Prime95 Error: ILLEGAL SUMOUT - Bad RAM Voltage/Timing Settings?

JOOK-D · Jan 4, 2014

Hello,

I've been having some issues lately that I assume are down to my RAM, consisting of random freezes and occasionally (once or twice) failing to POST on restart, as well as a few beep codes.

I noticed that upon startup my DRAM_LED flashes red and then goes out, and I cannot find what is the cause of this or whether it is cause for concern. My motherboard is the ASUS M5A97 R2.0, with an older BIOS but it does support my CPU (FX 8320).

So far I have resat the DIMMs to no avail. Prime95 large FFT and Blend tests fail with the error shown in the title - usually 1 or 2 workers fail. I've manually set my DRAM voltage to 1.65V and the speed to 1600 MHz. My CPU is stable for 4 hours in Prime95 on a small FFT test @ 1.34v, core temps don't exceed 62c and socket temps go slightly over the max 70c as they reach around 76c.

Here is my exact RAM Model: http://www.kingston.com/datasheets/KHX1866C9D3K2_8GX.pdf
I know nothing of RAM timings and haven't had any problems with it in the past, I only noticed this recently (following my CPU overclock / setting my BIOS to optimised defaults).

Does anyone know whether this could be an issue with my DRAM voltage or timing settings? I appreciate any response, thanks in advance.

Tradesman1 · Jan 4, 2014

Try manually setting to 1866, 9-11-9-28, 2N, 1.65 for DRAM voltage and raise CPU/NB voltage to 1.2

JOOK-D · Jan 4, 2014

How do I manually set the timings? Total noob question.

I did this just before you replied, I assume this isn't what I should do? If it is, I tested in p95 and it crashed again.

EDIT: I assume it's in DRAM Timing settings? And it's the first 4. I'm not sure what the 2N refers to though - the Command Rate? I only have 1T and 2T available.

JOOK-D · Jan 4, 2014

I manually set to 1866MHz and 9-11-9-28. DRAM Voltage is 1.65V.

After about 10 seconds on a Blend test worker #5 fails, followed swiftly by #6, #7 and #8. The same occurs in large FFT tests.

Tradesman1 · Jan 4, 2014

Get the latest BIOS first....CR is the Command Rate or Command Timing, should already be at 2N or 2T but some mobo will set it to 1T or N, and may need to raise the CPU/NB voltage as mentioned might have a weaker 8320

JOOK-D · Jan 4, 2014

Alright, I set it to 2T and it lasts for a few seconds longer before I get the error. I'll slowly bump up the CPU/NB and see if it makes a difference - do you know the limit for it?

JOOK-D · Jan 4, 2014

Tradesman1 :

I'll be updating the BIOS soon. Bumping the CPU/NB to 1.22 seems to have done the trick

It's all stable.

Thanks a lot!

Tradesman1 · Jan 4, 2014

CPU/NB can be tricky but is normally safe up through about 1.35 on most mobos some go higher, others not quite that much (that I've found anyway) it handles the MC (memory controller)

JOOK-D · Jan 5, 2014

The tests are stable up to around 10 minutes in when 1 worker fails @ 1.25v, I'll bump it a little further and see how it goes. Thanks for the info.

Tradesman1 · Jan 5, 2014

No worries, keep us updated, sounds like on the right track

4Ryan6 · Jan 5, 2014

JOOK-D :

As an additional option drop your memory speed to 1333mhz and use timings of 9,9,9,24, 2T @ 1.5v manual setting, at that point the CPU/NB can stay on auto, your CPU temperature should drop noticeably and it should be able to stabilize at the 4.4ghz overclock you were running.

You may actually be able to bump your multiplier overclock higher and still be stable only requiring additional CPU voltage bump up to stabilize a higher multiplier.

JOOK-D · Jan 5, 2014

So I'm now getting an error of "FATAL Error: Rounding was 0.5, expected less than 0.4."

CPU/NB: 1.3v
DRAM: 1.65, 9-11-9-28-2T, 1866MHz.
Temps completely in check.

I guess I'll continue to bump the CPU/NB a little. My CPU is completely stable in smallFFT.

JOOK-D · Jan 5, 2014

4Ryan6 :

I'll try that if my CPU/NB gets too high. It definitely seems as if the RAM is the issue since smallFFT is completely stable. Also, any idea on what I should set the CPU/NB LLC to? It's currently on Auto.

4Ryan6 · Jan 5, 2014

JOOK-D :

Just leave it on auto and test those 1333mhz timings, the Kingston information you linked says those modules will run at those timings and at 1.5v DDR3 voltage.

That will drop your CPU temperature.

Those timings and speed should give you a rock solid memory foundation to build on, so test them and see what happens, you can always tweak tighter timings later once you reach a rock solid overclock memory foundation.

If your memory is off, it throws everything off along with it.

But lets get you rock solid first and go from there.

JOOK-D · Jan 5, 2014

Yeah I've been testing it for the past 20 minutes and 1333MHz just failed. It's when the Self-test 1024K starts and I think my PSU starts making a very high-pitched squealing sound. Worker #5 failed due to Rounding being 0.5 when it was expected to be 0.4, and Worker #7 failed because Rounding was 0.484375 when it was expected to be 0.4. It seems the ILLEGAL SUMOUT error is gone though. Where do I go from here?

JOOK-D · Jan 5, 2014

As an update, I tried testing again and the ILLEGAL SUMOUT errors are back, I start the test and it fails.

Would it be worth resetting everything to optimised defaults and just starting my overclock over again?

JOOK-D · Jan 5, 2014

It seems that it keeps failing at the same point, following the high-pitched (PSU?) squeals and after passing the Self-Test 1024K. Worker #5 gets the ERROR: ILLEGAL SUMOUT.

This is with:
1600MHz
1.3v CPU/NB
9-9-9-28-2T
1.65V DRAM

Any ideas on what to do? Thanks.

This is from the readme.txt: "Windows NT/2000/XP/Vista protects prime95 from bad device drivers. In that case ILLEGAL SUMOUT is very likely a true hardware error. The good news is that prime95 recovers very well from ILLEGAL SUMOUT errors. Try seeing if the problem only occurs when a specific device is active or a specific program is running." I'll try closing any startup programs that may be interfering.

4Ryan6 · Jan 5, 2014

What are your power supply voltage readings regarding 12v, 5v, and the 3v readings, you should find those readings under system health or something similar in your BIOS, your power supply squealing is not good that could be the cooling fan failing or much worse, it could be intermittently failing and causing the problems you're experiencing since at one time you were stable with higher memory settings?

4Ryan6 · Jan 5, 2014

http://forum.corsair.com/v3/showthread.php?t=85166

JOOK-D · Jan 5, 2014

I ran @ 4.4 for about 2 weeks and completely forgot about testing the memory. Then I started getting the random lockups etc. and found the memory to be the potential culprit and remembered the blend test etc.

12v, 5v and 3v have always been within the acceptable ranges.
3.3v: 3.264
5v: 4.992
12v: 12.046

As I said I have the Corsair TX650 and after doing some research, many people seem to have had this same issue or it has just been DOA.

JOOK-D · Jan 5, 2014

Thanks for the link. Unfortunately I don't have another system to test in but the sound is definitely the PSU. All power saving features have been disabled so as to not interfere with my overclock too.

Rather than RMAing I'd rather just get a new fully modular PSU if it is the problem, since I'd have to pay shipping fees etc.

4Ryan6 · Jan 5, 2014

You definitely have some kind of hardware problem to suddenly arise like this, if it was memory related you should have easily been able to run it at 1333mhz at cas 9, that would have relieved any OC load on the CPUs memory controller, the squealing coming from the power supply does indicate somethings wrong there, and if is not the P/S cooling fan making the noise, it could be the capacitors?

JOOK-D · Jan 5, 2014

Yeah, I'm not sure what the hardware problem is though. All I know is that it's RAM/PSU related.

My RAM is finally somewhat stable:
1600MHz
9-9-9-27-2T
1.65v
1.3v CPU/NB

Finally managed to get over the hurdle that I kept getting stuck at before and now I have this potential PSU issue. It's always had some kind of whine but it's never been this loud. SR-71 Blackbird mentioned on my other thread that he still thinks it's coil whine but it just ramps up due to the prime95 stress.

Considering I couldn't get the RAM stable at 1333MHz could it be a RAM problem? Or just the PSU delivering power incorrectly?

As it's sitting idle now I still hear some coil-whine and I'm not sure whether I should continue testing in prime95 :lol: It would cost me £73 for a new PSU, and it isn't even fully modular. So I'm kind of stuck deciding whether to risk it and play it off as coil whine, or just get a new PSU (was planning to get an SSD instead).

4Ryan6 · Jan 5, 2014

JOOK-D :

Possibly or (and I know you don't want to hear it), it could even be the motherboard, in most situations you're at the point of swapping out known good parts to eliminate the hardware failure possibilities, do you have any local friends that are running DDR3 memory you could borrow and swap out to test?

Or a power supply?

4Ryan6 · Jan 5, 2014

I only have a few more minutes online today, I'm sorry I wasn't more help to you!

I'll check back at another time, Take Care and best to you! Ryan

Prime95 Error: ILLEGAL SUMOUT - Bad RAM Voltage/Timing Settings?

Dignified

Legenda in Aeternum

Dignified

Dignified

Legenda in Aeternum

Dignified

Dignified

Legenda in Aeternum

Dignified

Legenda in Aeternum

Champion

Dignified

Dignified

Champion

Dignified

Dignified

Dignified

Champion

Champion

Dignified

Dignified

Champion

Dignified

Champion

Champion

Share this page