Hey All.

So I've been running Seti@home as well as a couple other projects at home. These are on an MSI GP72VR Leopard Pro laptop with an i7-7700HQ and Nvidia GeForce 1060. I keep it on it's comfort tune when crunching. In addition, I have 2 exhaust port fans and a laptop cooler pad for it.

The other machine is one I put together consisting of an i9-9900k on an MSI Gaming Carbon Pro AC mother board, MSI GeForce 2080 gaming x trio , Corsair H150i Pro, and 15 case fans with a positive pressure case. I am also only using 75% of the processor.

Typically, I can run any of the other projects bouncing between 55-65 C (highest spike being 65C). The other projects I run are Milkyway@Home, Enigma@home, and MindModeling@beta.

The problem I run into is when SETI@home kicks in, it will run me up into the mid 70's on both systems, and then crash my i9 system. That doesn't even happen when I put it on a heavy system stability test. Is there a particular reason that SETI runs the temps so much higher? Does anyone else have similar problems running just SETI@home?

Thanks!

Bill

Positive pressure is not really a good thing in a case. What you want is airflow or air exchange. Amount of air leaving should equal the amount of air entering or just slightly below. Most of the air should be leaving through the radiator.

A good case with proper airflow should be no problem for any computing, even Seti. I don't see any reason or need for 15 case fans. It sounds like you are preventing any air from flowing through the AIO radiator. Have you pulled the AIO cold plate off to check for proper application of heat sink compound and mounting pressure?

Seti does floating point math and that works a cpu the hardest. Try to avoid any AVX application.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Seti does floating point math and that works a cpu the hardest. Try to avoid any AVX application.

The AVX application in the Lunatics installer gives about a 30% improvement over the best stock application.

However Apharot, you have a serious problem with your system cooling if it's hitting 70°c with the stock applications. My previous 2600K used to sit around 70°c with the stock cooler with the ambient temperatures around 30-33°c and the side off of the case. There is no way your system could cope with the AVX application with it's present cooling problems.


Edit- if the motherboard doesn't have a dedicated water pump header, at least make sure the radiator fans are connected to the CPU Fan header, and the water pump to a Sys Fan header.

---

Grant
Darwin NT

Seti does floating point math and that works a cpu the hardest. Try to avoid any AVX application.

The AVX application in the Lunatics installer gives about a 30% improvement over the best stock application.

However Apharot, you have a serious problem with your system cooling if it's hitting 70°c with the stock applications. My previous 2600K used to sit around 70°c with the stock cooler with the ambient temperatures around 30-33°c and the side off of the case. There is no way your system could cope with the AVX application with it's present cooling problems.


Edit- if the motherboard doesn't have a dedicated water pump header, at least make sure the radiator fans are connected to the CPU Fan header, and the water pump to a Sys Fan header.

If there were a problem with my cooling system, I would see those temperatures on everything else. SETI is literally the only thing I see this on. I can run Milkyway on 100% and I won't see those temperatures. In addition, it handles other AVX applications with no problems and there is no thermal throttling. It's a brand new system, with brand new drivers, and the blue screens occur well before the temperature would be an issue (throttling doesn't kick in until 100 C, and if it were anywhere near that, I'd agree there's a cooling issue). This isn't a cooling issue.

The MB does have a dedicated pump header, and it's fans are connected to the supplied connections that come on the pump itself. All other fans run through Corsair Commander Pros.

Positive pressure is not really a good thing in a case. What you want is airflow or air exchange. Amount of air leaving should equal the amount of air entering or just slightly below. Most of the air should be leaving through the radiator.

A good case with proper airflow should be no problem for any computing, even Seti. I don't see any reason or need for 15 case fans. It sounds like you are preventing any air from flowing through the AIO radiator. Have you pulled the AIO cold plate off to check for proper application of heat sink compound and mounting pressure?

Seti does floating point math and that works a cpu the hardest. Try to avoid any AVX application.

Positive pressure does exchange air. It's constantly pulling in cold air, forcing warm air out, and because it's running positive, keeping dust out as well (which can be a major cause of heat build up). The problem isn't really the heat. This chip should be able to operate at 70 C with no problem. In addition, SETI is the only thing it reaches that temperature on, even when stress testing on other programs. But again, I'm not horribly worried about the heat. 70 is hot, but not out of operating range according to Intel. The problem is the blue screens on a brand new system running the latest drivers for everything. I've tested and retested this setup multiple times and SETI is the only thing this happens on.

I'm not blocking flow through the radiator, I'm actually using a push pull through the radiator, and I thought of the thermal compound as well. I pulled it off this past weekend and re-applied. It did not seem to have any bare spots on it when I looked, but in pulling it off, the nature of it pulls thermal paste back a bit like a suction, so there could have been small pockets of air. If there were, however, I haven't seen any temperature drops.

Maybe I'll do some tuning work on it this weekend and see if I can find the issue.

(throttling doesn't kick in until 100 C, and if it were anywhere near that, I'd agree there's a cooling issue). This isn't a cooling issue.

The whole point of a cooling system is to remove heat, and that isn't happening.- that makes it a cooling issue.
A water cooled system shouldn't get over 70°c. With the fans and pumps all running at max, even with the ambient temperatures in the low 40°c it should still remain below 70°c

Different applications work different parts of the CPU. Applications (such as Seti) work very specific parts of the CPU, and they work it very hard. And the CPU cooling system shouldn't have any difficulty removing that very localised heat.

My i7 8700k systems run the AVX application, all cores with full boost with a mid range AIO liquid cooler. Even when the ambient temperatures reach almost 40°c, the CPU temperature remains around 70°c.

I'd check you Commander Pro settings- try running Seti with the pump & radiators fans set to run flat out. If the temperature is OK then, then it's just a case of having the appropriate controler settings. If not- there is something seriously wrong with the system that it's not able to remove the heat.


Edit- having had a quick look at the Commander Pro (CP) system, it would be worth checking how the temperature sensor inputs on the CP relate to the temperature sensor levels from the CPU itself.
If the CPU is at 70°c, but the input that controls the pump & radiator fans is only at 55°c, then they won't be working hard enough to get rid of the heat the CPU is actually producing.

Edit- Ignore the edit above.
I re-read what you posted, and you need to check the BIOS settings as they are what are controlling the pump & radiator fans.

---

Grant
Darwin NT

The problem is the blue screens on a brand new system

And what do the blue screens say?
If you don't know or don't remember, use BlueScreenView to recollect.

---

(throttling doesn't kick in until 100 C, and if it were anywhere near that, I'd agree there's a cooling issue). This isn't a cooling issue.

The whole point of a cooling system is to remove heat, and that isn't happening.- that makes it a cooling issue.
A water cooled system shouldn't get over 70°c. With the fans and pumps all running at max, even with the ambient temperatures in the low 40°c it should still remain below 70°c

Different applications work different parts of the CPU. Applications (such as Seti) work very specific parts of the CPU, and they work it very hard. And the CPU cooling system shouldn't have any difficulty removing that very localised heat.

My i7 8700k systems run the AVX application, all cores with full boost with a mid range AIO liquid cooler. Even when the ambient temperatures reach almost 40°c, the CPU temperature remains around 70°c.

I'd check you Commander Pro settings- try running Seti with the pump & radiators fans set to run flat out. If the temperature is OK then, then it's just a case of having the appropriate controler settings. If not- there is something seriously wrong with the system that it's not able to remove the heat.


Edit- having had a quick look at the Commander Pro (CP) system, it would be worth checking how the temperature sensor inputs on the CP relate to the temperature sensor levels from the CPU itself.
If the CPU is at 70°c, but the input that controls the pump & radiator fans is only at 55°c, then they won't be working hard enough to get rid of the heat the CPU is actually producing.

Edit- Ignore the edit above.
I re-read what you posted, and you need to check the BIOS settings as they are what are controlling the pump & radiator fans.

Fans run max, but the pump doesn't. The pump is plugged into the MB, but I can partially control it through the Corsair commander. It doesn't run at max unless I tell it to. I'll check the BIOS to see if there's a way to have it flip over to max at a certain temperature.

If not, I'll plug it directly into the commander pro's USB input and see if there's a way to run it that way. I think I tried that before, but switched it for some reason.

I'll give it this, SETI@home is, apparently, a better stress testers than the actual stress testers!

Thanks!

The problem is the blue screens on a brand new system

And what do the blue screens say?
If you don't know or don't remember, use BlueScreenView to recollect.

I'll get that info when I get home. I'm currently at work. Thanks for that link!

Also, made a big mistake and left something out of my build that does matter. It is overclocked to 5.1 GHz at 1.33 volts, which at least partially accounts for the heat generation (even going to solder didn't seem to help Intel with the heat issues).

I was talking to the guys in the local computer repair shop a few months ago. After having seen how SETI triggered a strange fault on one of my computers they decided to investigate further. Sufficient to say they now use SETI to stress-test computers that come in with "random faults" due to the way SETI stresses RAM, CPU & GPU to the limit.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

I was talking to the guys in the local computer repair shop a few months ago. After having seen how SETI triggered a strange fault on one of my computers they decided to investigate further. Sufficient to say they now use SETI to stress-test computers that come in with "random faults" due to the way SETI stresses RAM, CPU & GPU to the limit.

That's what I'm thinking. I'll consider this an issue with my PC, since no one else is saying there's an issue with theirs, and use it to trouble shoot.

The project that I know of that run CPU very hard is PRIMEGRID
A very hard CPU stress-test are PRIME95 https://www.mersenne.org/download/
If you run PRIME95 for at least one hour without errors - then you are god to go on ANY project in BONC - family as far far I know
Seti on mine i9-9900K run about 5-10C lower temps than PrimeGrid

Hey all. Thanks for your help! I did a little messing around and it turns out the voltage seemed to be a little low. I noticed what I programmed it for and what it kept trying to reach were about .005 off. I bumped it up by .1 and it seems to be fine now. Oddly enough, temperatures on both Milkyway and SETI dropped by 10C by raising the voltage, but that might be because the system is having to compensate less.

Also, I was at 1.32 volts, not 1.33. I'm at 1.33 now. Thanks for your help!

Bill

You should test for stability with your actual intended load. If Seti is your intended load, test with that. I can run hours of Prime95 with no errors, but my actual load of Seti will trip up and fault if the voltages, LLC and timings aren't correct for Seti.

And I am not a fan of Corsair AIO's since they have mostly always used Asetek hardware which is the weak link. Asetek hardware fails often and fast in my experience. It can appear to be working correctly in the BIOS and desktop idling but spike the cpu temps instantly under Seti loading if the pump is not working correctly.

The pump should be run at full speed at full +12V voltage. Trying to run it any lesser speed or under PWM control is asking for trouble.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Just a note, it may not neccessarilly bbe your CPU temp, but the board temp.
With that being said, airflow through the case IS important. Try turning a couple of fans around to move air through the case rather than pressurize it.
Even a 92-120mm fan near the MB just to move some air around - you lose that airflow with an AIO cooler.

Also try taping up cracks/holes in the case. It does make a difference.
Like that 140mm fan hole that has a 120 in it, there is an unneeded space.
And PSU inputs are usually much larger than the PSU, tape up a little.

I was surprised how much you can change things with a little time, when your cooling is borderline.

---

It is overclocked to 5.1 GHz at 1.33 volts, which at least partially accounts for the heat generation

Yep.
And most likely the blue screens as well. Back off the clock/voltage & it should then be happy.
Even so, I would have thought a high end AIO such as yours would cope better even with that excessive heat- Annand Tech did a review about a year ago and tested it with a 340W load, which it handled without issue.


Edit-

Oddly enough, temperatures on both Milkyway and SETI dropped by 10C by raising the voltage, but that might be because the system is having to compensate less.

Also, I was at 1.32 volts, not 1.33. I'm at 1.33 now.

I can understand the extra voltage sorting out the bluescreens by helping with the overclock, but the temperatures being lower makes no sense at all...

---

Grant
Darwin NT

It is overclocked to 5.1 GHz at 1.33 volts, which at least partially accounts for the heat generation

Yep.
And most likely the blue screens as well. Back off the clock/voltage & it should then be happy.
Even so, I would have thought a high end AIO such as yours would cope better even with that excessive heat- Annand Tech did a review about a year ago and tested it with a 340W load, which it handled without issue.

The voltage was a little too low and that's where the stumble came from. For the most part, the AIO does well at keeping the system cool unless it's SETI@home, which was kind of odd.

Edit-

Oddly enough, temperatures on both Milkyway and SETI dropped by 10C by raising the voltage, but that might be because the system is having to compensate less.

Also, I was at 1.32 volts, not 1.33. I'm at 1.33 now.

I can understand the extra voltage sorting out the bluescreens by helping with the overclock, but the temperatures being lower makes no sense at all...

Everything I know, and I've been in IT for 23 years including a lot of data center work, says adding voltage raises heat. So when the temps dropped I was a bit puzzled.

There's a couple other things I still want to experiment with, such as dropping the ring ratio and seeing how that works.

Honestly, I wish I had went with MSI's Godlike instead of the gaming carbon pro due to the extra power phases and the heat sinks above them. I have an i3-3570k I'm getting ready to upgrade to an i7 or i9. I may use this MB in that, and get the Godlike for my every day machine.

Lets simplify like this:
- with 10% clock increase, 10% more power is needed
- with 10% more voltage (on same clock), power increase is 20%
- with 10% clock increase and 10% voltage increase (so to get 10% faster CPU), power increase is 30%.

So it doesn't make much sense (except for some benchmarking) to increase a voltage, unless you have a really good cooling system and not care about power consumed.

IMO, reasonable overclock is what you can get on stock voltage while keeping a 100% stability....

---