gpu with one with a fan and a lower heat footprint, but I am lazy and things are looking good right now...

Thank you parsec and SelfGovern for all the help, I think my problem is fixed! Now on to the other machine, gonna get that going as well!!

Also, I want to show what is going on with the psensor, the results seem to make more sense now, and by the way just for reference, psensor still always gives me an error on startup, now it complains about the temp1 value. But, psensor is still at an early stage of development and that doesn't worry me much.

Here is what things look like now:

Here, notice that several of the values go way up into the 50s, but the SYSTIN value stays down in the 30s at all times. Also notice the fan working responsively to lower temps. What was weird about the info before is that the working computer at the office had much *higher* levels for a bunch of components, which seemed strange, the working one is hotter is some respects than the broken one. And I focused on the AUXTIN value, but that seems to be some ghost value, even going down into frozen territory, it can't be trusted.

Let me compare with the old info, the two computers side by side, the working one at the office and this one before the fan was fixed, I need to make a new post to attach a second picture.

[MBwoes]

Re: Defective Mobo - What to do?

So here is the old pic and my analysis:

Here, the working one has similar values to the situation now, with many components getting up very high, but this one when it was broken, notice none of the max values get up into the 50s at all. That, I guess, was because the system was getting overheated before any of the components could really get going. But notice especially the SYSTIN values.

In addition, it looks like the office computer is also under a lot of heat stress. The fan is going at much higher rates than the fixed one, and that was true even before I put in the case fans. That system gets some stress because we run virtualbox, but not quite as much as my home computer though. But look at the SYSTIN value, quite high, so I will need to get more case fans.

Anyway, thanks again!

[parsec]

Re: Defective Mobo - What to do?

Great, so glad you worked this all out!

The PSensor error on start up is probably just a bug in the sense that the temperature value that causes the error message is set so low that it displays the message when it's probably not valid.

While a passively cooled video card is unusual, actively cooled video cards (with fans), both Nvidia and AMD/ATI routinely go above 60C under heavy load. The specs of your particular card I don't know, but I would not be surprised if it is normal for your card to operate at 60C or more.

I mentioned earlier that all the sensor readings you see are not necessarily true readings of a component. Then of course there is the question of what the generic name (SYSTIN) is referring to. As I also said, besides CPU temperatures, unless the programmers research what each mother board uses for the sensor connections, they can only use generic names for identification.

Now you can see why business oriented PCs usually don't use high end processors when the PC is sitting next to a desk. The user does not understand why they hear fans speeding up and slowing down.

Anyway, keep those fans running and watch those temperatures!