Urgently need computer savvy experience to determine does my CPU have a problem. Ref overheating ?s

OK I installed my Xeon E5 1650 V3 using a Noctua NH-D15 with two 150mm onboard PWM fans. The PC case deliberately is clear for airflow from the right intake to the MB, except a Samsung evo 850 bolted to the side wall. In addition there are 5 more 140mm Noctua PWM fans and 1 more 120mm fan. ( 3 intake and 3 exhaust ).
In Summary I imagined this would be massive overkill. See end - for airflow concept.

But when I turn on the PC the CPU hits 40C in a few minutes, and the PCH is at say 36C. Hmmm not so happy, I am just in the Bios looking around so no heavy workload at all.
I was surprised that with such good airflow, and a processor only cranking at 3.5ghz it was so hot, Honestly I was pretty disappointed the CPU temp was not nearer room temp!
Over the next 20 minutes the CPU temp will gradually rise to 52C and the PCH to 40C, and then all fans hit 1,500 rpm until the temps come back down to CPU 44C and PCH 37C.

If I tweak the PMW settings to Full Power all the time i.e. 8x fans at full tilt eg 1,500 rpm then it will maintain CPU at 42C and the PCH at 37C but TBH its unpleasantly noisy for the money.

Intel Ark says max temp at the IHS is 66.7 C
And the CPUs TDP is 140 Watts

Q1 is this too hot for a Xeon, basically doing nothing at 3.5 ghz i.e. Just looking at the bios settings!
Q2. Am I right that when the Std. turbo mode kicks in taking the CPU to 3.8 Ghz it will get hotter and the voltage supplied will be bigger.

The most worrying part for me is the CPU Voltage shows between 1.42V and 1.473V - note these are all at 3.5Ghz!
When the standard Turbo mode ( 3.8Ghz ) is required will the Voltage supply also go up?
Intel Ark says the correct voltage can be between 0.65V and 1.3V. So Even at a basic clock speed its way overvoltage, I would have hoped to see 0.66V or 0.7V considering the low workload and, that its running at its base specification not turbo mode.

I spoke to Intel support - just got the wrong person. Less knowledge than me!!
What I wanted confirmation of was :-
1. As the CPU speed remains / is rock solid at 3.5GHz, and one sees the temp and voltage vary a lot, is the Bios doing its job properly? Namely sending enough power to keep the CPU at 3.5ghz ?
2. Say the Bios was faulty and I updated it, and then The voltage regulator then only sent say 0.7 Volts to the CPU as not much is being done. Would that mean my CPU would slow down?
3. When it goes into Turbo mode , I assume the rise in speed will be matched by a corresponding Voltage increase?
4. If Intel say the max. voltage the CPU can operate at is 1.3V , is 1.472V harmfull long term?

Any prompt advice very much appreciate - should this be swapped.

In summary can I argue ( in Computer speak ) that under no load the CPU is out of spec, and its highly likely that when it goes into the standard torbo mode it will only get worse? Add some heavy video rendering and this will be come a real problem?

Planed Air Cooling Concept - backgrounf info:-
Was to let lots of fans run slowly ( Noctuas go down to 320 rpm ) where they are at there quietest; but have the capacity to create a hurricane if required. Also By using PWM one could let fan management target specific areas, say boost the CPU cooler fans speed and still run the case fans slowly.
And it works just as I hoped bar the 2 problems above . By that, the fans can all start at 300 to 350 rpm = beautifully quiet
As the CPU temp rises, then the CPU fans may speed up to say 900 rpm and the other 6 case fans stay at circa 300 rpm. Still a very pleasant noise level.
When the CPU gets very hot, all fans hit 1,500 rpm until the temp drops, and the cycle continues. Quite loud all at 1,500 rpm but that does not last for long, and anyway extreme cooling is reuired ... so the noise level is justified.
Perfect I am a happy bunny withthe way airflow works.

1. Why did you install your cooler in a vertical orientation? Fans are better at pushing than pulling in my experience.
2. What thermal paste and how much did you use?
3. What are you using to monitor voltage and temp?

Hi

1. Why did you install your cooler in a vertical orientation? Fans are better at pushing than pulling in my experience.

- I needed the extra 11mm or it would have gone right up to the 1st slot. In their wisdom SM engineers didnt think one would want to desperately use one of three Gen3 x16 slots!

2. What thermal paste and how much did you use?
Noctuas NH1 - came with Cooler - non conductive and had good reports
I used a pea size, measured as best I could at about 4.5mm diam. The instructions said 4-5mm diam. So 4.5mm was a balance between using as little as possible but on the higher scale as the 2011-3 cpu is much bigger physically than the i7 Core ones.
I spread that lengthwise a litlle like a a grain of rice as the CPU is slightly longer than wide. Put the cooler on as square as possible and bolted it up a few turns ine side and then a few turns the other side. And used a torque screwdriver to 0.4 Nm - as Noctua use that setting to tighten the cooler to the metal securing plate. The springs did compress fully somassume thats the best I could do. Surprisingly Noctua had not worked out the perfect torque for doing this up.
0.4 was also the number Supermicro said they used to secure the socket to the MB.

3. What are you using to monitor voltage and temp?
I am using the bios.

Check that core voltage! (vcore) Most current cpus default around 1.0 to 1.2v.
I overclock my 4770k to 1.30v, and with a similar (archon) cooler, idle at 35c.
Are the top two fans blowing air IN? Same direction as the cpu cooler?

Next to check is the heatsink mount. Pull it off, and check how the paste spread out. Current thinking is to spread two *thin* lines in an 'X' pattern to get the best coverage. Then clamp that heatsink down tight. I actually added shims to get mine even tighter.

Next, make sure all case fans spin at a reasonable speed, to exhaust the heat from the case. Make sure the case isn't building up heat under a table/corner, etc.

If the Vcore voltage is 1.42V - 1.473V you need to change it in the BIOS. That is way to much Vcore unless you are overclocking with liquid nitrogen. With everything setup properly the CPU should run at a lower speed and voltage at idle, the CPU will automatically speed up and draw more voltage as needed, but should not exceed 1.3V unless you set it higher for overclocking. Basically, you need to setup the CPU properly in the BIOS.

First steps I would take. Check for/update to the latest BIOS available, reset BIOS defaults and finally setup the BIOS properly for your CPU and RAM.

[QUOTE=skid00skid00;17​908661]Check that core voltage! (vcore) Most current cpus default around 1.0 to 1.2v.
I overclock my 4770k to 1.30v, and with a similar (archon) cooler, idle at 35c.
Are the top two fans blowing air IN? Same direction as the cpu cooler?

Next to check is the heatsink mount. Pull it off, and check how the paste spread out. Current thinking is to spread two *thin* lines in an 'X' pattern to get the best coverage. Then clamp that heatsink down tight. I actually added shims to get mine even tighter.

Next, make sure all case fans spin at a reasonable speed, to exhaust the heat from the case. Make sure the case isn't building up heat under a table/corner, etc.[/QUOTE

Thanks
Check that core voltage! (vcore) Most current cpus default around 1.0 to 1.2v.
I overclock my 4770k to 1.30v, and with a similar (archon) cooler, idle at 35c.
Are the top two fans blowing air IN? Same direction as the cpu cooler?

Top 2 fans are sucking air out, i.e. In the same direction as the cooler which is blowing upwards

Next to check is the heatsink mount. Pull it off, and check how the paste spread out. Current thinking is to spread two *thin* lines in an 'X' pattern to get the best coverage. Then clamp that heatsink down tight. I actually added shims to get mine even tighter.

>> Will do that next

Next, make sure all case fans spin at a reasonable speed, to exhaust the heat from the case. Make sure the case isn't building up heat under a table/corner, etc

>> Fans are set at full speed circa 1,500 rpm
And the case temp is 28C tested with one of those remote thermometers.

-dave-m- wrote in post #17908847
If the Vcore voltage is 1.42V - 1.473V you need to change it in the BIOS. That is way to much Vcore unless you are overclocking with liquid nitrogen. With everything setup properly the CPU should run at a lower speed and voltage at idle, the CPU will automatically speed up and draw more voltage as needed, but should not exceed 1.3V unless you set it higher for overclocking. Basically, you need to setup the CPU properly in the BIOS.

First steps I would take. Check for/update to the latest BIOS available, reset BIOS defaults and finally setup the BIOS properly for your CPU and RAM.

Thanks
Just what I thought. The board came with Bios version 1.0b. Looking on SM support there seemed to be a ver2.0 - somI explained the problem,mand doublemchecked the ver 2.0 was for my board as there is a very similar X10SRA -F ( a remote management version , mine is -O ) the supprt tech checked and sent me the link to make sure . It was the same one I found. So loaded it, and got the same heat and voltage.

Before getting hold of SM support I spoke to a specialist server company. They do a lot of work with SM and Xeon's. He selected the same bios. And said a lot of SM server servers have a super user that allows one to open up the Bios to change the voltage and the frequency??.

The new bios is very different and does have the Superuser button but it does not operate here. So There is no way to change the actual voltage.
Imwent around to a mates and he has an MSI Deluxe. Same sort of thing but you have tons more controls, like Voltage!

He came over and agreed no way to change it.

Now I am worried by the time SM find a fix I am concerned this CPU will have been damaged. As my friend said to add 0.175 vots to a CPU's top Voltage would give you a thumping overclock. Something In Dont want, I want longevity, unstressed reliability! :
He has taken his i7 6700K from 4.2 Ghz to 4.6 ghz but he has massive dual radiator water cooling crap.

One thing I didnt do was reset it to the Optimum Defaults, as I thought flashing it from 1.0b to 2.0 would do that as standard. Best check though.

How bad do you think the damage would be running at 1.42v where the max is 1.3V BUT the temperature is 48C where they say the max temp should be 66.76C . As its not overheated does that mean no damage done??
Or just as I suspect no catostrophic damage yet.

As this is a Supermicro certified processor and they allegedly are Xeon server experts. One would have thought, if they provide no tweaking controls, they should have built a pretty "conservative" Bios setting.
Hence could I just have a duff CPU?

Yes I will check the thermal paste, but my friend said as with all the fans on ( apart from being horrendously loud ) and all at 1,500 rpm. He thinks the TP must be done pretty well, or at 1.47V he said I could have been looking at 75C to 80C until it burnt out.

I assume one can reflash the bios withthe same code again?. Thought I might do that as it did say sucessful and then seemed to reboot before it came to the c:\ prompt.
They warned me to wait for this at all costs.

2 other weird things
A) The mouse was fast on bios 1.0b but hung after about 20 mins in the old bios. And the keyboard would freeze after another 10 mins or so.
With the new bios ver 2.0, its like running in slow motion, and not very accurate - it was very snappy and precise on the old ver 1.0b
Its a standard MS 6000 wireless mouse.
The mouse has not hung yet but due tomthe voltage I tried to get out real fast.

B) its now found the Intel PCIe Nvme SSD, and even offered me the opportunity to Boot from it!

But under the I/O , PCIe setting there are 5! Not one as in reality!!
See below

I set the Intel 750 and the Asus PCIe to M.2 adapter cards to EFI.
And the boot mode to UEFI. Normally

To do the Bios flash I set the Boot mode to Legacy.

Should once the Bios is loaded, the GPU, and Intel Wi Fi cards be set to EFI or Legacy.

Slot 2 has 8 Lane one redirects to Slot 4 where the GPU is. So that now becomes x16 ( lanes and physical )
Question is should that unused slot 2 be set Legacy or EFI or disabled?? Bearing in mind its redirecting the lanes to slot 4.

Likewise slot 3 is overhung by the MSI GTX970 but unconnected - should I set that to disable are leave as Legacy.

Or should as I am useing a UEFI , set everything to UEFI or actually EFI in this case.??

Man this is complicated for a novice!

What's your room temperature?

If it doesn't hit max (66.7) you're probably ok. My older i7 has gone from 33 to 48 since it was last turned on, and it has only done about a minute of image processing, but it's max is much higher - 90c or so.

You probably haven't damaged it. Run Prime95 overnight to check. It's not conclusive but it's good. There will be plenty of CPU testing programs around.

Also run HCI memtest to check your RAM. Each instance can only address 2GB so you will have to run many instances. If you have 16GB of RAM, and 1GB is used by the OS, you need to run ((16 - 1) / 2) = 7.5 instances - ie 7 at 2GB and one at 1GB, but play with that last one so you don't swap.

I have no idea why you decided to build your own machine. The number of threads I've seen about this is just unreal. You should've bought a premade machine. Your fan setup is huge overkill. Everything is overkill really. But you didn't pay attention to what anyone said before, and it's too late now, so good luck to you.

Thanks for the info on testing everything.

As for over kill on the fans, probably a good thing ref the running relatively hot.

By adding another £60 in fans, would have worked beautifully if this processor wasnt overvoltaged. ( for whatever reason ). Because in Standard fan mode the machine is virtually silent running all fans at circa 300 rpm. You have to have your ear virtually on the case to hear it. Yet still have the potential to deliver a hurricane of cooling if needed say in the future rendering 4K video.

Sorry about all the posts - isnt that what forums are for. One has to learn somehow.? No?

Bluntly I now have a few problems that I feel even the most experienced would have had. Either Supermicro need to deliver a Bios that works or Intel need to replace a faulty cpu.

The question is who is to blame - intel ark says " the processor tells the nimum and maximum voltage values at which the processor is designed to operate. The processor communicates VID to the VRM (Voltage Regulator Module), which in turn delivers that correct voltage to the processor."
So it looks like Intel are at fault. Or SM's bios implements the communication incorrectly.
I Assume the voltage regulator module is on the mother board or coded into the bios?

It looks like the bios is saying the CPU needs 1.47V to meet the min. speed spec. Of 3.5 Ghz. My bet is its a faulty processor. Because despite the warm 44C running and pushing 1.47V my cooling system is still keeping it way under its max temp of 66.67C.
Sadly the X10SRA doesnt have the facilities like my friends MSI gaming board to tweak the voltage, as it would be interesting to drop that to say 1.2V and see what the speed was.

I would definitely investigate the voltage issue further before running any stress tests like Prime 95. Overvolting a CPU will shorten it's life. If you have Windows installed you can use a program like CPU-Z to check CPU speed and voltage. I would contact Supermicro and have them try and diagnose the problem. It could just be a wrong reading from the voltage sensor.

Here's a link for it http://www.cpuid.com …s/cpu-z/cpu-z_1.75-en.exe

Have you thought of switching to a liquid cooled CPU fan, may help keep temp down and protect CPU until you figure out voltage problem?

From: http://ark.intel.com …650-v3-15M-Cache-3_50-GHz

VID Voltage Range 0.65–1.30V

Your mobo is frying that cpu. It's possible that your mobo doesn't support that cpu...
There should be specs listing what it is certified to use.

When I first hooked up my i7 3820 the bios (Asus Sabertooth z79 board) clocked the voltage to 1.41v, don't remember the temp because I knew right off the bat that was too high so I just manually clocked it down to 1.28 and put an OC on it to 4.4ghz. My idle temp is usually around 38C.

I would just set your voltage manually and don't worry about. You're not hurting anything until you get closer to your max temperatures anyway, which is much higher than what you're seeing right now.