Solution : reason for failing conroes at prime

Printable View

Show 100 post(s) from this thread on one page

08-30-2006, 10:20 AM
M.Beier

Quote:

Originally Posted by Revv23

yeah i remember early AMD winchesters were doing something similar, they would faiil prime even at stock speeds sometimes, but there wasnt any realy problem with the cpu's.

I belive that was due the IMC being weak.. - And timmings with BH5 being to aggressive, not 1.5-2-2-5, but some of the others..
08-31-2006, 03:28 AM
Entsafter

My E6600 is strange under dual-prime load...
2950 MHz with 1.27V dual-prime stable over hours...
3000 MHz with 1.3 - 1.6V unstable!

I lowered the multi from 9 to 7 and started testing with FSb 400+ ... same problem!
Mobo: P5B Deluxe Wifi
I tryd different settings:
Vcore: 1.3-1.65V
FSB Termination: Auto-1.45V
NB: Auto- 1.65V
SB: Auto - 1.7V
I tryd also every Multi but the problem still exists.. I can't run prime over 2950 MHz... but I can run Super-Pi till 3510 MHz with 1.39V.. and above 3600Mhz with more Voltage... really strange...
I use Corsair 8000UL and 1:1 divider, I removed also one stick ... but the situation is still the same...
I don't know where's the problem! Probably it's only one setting in bios!

btw: speedstep and other voltage reducing and multi reducing features are off...
08-31-2006, 05:08 AM
Vric

Dual prime fail after 2 hours on my E6600 at 2.9Ghz, but Orthos pass for over 8 hours.
08-31-2006, 06:18 AM
Artmic

Interesting, i OC'ed my x6800 to 3.46Ghz and it failed Orthos after only 3min testing CPU+Memory.

I got a TR SI-128 heatsink and panaflo M1 fan, so i figure my fan blows(just not air) so i'm replacing the fan with the panaflo H1 fan right now, i figure 103.8 cfm should do the trick and cool it down even more. (maybe)

I can see how the people that purchased the e6600 can love their processors, for me the X6800 leaves a bad taste behind, you'd figure after spending so much money and having an open Multiplier, you'd be able to do 3.46Ghz with even the stock cooler, never mind an aftermarket SI128 with a fan that sounds like a tractor pushing so much air onto it.

Well my retail X6800 is now running 3.2Ghz, stably. I will try to attempt the 3.46 with the new fan.

There is a reason Intel lowered the fastest Conroe from 3.2Ghz to 2.93Ghz, ... and it has nothing to do with marketing or squeezing more money out of joe-six-pack..... i figure there is only so far they can take these C2D on air, before they start roasting pigs on an open fire.
08-31-2006, 08:07 AM
chrisf6969

Quote:

Originally Posted by Artmic

There is a reason Intel lowered the fastest Conroe from 3.2Ghz to 2.93Ghz, ... and it has nothing to do with marketing or squeezing more money out of joe-six-pack..... i figure there is only so far they can take these C2D on air, before they start roasting pigs on an open fire.

Thats what I'm thinking. Before I thought they did it just so they could gradually roll out new products instead of blowing their whole wad at the launch. But I'm having problems right around 3Ghz too.
08-31-2006, 08:11 AM
Entsafter

Hmm I've tested my second E6600... same problem.. but 100 Mhz higher... STRANGE!!!
I sell all Core2Duo stuff and buy an old D950... it's more stable than this new superduperfast CPUs... I can run Super-Pi 600 MHz higher than Dual-Prime.. that sucks hard... the point is that I can only use this cpus @ 2.95 GHz for daily use... with 1.25V or less... raising vcore doesn't help to improve the stability... Maybe the board ist crap... expensive P5B Deluxe Wifi! Bios 0507..
08-31-2006, 11:58 PM
realsmasher

EVERYONE who uses conroe overclocked, leave this thread IMMIDIATLY.

this here discussed problem has NOTHING to do with OC, so go away. people like you bring trolls here that say : "hoho not a bug, it's you OC".

i repeat it again for all those "wise guys" : this is not related to OC or any stability.

it appaers at ANY clock with ANY voltage but only on SPECIAL Chipset/bios/ram settings, but there is actually no way to find out why.

so this flag-bug COULD be a good reason.
09-01-2006, 01:42 AM
astaris

Quote:

Originally Posted by realsmasher

EVERYONE who uses conroe overclocked, leave this thread IMMIDIATLY.

this here discussed problem has NOTHING to do with OC, so go away. people like you bring trolls here that say : "hoho not a bug, it's you OC".

i repeat it again for all those "wise guys" : this is not related to OC or any stability.

it appaers at ANY clock with ANY voltage but only on SPECIAL Chipset/bios/ram settings, but there is actually no way to find out why.

so this flag-bug COULD be a good reason.

This flag bug has nothing to do with the reason Prime fails, if the two things where connected then all people here (overclocked or not) should experience Prime rounding error but it's not this way...
If Prime fails then your system is unstable (due to overclocking, bad memory, poor bios, poor bios settings...)
You started a thread titled "Solution : reason for failing conroes at prime", then before saying "EVERYONE who uses conroe overclocked, leave this thread IMMIDIATLY", then change the title thread to something else i.e. "guys look, Conroes has a bug...", so the one trolling here is only the thread starter, of course due to the title of your thread, due to the fact that we are on xtremesystems and not on underclockedsystems, there are lot of people that overclocks that post in this thread...
And about your statement "it appaers at ANY clock with ANY voltage but only on SPECIAL Chipset/bios/ram settings, but there is actually no way to find out why", what happens, that prime fails? -> speak for yourself / your system, that conroe has a bug? -> who cares?
09-01-2006, 06:27 AM
uOpt

I ran dual mprime for like a week after I stabilized my E6600. At 3600 MHz...

This CPU bug has no impact on mprime. I read mprime's source code. It's not using that flag that might be incorrectly not raised by this CPU bug.
09-01-2006, 07:46 AM
mds47

Quote:

Originally Posted by uOpt

I ran dual mprime for like a week after I stabilized my E6600. At 3600 MHz...

This CPU bug has no impact on mprime. I read mprime's source code. It's not using that flag that might be incorrectly not raised by this CPU bug.

QFT

The flag has absolutely no effect in prime95. It's just a status flag you can optionally check on the FPU, and prime95 does not check it.

If you're getting errors in prime95 at stock, it's either a different cpu bug (not listed in the errata) or a problem somewhere else in your system.

mds
09-01-2006, 07:57 AM
[XC] hipno650

why use prime at all? why not get the ultimate stabilty test and help fight cancer and aids!!!!!!! come over to the dc valut and go to the WCG section and we will set you up from there!!!!!:) if your conroe is unstable you will know it by running WCG!!!!!!

Uncle Sam wants you to fold WCG!!!!!!!!
09-01-2006, 10:42 PM
realsmasher

so you guys say it cannot be a bug as not everyone gets this.

but what about that : the bug only appears together with other stuff ?

as you see, i'm not the only one having this "problems" and if prime would recognize an error SO early, don't you think the computer would crash at other stuff ?

how can it make sense, that only ONE program crashes due to instability ?

and the reason why i write this here : there are several threads asking for a reason/solution for this BUG.

yes in my opinion it is a bug, cause my system is even stable with 0.2V lower with all other programs(and prime small ftt), but not even at stock or higher vcore at large ftt. there is no other logical reason then a bug.

plus : is comes on different rams,, ram settings, different boards und different cpus(not only conroes, also allendales).
09-02-2006, 02:19 AM
Shintai

Quote:

Originally Posted by realsmasher

so you guys say it cannot be a bug as not everyone gets this.

but what about that : the bug only appears together with other stuff ?

as you see, i'm not the only one having this "problems" and if prime would recognize an error SO early, don't you think the computer would crash at other stuff ?

how can it make sense, that only ONE program crashes due to instability ?

and the reason why i write this here : there are several threads asking for a reason/solution for this BUG.

yes in my opinion it is a bug, cause my system is even stable with 0.2V lower with all other programs(and prime small ftt), but not even at stock or higher vcore at large ftt. there is no other logical reason then a bug.

plus : is comes on different rams,, ram settings, different boards und different cpus(not only conroes, also allendales).

BIOS update will fix with microcode corrections. And the reason SP2004 fails is roughly due to it being coded wrongly. But ofcause it should still work.
09-05-2006, 05:00 AM
uOpt

Quote:

Originally Posted by realsmasher

so you guys say it cannot be a bug as not everyone gets this.

Nobody said that. All we say is that Mprime does not use the CPU feature that has this particular bug.

Quote:

Originally Posted by realsmasher

but what about that : the bug only appears together with other stuff ?

as you see, i'm not the only one having this "problems" and if prime would recognize an error SO early, don't you think the computer would crash at other stuff ?

how can it make sense, that only ONE program crashes due to instability ?

Two reasons:

1) Mprime ist just very intense heat-wise

2) Mprime in self-test mode does a test (duh) of its results and detects when things are wrong. Most other programs continue to compute with previous wrong results without noticing.

Quote:

Originally Posted by realsmasher

and the reason why i write this here : there are several threads asking for a reason/solution for this BUG.

yes in my opinion it is a bug, cause my system is even stable with 0.2V lower with all other programs(and prime small ftt), but not even at stock or higher vcore at large ftt. there is no other logical reason then a bug.

plus : is comes on different rams,, ram settings, different boards und different cpus(not only conroes, also allendales).

This CPU bug CANNOT have to do with the mprime instability when overclocking core2. Mprime does not use that flag that is sometimes not raised when it should.

You also seem to forget that mprime gets stable if you take the clocks back enough (or raise voltage when appropriate).

Face it: Conroe just has more headroom for usable but not entirely stable overclock. But to be entirely mprime stable you have to take those clocks back to stables levels simple.

I also remind people that the math errors disappear when you take back the clocks.
09-05-2006, 05:19 AM
uOpt

Quote:

Originally Posted by realsmasher

so you guys say it cannot be a bug as not everyone gets this.

Nobody said that. All we say is that Mprime does not use the CPU feature that has this particular bug.

Quote:

Originally Posted by realsmasher

but what about that : the bug only appears together with other stuff ?

as you see, i'm not the only one having this "problems" and if prime would recognize an error SO early, don't you think the computer would crash at other stuff ?

how can it make sense, that only ONE program crashes due to instability ?

Two reasons:

1) Mprime ist just very intense heat-wise

2) Mprime in self-test mode does a test (duh) of its results and detects when things are wrong. Most other programs continue to compute with previous wrong results without noticing.

Quote:

Originally Posted by realsmasher

and the reason why i write this here : there are several threads asking for a reason/solution for this BUG.

yes in my opinion it is a bug, cause my system is even stable with 0.2V lower with all other programs(and prime small ftt), but not even at stock or higher vcore at large ftt. there is no other logical reason then a bug.

plus : is comes on different rams,, ram settings, different boards und different cpus(not only conroes, also allendales).

This CPU bug CANNOT have to do with the mprime instability when overclocking core2. Mprime does not use that flag that is sometimes not raised when it should.

You also seem to forget that mprime gets stable if you take the clocks back enough (or raise voltage when appropriate).

Face it: Conroe just has more headroom for usable but not entirely stable overclock. But to be entirely mprime stable you have to take those clocks back to stables levels simple.

I also remind people that the math errors disappear when you take back the clocks.

Show 100 post(s) from this thread on one page