Prime95: Worker #5 Stopped.

[Synomenon]

I just "rebuilt" my system. Switched motherboard out for a new one. Anyway, my previous board was an ASRock Z77E-ITX. I never could get a stable running system with that board for some reason. Switch over to an ASUS P8Z77-I Deluxe/WD. First thing I did with this new board was run Memtest86+ for 24 hours to make sure the memory slots didn't have any problems. 8 tests after 24 hours and no errors. I then fired up Prime95 and am running a blend test now. It opened up 8 workers. I'm 2 hours into it and worker #5 has stopped running with the message:

FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture test completed 100 tests in 1 hour, 47 minutes - 1 errors, 0 warnings.


I'm not even overclocking. I only have Turbo Boost enabled in the BIOS and have enabled my RAM's XMP Profile.

Should I worry about this? I haven't had any BSODs since switching to this ASUS board.

 

[sangyup81]

Is your LLC on? You could be having a VDrop/VDroop issue

 

[Synomenon]

What's LLC? How do I turn it on if it's off?

Also, worker 7 has now stopped as well:
FATAL ERROR: Rounding was 0.4992675781, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture test completed 127 tests in 2 hours, 15 minutes - 1 errors, 0 warnings.

 

[sangyup81]

What's LLC? How do I turn it on if it's off?

Also, worker 7 has now stopped as well:
FATAL ERROR: Rounding was 0.4992675781, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture test completed 127 tests in 2 hours, 15 minutes - 1 errors, 0 warnings.

It should be a BIOS setting

 

[coffeejunkee]

You don't need to use llc when running stock. In fact, running stock means no llc.

And yes, I would worry, a non-overclocked cpu shouldn't show errors in Prime95.

Considering you tried 2 different mobo's the problem might actually be the cpu itself. Maybe try testing with the memory set to standard 1333Mhz to definitely rule that out.

 

[Kenmitch]

HCI memtest is superior to LinX, Prime (large or small), and memtest86+ for determining memory stability.

• Launch one instance per thread supported by CPU (8 instances of HCI memtest for 2600K)
• Set each HCI instance to use an appropriate fraction of the memory...16GB on a 2600K means each HCI instance (there will be 8 instances) are to use 2048 MB.
• Stability Criterion: Must pass 200% coverage minimum, passing 1000% coverage is preferred (considered gold standard)

Maybe checking the memory some more might help out.

 

[Synomenon]

I hope the CPU isn't bad. It's an i7-3770T. It has to be special ordered. I got mine from Provantage 2 months after I ordered it.

I'll try testing with HCI later.

Another update. Workers 1, 2 and 3 all stopped within minutes of each other, all with similar errors. Those three workers ran for around 6 hours 46 minutes.

 

[Kenmitch]

Are you running the latest bios?

Guessing you are but I see no mention of it unless I missed it. Seems like the last couple state improves system stability.

Have you tried bumping the ram voltage up a couple of tics above stock spec?

 

[Synomenon]

I haven't updated the BIOS yet. First thing I did after putting the system together was to start testing it.

I'll try updating the BIOS and re-running the tests. I haven't tried fiddling with the voltage either. I though running at stock clocks and voltages would equal a stable system.

 

[2is]

I had an issue with a particular version of Prime 95 that would lock up after just over 2 hours of stress testing. It didn't matter what my clock speed was. Default, under-clocked or overclocked, it would always lock up shortly after two hours. I don't remember what version it was, but I re-downloaded what I made sure was the latest version and that one worked normally.

So before messing with your build, try downloading Prime 95 again, make sure it's the latest version/build

 

[fixbsod]

I had issues with P95 blend and had to downclock my RAM from 1866 to 1600, prolly would work with a little more voltage, but as 1600 has solved all my issues I'm not concerned with it.

 

[rgallant]

I had problems with blend also but for me it was running my ram @ 1866 to tweak the cpu ,but when only prime blend failed a worker and not any of the non blend test, I turned on the xmp to 2400 t2 and prime blend passed. lol
so maybe run the non blend prime , if it does not fail check the memory settings again.

 

[philipma1957]

i have a 3770t in an asrock z77e itx it is an amazing machine.


what is your ram?

also run the prime torture with small ftts if it passes i would suspect ram not the cpu or mobo.

 

[Synomenon]

I'm using this Corsair RAM:
CMD16GX3M2A1866C9
Dominator Platinum - 1.5V 16GB (2x 8GB)

It only has one XMP profile in the SPD.


http://www.corsair.com/us/memory-by...hannel-ddr3-memory-kit-cmd16gx3m2a1866c9.html

 

[Ed1]

try leaving the ram at lose timing/SPD slower rates and see if that doesn't error . Then if good slowly increase ram speed and timings .

 

[JAG87]

I haven't updated the BIOS yet. First thing I did after putting the system together was to start testing it.

I'll try updating the BIOS and re-running the tests. I haven't tried fiddling with the voltage either. I though running at stock clocks and voltages would equal a stable system.

You would think.

The simplest thing to do is to increase your CPU voltage starting with an offset of +0.05 and see if the problem goes away. If it does, don't wasting time and exchange the CPU.

 

[2is]

I'm using this Corsair RAM:
CMD16GX3M2A1866C9
Dominator Platinum - 1.5V 16GB (2x 8GB)

It only has one XMP profile in the SPD.


http://www.corsair.com/us/memory-by...hannel-ddr3-memory-kit-cmd16gx3m2a1866c9.html

You can manually adjust your timings. An XMP profile need not exist to do that.

 

[Puppies04]

I hope the CPU isn't bad. It's an i7-3770T. It has to be special ordered. I got mine from Provantage 2 months after I ordered it.

Not the case anymore, my supplier (in UK) is showing stock avaliable for next day delivery. Did you order yours when it was released by any chance.

 

[Synomenon]

Ordered mine back in May and got it in July.

Updated to the latest BIOS for my board and this time the 8 workers ran for 4 hours 24 minutes before all experiencing a fatal rounding error.

I don't know how to adjust voltages for the CPU or adjust timings for the RAM. I've never done any of that before. Anyone have some kind of beginner's walk-through for those?

 

[Ed1]

Its very straight forward .

Go to Asus site, find your MB and download manual .
read the bios section and it will tell you how to do all .

You can set the CPU voltage in 3 ways , default = auto, manual and offset method .

You can also adjust Dram voltage easy to and the ram timings are in a separate memory section with all timings listed .
I would just put ram back to SPD, 1333 and rerun tests .

 

[Puppies04]

OP did you read post #10? I have heard about this more than once.

 

[Synomenon]

I have version 27.7. That's the latest version I saw on the Prime95 website.

Also, when the workers stop, Prime95 doesn't lock up. The workers just stop because they encounter a fatal rounding error.

 

[Synomenon]

I'm convinced the problem is not with my RAM. What's the best way to test my CPU w/o actually damaging it if is not what's causing the Prime95 blend test failures?

 

[2is]

What exactly convinced you it's not the ram? The CPU is typically the most fault tolerant component in the entire box. It doesn't sound like you have done any if the suggestions outlined in this thread. There's no magic bullet, gotta go through real troubleshooting. Hope & faith that it isn't your memory isn't troubleshooting.

 

[Synomenon]

48+ hours with no errors on the latest version of Memtest86+ as well as HCI Memtest. I've redownloaded v27.7 of Prime95 with a different browser at a different time of day. I've also updated to the latest BIOS. I haven't changed voltages and timings though. I can't find a decent walk-through guide on making those changes.