Claymore and GpuMiner cu_k02 failed xx - G3930 not enough

[Iomgui]

Configuration
ASUS Z270 P + G3930 + 4G RAM (64G virtual mem) + 8 GTX 1060 (Samsung) + Win 10 Pro 1709

Ok its been some time now i was bothered by having errors leading to either a restart of Claymore (more than 95% of the time), the loss of one graphic card in Claymore (lets say 4%), or the crash of computer (1%). The error i was getting in Claymore log was "GPU miner cu_K02 failed error 30" or cu_k01, and error 73 too. Internet research leaded to : too much OC, risers faulty, psu not enough, card faulty and silicon bad luck. And i can tell if i lost a graphic card or crashed because each time claymore get this error, it starts a new log so you just need to look at the number of logs you get and check the new log to know if you lost a card. Of course crash you are aware without log.

At start i was thinking that my OC settings were too much, or that i didnt setup properly the motherboard (i started to FIX GEN1 in bios for some PCIE Values).I was doubtfull of my OC being bad because i was still getting the error at a ridiculous + 400MHZ (ridiculous for Samsung). I even tested the cards one by one and got complete random results to reproduce the errors. I finnaly eliminated the risers, the PSU and a faulty card. What i wasnt able to eliminate was the M2 to PCI converter as i dont have different brands. Though, i was certain there was no error from the graphic card as HW64 was saying so (i wasnt OC too much).

But i fanlly and by luck found how to produce this error. It seems that if my CPU is for any reason at full load (Windows service, other things) during more than a few seconds, i had high probabilty to get one of these errors.


And i managed to provoke the error multiple time by saturating the CPU load. What i cant tell is if it is a Claymore only error.

Do some of you had this error ?
Should i consider replacing the G3930 by another one ? and wich one ?
How could i lower the load on the G3930  ? (most of the time Claymore takes 2 to 15%, afterburner between 15 and 25%, and some random Windows service can come and frak my cpu load)

Any advice very welcome
(and yes i want to stay Under Win10 for the ease of use, my linux is too old)

[halker2010]

Configuration
ASUS Z270 P + G3930 + 4G RAM (64G virtual mem) + 8 GTX 1060 (Samsung) + Win 10 Pro 1709

Ok its been some time now i was bothered by having errors leading to either a restart of Claymore (more than 95% of the time), the loss of one graphic card in Claymore (lets say 4%), or the crash of computer (1%). The error i was getting in Claymore log was "GPU miner cu_K02 failed error 30" or cu_k01, and error 73 too. Internet research leaded to : too much OC, risers faulty, psu not enough, card faulty and silicon bad luck. And i can tell if i lost a graphic card or crashed because each time claymore get this error, it starts a new log so you just need to look at the number of logs you get and check the new log to know if you lost a card. Of course crash you are aware without log.

At start i was thinking that my OC settings were too much, or that i didnt setup properly the motherboard (i started to FIX GEN1 in bios for some PCIE Values).I was doubtfull of my OC being bad because i was still getting the error at a ridiculous + 400MHZ (ridiculous for Samsung). I even tested the cards one by one and got complete random results to reproduce the errors. I finnaly eliminated the risers, the PSU and a faulty card. What i wasnt able to eliminate was the M2 to PCI converter as i dont have different brands. Though, i was certain there was no error from the graphic card as HW64 was saying so (i wasnt OC too much).

But i fanlly and by luck found how to produce this error. It seems that if my CPU is for any reason at full load (Windows service, other things) during more than a few seconds, i had high probabilty to get one of these errors.


And i managed to provoke the error multiple time by saturating the CPU load. What i cant tell is if it is a Claymore only error.

Do some of you had this error ?
Should i consider replacing the G3930 by another one ? and wich one ?
How could i lower the load on the G3930  ? (most of the time Claymore takes 2 to 15%, afterburner between 15 and 25%, and some random Windows service can come and frak my cpu load)

Any advice very welcome
(and yes i want to stay Under Win10 for the ease of use, my linux is too old)

if updating drivers doesn't fix the cpu load , you need to bump up some more ram and get a better cpu and see that what is using the cpu in task manager if claymore you need to get a better cpu.

[CryptoWatcher420]

ROFL 64gb of virtual memory, dude that's beyond stupid. GET MORE RAM. clearly you don't know that setting that much virtual memory really affects the PC performance, but hey stupid is as stupid does

[gt_addict]

4gb of ram?!! Minimum i read to use was 8gb but now ive switched to 16gb and no major problems. I was suffering from a similar issue to you but not anymore.

[Iomgui]

ROFL 64gb of virtual memory, dude that's beyond stupid. GET MORE RAM. clearly you don't know that setting that much virtual memory really affects the PC performance, but hey stupid is as stupid does

Thanks for your answer. Yea i know i put 64 only because i do all rigs the same way and 64 is only required for 8 1070. Though with 1070 i have zero problem with the same config though i do zcash and not claymore but it points out i need to test the 1070 with Claymore.
Get more RAM ? and I am beyond stupid ? Let me ROFL too
But let me explain also, what you didnt take time to, the main use of virtual ram in a mining RIG Under win10 is for graphic card RAM dump, so basically you size Virt RAM by the sum of your graphic card RAM and its enough. Also you'll learn that its the MAX size of virtual RAM that you setup if you know wath you do but Windows dynamiccaly resize it when required. It was my fault, saying i size virtual memory at 64G Max is absolutely irrelevant to the problem i m facing. And for your information, the MAX size of Virtual RAM that Windows use is as expected around 22G.
Considering the RAM use : everything else runs at minimal, basicly not more than 2 Go of RAM is used by the process and services. My problem is not RAM related, i have 50% of my ram free. My problem is related to CPU load and only when it reaches 90%+ which is like 1 time per hour. The process that consumes the more RAM is Afterburner, its also the one generally consuming the more CPU load, unless i have CPU load problem and one of the Windows service started to mess whith my rig. I know one of them is related to Windows update and its my fault for not blocking it.


So let me rephrase my questions :
anyone of you had the same type of error and how did he solutioned it ?
Has anyone identified CPU load problems generating errors in Claymore ?
and please remind that i dont say my CPU load is problem all the time, i m saying that for a few seconds every hour my CPU load reach more than 90% and it could result in generating an error. If you experienced same problem feel free to share. If not either remain polite or at least know your subject. I have no problem to be named stupid as long as you know your subject and you explain it.

FYI i tried 8G of RAM just because i have some (i dont have better CPUs) and the result is absolutly the same. It also prooved me that putting 8 RAM for such a configuration was unecessary.

[gt_addict]

Im currently running 4x1070ti on my rig with an i5-7500t, 16gb ram and Win10. On it i run claymore mining eth/pasc and claymores cryptonite cpu miner using 3 out of 4 threads. My cpu reaches 100% load fairly often and no problems.

[Iomgui]

Ok i have new information.

The errors reported by Claymore are related to latency problems with drivers and or CUDA of nvidia cards. In my case and others i v seen on the net its related to high CPU load and its always the G3930 that is guilty. It has been proven that AMD graphic cards doesnt have this problem with this CPU. It seems the problem was even worse for ethminer and has been fixed for ethminer but not enough feedback on Claymore (i think people dont see the errors as you need to check your log, considering it hardly is the cause of a crash once in a week).
Experts are saying that pentium i5 or i7 at full load wont generate the error where the 3930 will cause the error, they are talking about limitation of the 3930 cpu cache.

It is unclear for now if the motherboard settings is also guilty. I have to try a new bios setting as expert on this subject explained that setting to pcie to GEN1 was an error and could be part of the latency problem. They advise to use GEN2 for both nvidia and amd cards.

I ll let you know if i have more information.