I can't get Boinc to a reliable state with ANY kind of OC going on, even the slightest bit. Is there something specific I need to tweak for Boinc or is it just a bad idea to OC for it? I have an otherwise fairly stable overclock, passed multiple stress tests for 6-8 hours each and the temps are also pretty comfortable. Running Boinc though, will result to random BSODs, from 5 minutes in to 3 hours in, I can't really replicate it. Anyone has had any better luck with overclocking and crunching?

PS: It's a 5820k, pushed to 4.5ghz. Boinc wasn't stable at even 3.9ghz, which is just 0.1 more than the default turbo speeds of this cpu.

How well a particular CPU will stand overclocking is impossible to predict. Even within the same batch of processors some will stand far higher clock speeds than others. If one is unfortunate to have a processor that is sitting at the bottom end of the stability curve then it will fail at far lower clock speeds than one sitting at the top end.
SETI@Home is "famous" for stressing processors, and overclocking will only add to that stressing....

---

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

How well a particular CPU will stand overclocking is impossible to predict. Even within the same batch of processors some will stand far higher clock speeds than others. If one is unfortunate to have a processor that is sitting at the bottom end of the stability curve then it will fail at far lower clock speeds than one sitting at the top end.
SETI@Home is "famous" for stressing processors, and overclocking will only add to that stressing....

Hey there Bob, thanks for taking the time to reply. As I have been into overclocking for a good while, I understand your point completely. I'm actually pretty lucky with this chip, it overclocks fairly well and stable compared to the average. Like I mentioned in the OP, it is only Boinc that doesn't like it. Anything else I've tried is happy up to 4.5ghz, while Boinc won't even play nice with the absolute minimal. At first I thought I'd just lower my overclock to whatever is stable with Boinc, but it seems that's not an option.

About stressing, I think the software I use for testing actually is heavier on the CPU than Boinc. I've used Prime95, AIDA64, RealBench and Handbrake. All of which ran for a minimum of 6 hours to 8-9 hours maximum with zero errors.

So that gets me thinking, is there anything specific to how Boinc works that I'm missing? A +36% boost is a lot when crunching numbers and I really want to get this machine going and get my numbers up :P

So is it specifically the boinc client, manager, or applications that are choking first? which one would point to different possibilities.

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

So is it specifically the boinc client, manager, or applications that are choking first? which one would point to different possibilities.

I'm not sure what you mean by manager or applications. The Boinc client itself runs fine, but when it starts running the tasks the computer will randomly BSOD. It can happen early in, but also after few hours of crunching.

So is it specifically the boinc client, manager, or applications that are choking first? which one would point to different possibilities.

I'm not sure what you mean by manager or applications. The Boinc client itself runs fine, but when it starts running the tasks the computer will randomly BSOD. It can happen early in, but also after few hours of crunching.

The Tasks run as separate applications, launched by BOINC. You can see the individual executable names spawned as child processes from boinc.exe, by using a tool like Process Explorer

So is it specifically the boinc client, manager, or applications that are choking first? which one would point to different possibilities.

I'm not sure what you mean by manager or applications. The Boinc client itself runs fine, but when it starts running the tasks the computer will randomly BSOD. It can happen early in, but also after few hours of crunching.

The Tasks run as separate applications, launched by BOINC. You can see the individual executable names spawned as child processes from boinc.exe, by using a tool like Process Explorer

Oh I see. Should I check the console log to specify the errors? Or is there a Boinc specific log I could look for?

So is it specifically the boinc client, manager, or applications that are choking first? which one would point to different possibilities.

I'm not sure what you mean by manager or applications. The Boinc client itself runs fine, but when it starts running the tasks the computer will randomly BSOD. It can happen early in, but also after few hours of crunching.

The Tasks run as separate applications, launched by BOINC. You can see the individual executable names spawned as child processes from boinc.exe, by using a tool like Process Explorer

Oh I see. Should I check the console log to specify the errors? Or is there a Boinc specific log I could look for?

There are various logs available, but if your main problems are BSODs, I don't think any app writes fast enough to leave any helpful traces behind. I use blue_screen_view if needed to see the dump files that Windows creates - you might be able to find which driver or application triggered the crash from those.

I can't get Boinc to a reliable state with ANY kind of OC going on, even the slightest bit. Is there something specific I need to tweak for Boinc or is it just a bad idea to OC for it? I have an otherwise fairly stable overclock, passed multiple stress tests for 6-8 hours each and the temps are also pretty comfortable. Running Boinc though, will result to random BSODs, from 5 minutes in to 3 hours in, I can't really replicate it. Anyone has had any better luck with overclocking and crunching?

PS: It's a 5820k, pushed to 4.5ghz. Boinc wasn't stable at even 3.9ghz, which is just 0.1 more than the default turbo speeds of this cpu.

Hi,

the Seti CPU applications, both Multibeam and Astropulse, are higly optimized. They utilize the CPU to the maximum and there is a lot of heat generated. Most applications sit idle and have time for the heat to dissipate. Thats why your OC is unstable with Seti.

---

To overcome Heisenbergs:
"You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones

Is it crashing with any kind of WUs (CPU/GPU/MB/AP)?
You could try and isolate what exactly is triggering it.

Anything else I've tried is happy up to 4.5ghz
About stressing, I think the software I use for testing actually is heavier on the CPU than Boinc.
I've used Prime95, AIDA64, RealBench and Handbrake. All of which ran for a minimum of 6 hours to 8-9 hours maximum with zero errors.

Try LinX - The simpliest Linpack interface
http://www.xtremesystems.org/forums/showthread.php?201670-LinX-A-simple-Linpack-interface
http://www.xtremesystems.org/forums/showthread.php?201670-LinX-A-simple-Linpack-interface&p=5265588&viewfull=1#post5265588

Be careful as it heats the CPU fast! (i.e. try first on stock GHz, no Overclock)
Free as much memory as possible (exit all programs) before the test and set it to use more RAM

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

One thing to check is that the CPU cooler is clamped to the CPU correctly - I had issues with a distorted CPU top, on my only Intel system, that meant the cooler wasn't clamped correctly.
Remove the overclocking, and see what happens.
As others have said SETI really pushes CPUs to their limits, far beyond the normal test programs, on many occasions over the years I've had systems that have "passed test" only to fail in a few hours when running SETI.





Another thing to look at is using the GTX970, it will far out perform the CPU - to see by how much take a look at the figures for this PC of mine: http://setiathome.berkeley.edu/show_host_detail.php?hostid=7619589
(I even turned off hyper threading as the CPU contribution is so low compared to the GPU)

---

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

I can't get Boinc to a reliable state with ANY kind of OC going on, even the slightest bit. Is there something specific I need to tweak for Boinc or is it just a bad idea to OC for it? I have an otherwise fairly stable overclock, passed multiple stress tests for 6-8 hours each and the temps are also pretty comfortable. Running Boinc though, will result to random BSODs, from 5 minutes in to 3 hours in, I can't really replicate it. Anyone has had any better luck with overclocking and crunching?

PS: It's a 5820k, pushed to 4.5ghz. Boinc wasn't stable at even 3.9ghz, which is just 0.1 more than the default turbo speeds of this cpu.

Hi,

the Seti CPU applications, both Multibeam and Astropulse, are higly optimized. They utilize the CPU to the maximum and there is a lot of heat generated. Most applications sit idle and have time for the heat to dissipate. Thats why your OC is unstable with Seti.

You sure? I always thought the stock apps are not very optimized, so they can run on almost any CPU.
Anyway, I don't see any reasons, why the CPU would pass a stress test for hours but immediatly chokes on Seti. That's really strange. There are/have been ppl running Seti on oc'd CPUs. Mark and his famous machine for example. Salad Fingers, I would try Lunatic apps instead of stock.

---

A blue screen (of death) is symptomatic of a significant problem, in this case it is almost certainly due to the CPU overheating, "Lunatics" will almost certainly make things worse, not better.

Looking back through the thread I see that even at the lowest level of applied overclocking there are blue screens, and this is the same as the "boost". The boost is only really available for short periods of time, and then when the temperature is low enough, not for sustained operation. This raises a question - what cooler is being used? (and what is the room ambient temperature?)




As for SETI not hitting things as hard as the stress test programs all I can say is the early days of my Intel based system the stress testing software did not show the problem as it passed all the tests, but just a few minutes running SETI and the CPU was getting very hot, very hot and the system either stalling or blue screening. The stress caused by SETI must be something to do with the mix of instructions in use being "just right" to get the temperature rising.

---

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

Just to chime in here.

I have overclocked my 1100T and stressed it with prime95 and it remained stable for days. As soon as I started Boinc running Seti it would BSOD
Boinc is now the way I test my Overclock. It stresses the system far more than any stress program out there.
Getting a CPU to remain stable is not just the CPU doing it as much as the staging of your motherboard. 4+1 is not suitable for OC. 8+1 or 8+2 is far better.
Ram is always the main consideration. Taking them off auto is a must. Getting your timing correct is something of a trial and error deal that each and every board, Ram and CPU combo will face. Whether or not you run x86 or x64 is another factor. For some reasons x86 runs with higher stable OC. Whether or not your ram slots are totally populated is another one. Two works better than Four
Weather or not you can Overclock by the CPU multiplier instead of pushing the FSB up is another big Factor, Always obtain unlocked CPU for the best OC.
Finally take the stock block and fan off and throw it in the garbage because it is useless. Get a good cooler. Liquid based is the best but some air coolers are just fine. Coolmaster makes great coolers for air.
I always take off my side plate to my box and place a small desk fan to cool it
Watch out for wire biting kitties though.
Speaking of Kitties you could do what Mark Sattler does. Use Freon based cooler and freeze your ass off

Sorry mark. Your great in my book.

Michael Miles

Thanks everyone for your replies. I will have a better look at it when I'm back home. To address some of your comments, though: Many of you are mentioning that it's a problem with temps or the overclock, while that is not the case. I have run tests for days, with different software and case scenarios. The CPU never exceeded 67C which was a spike, monitoring software gave it a 63C average. That temp shouldn't be a problem. Also about Seti being hard on the CPU, I understand that, but as I stated in the OP I have run tests that stress my CPU more than Seti, at least thermally, without problems. Seti only reaches 64C and mostly averages at 58C.

I didn't know that Boinc can make better use of the GPU card than the CPU, but the card has a flashed BIOS with an aggressive overclock which also freaks out when running Seti and I had kept it disabled. May fire it up with the stock BIOS and make some use of it, too. But I'm still pretty baffled with the CPU problem. WHen I'm back home I'll check the logs and test that program mentioned.

Thanks for all the help! I will eventually figure it out.

<snip>Getting a CPU to remain stable is not just the CPU doing it as much as the staging of your motherboard. 4+1 is not suitable for OC. 8+1 or 8+2 is far better.<snip>

Hey there, could you elaborate on this please?

Edit: Also my memory is set to use an XMP profile to reach it's factory speeds, should I try and get it there manually?

Edit2: @rob smith The cooler I am using is the H110 from Corsair, fitted with 2 Noctua's 3k rpm fans, on a somewhat aggressive curve, but not too much because it gets too loud. Ambient temp is around 20C this period of the year.

Sure bro

By staging I mean power phases

The capacitors beside you cpu socket determine how many phases the power goes through. If there are 5 then you have a 4+1 stage power phase. If there are 10 then you have a 8+2 stage power phase. The more stages the power goes through the more stability your CPU will have.

More phase means greater and cleaner electricity to be put into your CPU, therefor increases OC-ing ability and stability of your CPU.
XMP profile means it locked to that profile. Just fine as long as it is on the profile your good. Placing off profile and on auto is a problem
One other thing that was stressed to me. Do not be scared to use some voltage, just not too much

Michael

Well cooling should be OK with that cooler, but see my comment about an Intel processor that was misaligned, and so preventing the cooler sitting correctly. This is highly unlikely, but is worth bearing in mind if the system is unstable even at stock speeds.
My Intel system barely reaches 50C when running the same test programs as you have described, but hits 75C when running SETI - that's how much more SETI loads the CPU,as Michael Miles says, SETI is a monumental stress test program. (I had one system that I had built for me by a local shop passed all its pre-delivery test, but failed as soon as I started SETI, I went back to them and explained the situation, they then loaded SETI onto another system and THEY were amazed at the thermal load to such an extent that they now use SETI as a final burn in test for all their new mid/high performance systems.)

So back to your problem.
Return everything to its stock speed & voltage settings, make sure SETI runs, and is stable for at least a couple of days.
As you have loaded a non-standard BIOS onto the GPU revert it to stock - many of these non-standard BIOS are OK for gaming, but fail miserably when using the GPU for heavy duty calculations - a single pixel error in the picture in a game will be un-noticable, however a single bit error in a calculation can make a huge difference.

Once you are certain that the system is stable (for a few days) then start to think about overclocking. Do not rush in, one very small step at a time and let it settle for a day or so between steps. However, if you get the GTX970 working properly you will be amazed at its performance, and may well decide that overclocking is not required.

---

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

Just my 2 cents worth.

Have You taken a look at the photos on Tom's Hardware, those that shows the innards of Your CPU?

Then You will notice that different parts of Your CPU is used for different tasks.

Running Prime95 will utilize the FPU's at their maximum, while other parts of the CPU are less stressed.

CPU utilization at 100% does not necessarily mean that the CPU LOAD is 100%, so
running Prime95 without problems does not guarantee You that no other part of Your CPU die containining other registers will not fail.

If You want to see how much SETI work Your CPU is capable of, the only way to find out is to let it run SETI.

REM: Just curious. :) Why are You overclocking Your CPU that hard? This is definately not the way to increase You RAC.

ChrisD