weird problem eith voltage....

[Hugovsky]

weird problem with voltage....

I've been testing my XP1800+ running @ diferent vcore and i have weird results, for example:

@ 2.0v and 1923 (174 x 11) i can't run prime stabelç more than 5 min
@ 1.95v same speed i cn run it 24 hours or more (not tested beyond 24 hours)
@ 2.20v i can only run prime for a minute or two


less than 1.95v not tested yet

What do you think of this ?

 

[Opie420]

I had similar results with two XP2000's on both a prometia and a vapochill. At much over 1.95v the system became less stable. My 2400+ on the other hand seems to like the voltage and has been looping prime95 at 2525mhz 2.05v for more than 24hrs now.

 

[Hugovsky]

Do you think it's Prometeia related ?

 

[Opie420]

I think it's related to the palomino core. If you look around on the vapochill forums you might find something about it there, I think I saw some things on there about it.

 

[The Coolest]

It might be PSU related, what is your PSU, and what are your rails at the specified vcores?

 

[Hugovsky]

Do you think it could be the PSU ? Mine is a no special brand 350W and it's connected to 2 hard disks 2 cdrom and 3 80mm fans not working all time.
12V : 12.11V
5v : 4,91v +/-

Do you think the PSU cn't handle it at that voltages ?

 

[hellrazrblade]

its possible that its sucking too much power for the cpu and the rest of the system is becoming unstable or it might be a heat issue. the higher voltage is giving you more heat and it crashes.

 

[papounet]

Nothing weird here.

A CPU is designed to work at a specific voltage. pushing it beyond specs (or 10-15% over specs) is not garanteed to produce anything of value.

You have a very good cooling system, which is taking care of the excessive heat generated by overvolting and overclocking through frequency, but a cooling system does not change the amount of electrons needed to change a logic gate.
It happens than overvolting produces stronger signals (which helps when you are overclocking frequencies), but at some point too much voltage also cause random noise to be higher, causing failure.

(an analogy, if you put too much water pressure in a network of garden hoses, leaks will happen at various points, in CPU the higher the voltage, the higher the leakage current => at some points some gates will be operated by trickles of current trickling out of places they should n't be trickling from ;-)

Keep in mind that long operations at excessive voltage causes electro-migration to occur at a faster pace.

I beleive you have experimented with the voltage , and you should stick to the maximum safe value, which happen to be for your CPU and setup 1.95V. If you can lower it further and keep the same overclock, the better.

 

[Hugovsky]

Yes, i think you are probably right papounet, because i can't go any higher than that when i increase voltage. It just POST at higher speeds but no luck with windows no matter what voltage. I can POST at 2125 (11.5 x 185) at 2.0v or 2.20v, it's the same. I'm now at 1.95v and 1950 and i ran prime for 32 hours straight with no errors. I'll try a lower voltage maybe this weekend. Oh, about eletro-migration, i think it shouldn't be a problem, 'cause i plan to buy new tbred2 very soon nd i'll put my xp1800+ on a cheap mobo just to do F@H with no o/c or small O/C. Maybe till next month E.M. doesn't burn it.