Long time P95 user now experiencing errors with new version.

Sugam · 2016-03-23, 01:12

Hello First post.
I have been using prime 95 for YEARS and I have never seen these rounding errors before.

I was using 28.7 for the first time and it just shut down on me saying
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

Blend test

Now if I use 26.6 I get no errors.

I have ran memtest86 for 20 hours 7 passes with zero errors.

I don't know if i have an incompatibility problem with the new software or such.
I do have an odd possibly unsupported memory configuration

Hardware
Motherboard : Crosshair IV Extreme
CPU : AMD Phenom II X6 1100T Black Edition Thuban 6-Core 3.3GHz
Memory : 16 GB 2x G.SKILL Ripjaws Series 8GB (2 x 4GB) 240-Pin DDR3 SDRAM DDR3 1600 (PC3 12800) Desktop Memory Model F3-12800CL9D-8GBRL
GPU : MSI Radeon HD 7970 DirectX 11 R7970 Lightning
PSU: CORSAIR AXi series AX1200i 1200W Digital ATX12V v2.31 and EPS 2.92 SLI Ready CrossFire Ready 80 PLUS PLATINUM Certified Full Modular Active PFC Power Supply

Windows 7 or windows 10 whatever flavor i feel like.

Now Asus seems to have issues giving people a reasonable QVL for memory because they are selling a board that supports 16 gb but does not bother to actually list any 16 gb configurations or even a 8 gb dual channel configuration but that's a whole other topic

The board does not seem to like my ram much because when i use my memory at the 1.5v that g.skill says memtest86 errors

Increasing the voltage increases the errors as well causes BSOD's

Also the mother board actually slightly over volts the memory so if i say 1.50 in the bios its actually 1.53~1.54

I honestly wonder if the board is very inaccurate because it reports my 3.3 , 5v and 12 v low (IE 3.24) when my voltmeter and corsair dongle show spot on values.

So i need to under volt my memory to 1.45 (reported as 1.48) so that memtest86 passes long extensive testing. (20+ hours) it is the lowest i can go without errors.

It is very odd to me but i just go for it. So lets keep this in mind.

I really don't know what to say about the prime95 v28.7
I feel like my system is stable as is. I don't BSOD.
Is it giving me a false positive due to my odd configuration?

I am going to try 28.5 for a while and post those results

Can someone advise?

Prime95 · 2016-03-23, 04:07

I'm not an AMD expert. Frankly, its been so long since v26.6 was released that I'm not sure what all changed in the meantime. I do know that none of the optimization work was targeted to AMD chips. Thus, I suspect that even if you get version 28.7 stable, you will not see much of a performance gain. That said, there is no reason 28.7 shouldn't run reliably on all AMD chips.

Hopefully, some of our AMD users can comment more intelligently on your particular setup.

kladner · 2016-03-23, 15:01

I have been running 28.7 Build 1, on an FX-8350, for months without issues.

Asus Crosshair V Formula Z board
32 GB Crucial DDR3-1600
2x GTX 580 GPU
EVGA 1200 w PSU
Samsung 850 Pro, 256 GB
a bunch of HDDs
CPU and RAM at stock

I did not respond at once because the problem is puzzling. I still don't have an idea as to why the OP is encountering instability. I will think about it and keep track of this thread.

pepi37 · 2016-03-23, 15:05

Motherboard : Crosshair IV Extreme
CPU : AMD Phenom II X6 1100T Black Edition Thuban 6-Core 3.3GHz

1. You didnot "unlock 6-th core" on the CPU, I suppose?
If answer is no, so CPU has fabric 6 core , then your MOBO is defective.
Why I say this: I have virtually same situation, and in normal work I dont have BOSD, but on Prime 95 , I got errors.
At the end of story MOBO was defective.
1% can also be reason, and it can be funny one, dust on memory modules...

The board does not seem to like my ram much because when i use my memory at the 1.5v that g.skill says memtest86 errors

Then either is memory or again MOBO fault. But must be fault. since I never, will repeat never got error from Prime95 and to be false one. Something slowly dying, but will die...
Change memory... retest again: memory must work at 1.53 V, mine is at 1.65 and right now works 23 days in piece without any error.

Sugam · 2016-03-23, 23:24

First thing i want to say is unrelated but needs to be said. Besing someone with a disability it can take an hour or 2 or more to write even a small paragraph. When a website forum times you out and forces a log off "times out" it just makes a frustrating situation more frustrating. Thats all. Time to do this again in ms word!

I am happy to hear that other crosshair iv owners exist lol. This is an uncommon board compared to the forumlia. Its not a bad board but mine has some history that others should know.
When i first built the system i had a mid tower. I thought it would be large enough. I mean stuff fit in it but it was crouded. Eventually i found that the system was overheating behind the board. You could put a temp monitor behind the board and it would read 90C even on a infrared thermometer. I eventually realized that this was because the case was just not allowing for airfloor and I bought a new full tower HAF case. The back of the board is discolored slightly white.
But the board has been pretty good to me over the past 7 years. It is a TOUGH BOARD and can take quite a bit of accidental abuse.
Alright!
I know what the problem is.
The CPU does not native support 1600mhz ram. I always thought it does . It was hard to find reliable information on this processor.
So when i have been volting the memory down i was actually slightly underclocking it by starving it.
The speed diffrence is very slight but it is a factor.
Setting the memory to 1333 I no longer get errors in prime 95.
I have alot of testing to do but i may not ever be able to clock the ram at 1600?
Since increasing the voltage always increases the errors i am not quite sure where to start to even try to clock the memory at 1600.
I guess i could relax the timings.
SOMETHING needs to altered to run ram faster memory than the processor specifies but i dont know what controls this.
Would this be the memory controler?!?
The board supports ram thats much faster than what the processor supports.

I am using BIOS 3027 (Im not a fan of this bios its been in beta for 6 years lol)
also i dont know anything about unlocking a core. All cores should be unlocked by default as none are hiden. Also this is not the first board my processor has been in. I once had an MSI board which i returned.

I could enable core unlocker but why would i do something like that?
Could you explain in detail?

kladner · 2016-03-24, 02:03

Quote:

I guess i could relax the timings.

I think this is a good direction to explore. You might find something here:
Republic of Gamers at Asus.com
http://rog.asus.com/
Forum is here:
https://rog.asus.com/forum/forum.php

Sites like this might have helpful information:
http://www.overclock.net/

The articles section is worth checking, too:
http://rog.asus.com/i/crosshair-motherboards

You might have a look at the comments on this article. FlanK3r is a reliable voice on this site. His comment mentions some more obscure (to me) voltage settings, one of which he mentions regarding OCd RAM.
http://rog.asus.com/216552013/crossh...r-v-formula-z/

NBtarheel_33 · 2016-03-24, 10:06

Quote:

Originally Posted by Sugam

First thing i want to say is unrelated but needs to be said. Besing someone with a disability it can take an hour or 2 or more to write even a small paragraph. When a website forum times you out and forces a log off "times out" it just makes a frustrating situation more frustrating. Thats all. Time to do this again in ms word!

So sorry that this happened to you. It has happened to me before on numerous occasions, and yes, it is quite frustrating. What I do now is to copy my entire post to the clipboard (i.e. highlight the text with the mouse and select Copy) before clicking the Post button. That way, if I am inadvertently logged out, I can simply log back in and paste my text into the post creation window again.

Otherwise, welcome to the forum, and thank you for your contributions to GIMPS!

2016-03-23, 01:12	#1
Sugam Mar 2016 2 Posts	Long time P95 user now experiencing errors with new version. Hello First post. I have been using prime 95 for YEARS and I have never seen these rounding errors before. I was using 28.7 for the first time and it just shut down on me saying FATAL ERROR: Rounding was 0.5, expected less than 0.4 Hardware failure detected, consult stress.txt file. Blend test Now if I use 26.6 I get no errors. I have ran memtest86 for 20 hours 7 passes with zero errors. I don't know if i have an incompatibility problem with the new software or such. I do have an odd possibly unsupported memory configuration Hardware Motherboard : Crosshair IV Extreme CPU : AMD Phenom II X6 1100T Black Edition Thuban 6-Core 3.3GHz Memory : 16 GB 2x G.SKILL Ripjaws Series 8GB (2 x 4GB) 240-Pin DDR3 SDRAM DDR3 1600 (PC3 12800) Desktop Memory Model F3-12800CL9D-8GBRL GPU : MSI Radeon HD 7970 DirectX 11 R7970 Lightning PSU: CORSAIR AXi series AX1200i 1200W Digital ATX12V v2.31 and EPS 2.92 SLI Ready CrossFire Ready 80 PLUS PLATINUM Certified Full Modular Active PFC Power Supply Windows 7 or windows 10 whatever flavor i feel like. Now Asus seems to have issues giving people a reasonable QVL for memory because they are selling a board that supports 16 gb but does not bother to actually list any 16 gb configurations or even a 8 gb dual channel configuration but that's a whole other topic The board does not seem to like my ram much because when i use my memory at the 1.5v that g.skill says memtest86 errors Increasing the voltage increases the errors as well causes BSOD's Also the mother board actually slightly over volts the memory so if i say 1.50 in the bios its actually 1.53~1.54 I honestly wonder if the board is very inaccurate because it reports my 3.3 , 5v and 12 v low (IE 3.24) when my voltmeter and corsair dongle show spot on values. So i need to under volt my memory to 1.45 (reported as 1.48) so that memtest86 passes long extensive testing. (20+ hours) it is the lowest i can go without errors. It is very odd to me but i just go for it. So lets keep this in mind. I really don't know what to say about the prime95 v28.7 I feel like my system is stable as is. I don't BSOD. Is it giving me a false positive due to my odd configuration? I am going to try 28.5 for a while and post those results Can someone advise? Last fiddled with by Sugam on 2016-03-23 at 01:50 Reason: Wrong version specified

2016-03-23, 15:01	#3
kladner "Kieren" Jul 2011 In My Own Galaxy! 27AE₁₆ Posts	I have been running 28.7 Build 1, on an FX-8350, for months without issues. Asus Crosshair V Formula Z board 32 GB Crucial DDR3-1600 2x GTX 580 GPU EVGA 1200 w PSU Samsung 850 Pro, 256 GB a bunch of HDDs CPU and RAM at stock I did not respond at once because the problem is puzzling. I still don't have an idea as to why the OP is encountering instability. I will think about it and keep track of this thread.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Desktop Window Manager frequently crashing - anyone else experiencing this?	ixfd64	Lounge	10	2016-01-13 23:44
Will running prime95 for long time harm CPU?	dabaichi	Hardware	21	2011-12-25 06:29
Using long long's in Mingw with 32-bit Windows XP	grandpascorpion	Programming	7	2009-10-04 12:13
I think it's gonna be a long, long time	panic	Hardware	9	2009-09-11 05:11
Too long time to work ... ???	Joël Harismendy	Software	18	2005-05-16 15:05

2016-03-23, 04:07	#2
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 7526₁₀ Posts	I'm not an AMD expert. Frankly, its been so long since v26.6 was released that I'm not sure what all changed in the meantime. I do know that none of the optimization work was targeted to AMD chips. Thus, I suspect that even if you get version 28.7 stable, you will not see much of a performance gain. That said, there is no reason 28.7 shouldn't run reliably on all AMD chips. Hopefully, some of our AMD users can comment more intelligently on your particular setup.

2016-03-23, 15:05	#4
pepi37 Dec 2011 After milion nines:) 5·17² Posts	Motherboard : Crosshair IV Extreme CPU : AMD Phenom II X6 1100T Black Edition Thuban 6-Core 3.3GHz 1. You didnot "unlock 6-th core" on the CPU, I suppose? If answer is no, so CPU has fabric 6 core , then your MOBO is defective. Why I say this: I have virtually same situation, and in normal work I dont have BOSD, but on Prime 95 , I got errors. At the end of story MOBO was defective. 1% can also be reason, and it can be funny one, dust on memory modules... The board does not seem to like my ram much because when i use my memory at the 1.5v that g.skill says memtest86 errors Then either is memory or again MOBO fault. But must be fault. since I never, will repeat never got error from Prime95 and to be false one. Something slowly dying, but will die... Change memory... retest again: memory must work at 1.53 V, mine is at 1.65 and right now works 23 days in piece without any error.

2016-03-23, 23:24	#5
Sugam Mar 2016 2₈ Posts	First thing i want to say is unrelated but needs to be said. Besing someone with a disability it can take an hour or 2 or more to write even a small paragraph. When a website forum times you out and forces a log off "times out" it just makes a frustrating situation more frustrating. Thats all. Time to do this again in ms word! I am happy to hear that other crosshair iv owners exist lol. This is an uncommon board compared to the forumlia. Its not a bad board but mine has some history that others should know. When i first built the system i had a mid tower. I thought it would be large enough. I mean stuff fit in it but it was crouded. Eventually i found that the system was overheating behind the board. You could put a temp monitor behind the board and it would read 90C even on a infrared thermometer. I eventually realized that this was because the case was just not allowing for airfloor and I bought a new full tower HAF case. The back of the board is discolored slightly white. But the board has been pretty good to me over the past 7 years. It is a TOUGH BOARD and can take quite a bit of accidental abuse. Alright! I know what the problem is. The CPU does not native support 1600mhz ram. I always thought it does . It was hard to find reliable information on this processor. So when i have been volting the memory down i was actually slightly underclocking it by starving it. The speed diffrence is very slight but it is a factor. Setting the memory to 1333 I no longer get errors in prime 95. I have alot of testing to do but i may not ever be able to clock the ram at 1600? Since increasing the voltage always increases the errors i am not quite sure where to start to even try to clock the memory at 1600. I guess i could relax the timings. SOMETHING needs to altered to run ram faster memory than the processor specifies but i dont know what controls this. Would this be the memory controler?!? The board supports ram thats much faster than what the processor supports. I am using BIOS 3027 (Im not a fan of this bios its been in beta for 6 years lol) also i dont know anything about unlocking a core. All cores should be unlocked by default as none are hiden. Also this is not the first board my processor has been in. I once had an MSI board which i returned. I could enable core unlocker but why would i do something like that? Could you explain in detail?