X2 4400+ Core 1 Errors when Benching Core 0

[Theta]

I'm testing out a new 4400+ we got in today, and having some strange problems with it.

Right now, at 1.7v and 2800, Core 0 is stable as can be. Core 1 is fairly unstable at 2800. So, backing down to 2700 @ 1.7v, both cores seem rock solid.

However, I can only test Core 1 WITHOUT testing Core 0. As soon as I test Core 0 at the same time, POOF - error on Core 1. This seems to be more than a coincidence.

Anyone seen this, or am I just nuts?

BTW - 1.7v to not even hit 2800 on a 4400+ Toledo seems like a DOG...

[eBoy0]

I'm testing out a new 4400+ we got in today, and having some strange problems with it.

Right now, at 1.7v and 2800, Core 0 is stable as can be. Core 1 is fairly unstable at 2800. So, backing down to 2700 @ 1.7v, both cores seem rock solid.

However, I can only test Core 1 WITHOUT testing Core 0. As soon as I test Core 0 at the same time, POOF - error on Core 1. This seems to be more than a coincidence.

Anyone seen this, or am I just nuts?

BTW - 1.7v to not even hit 2800 on a 4400+ Toledo seems like a DOG...

If your running air back down on the voltage... make sure you're setting affinity correctly.

[Theta]

Affinity is set correctly.

About the voltage, what should my max be for a CNPS9500?

Load at 1.75v is below 55C. 1.7v is 51C, etc.

[eBoy0]

hmmm for air 1.75 is WAY too much man... your chip wont last you more than a month IMO

[Theta]

Yeah, I admit... it's worrysome.

It will be back at 1.6v or under once I figure out what the hell is holding me back...

[afireinside]

My chip starts locking on core 1 as soon as I hit 50C... back down

[Pyr0]

with what program are you testing?
if it's something like super pi or prime, make two folders and put a copy of the prog in each

[Theta]

Good suggestion about the seperate folders, did that immediately.

The voltage problem I'm having is that the chip can't even do 2600 stable at 1.55v, let alone any less.

It takes at least 1.7v to hit 2700, let alone bench it...

Here's the sad part - it looks like a good chip:

ADA4400DAA6CD
CCBWE0529SPMW

[Theta]

I'm thinking that I found the cause of this nonsense...

My board seems to be feeding some screwy voltage. Set at 1.65v on the board, it gives 1.61v when Core 1 is active, and only 1.56v when both cores are active.

No wonder it's failing - it's dripping an entire .1v under full load.

Geez...

[Pyr0]

no probs
they need their own folder to keep separate sets of data as they calculate. just running the prog twice from the same folder mixes the data up and so you get errors

wow i've just noticed you put 1.75v on your 4400
the most i've put on this chip has been 1.623v, normally it runs 24/7 at 1.586v

[edit] you posted as i was typing :P
how are you measuring your voltages?
try disconnecting any non essential devices to try to reduce the power drain, that does seem like a large drop

[DAK1640]

Holy Geez Batman...Your using AC line voltage almost...My 4400+ has seen a high of 1.475v and I was freaking out (my 1st AMD). I can hit 2.75Ghz at 1.500V and that's the highest voltage I've evr tried. You guys are talking about HIGH VOLTAGE...1.7V is enuf to power my furnace & clothes dryer. I don't know how the CPU can survive. I thought 1.50V was max "safe" voltage?

[eBoy0]

Holy Geez Batman...Your using AC line voltage almost...My 4400+ has seen a high of 1.475v and I was freaking out (my 1st AMD). I can hit 2.75Ghz at 1.500V and that's the highest voltage I've evr tried. You guys are talking about HIGH VOLTAGE...1.7V is enuf to power my furnace & clothes dryer. I don't know how the CPU can survive. I thought 1.50V was max "safe" voltage?

It'll survive.. but no on air, i ran a sandiego @ 3.35@1.72v under -30c.

[Theta]

Well, I'm glad I only suicide tested it at that voltage, then. 5-7 minnutes, max.

It's living at 1.6v now, but the damn thing still can't do anything at that voltage...

Ugh...

[Theta]

[edit] you posted as i was typing :P
how are you measuring your voltages?
try disconnecting any non essential devices to try to reduce the power drain, that does seem like a large drop

I'm using SpeedFan and CPU-Z to monitor the voltage. This board is EXTREMELY difficult to real-time monitor, as a needle is needed to test the vCORE point.

[Pyr0]

are the readings close to what you set in the bios?
in my experience all the software i have ever tried reads the voltages a bit lower than they actually are
for example, i have 1.586v on my cpu, but cpu-z sees it as 1.52v

[Theta]

The Epox 9NPA+ is notorious for voltage issues...

Hell... I might have only had 1.65 on it since the voltage is wacky.

Well crap, I'm going to reapply AS5, reseat the CNPS9500, then go back at it.

[jinu117]

Reason is quite simple. By running both cores, you are draining more power from mobo as well as putting more heat out through cpu. Temperature will go up obviously (about 5c at least if not more) around voltage you are playing with... :P Obviously your overclock will be joepardized. Reason why dual prime is much harder animal to pass on X2 with both core than just single core by easy 100mhz-200mhz margin.

[shoman24v]

My 3800+ Toledo will hit 2.65 @ 1.5v

[Theta]

My 3800+ Toledo will hit 2.65 @ 1.5v

Apples to oranges... that may be a similar core, but with 512k cache. Many 3800's have been getting great clocks, but 4400's and 4800's have been a bit lacking in the OC department.

[dng29]

just want to add my $.02 here. i have a similar problem you do with my x2 4400. core 0 is rock solid, but core 1 is a POS. however, now that winter is here, i can open the window a little bit for some cool air. i'm able to run pretty stable at 2.8Ghz on my dfi nf4 sli-dr with a big typhoon at 1.65vcore. it's pretty cool in my room with ambient at 20-30 degree C and case ambient is between 20-30 degree c. the cpu idles around 28 C. i was able to run dual prime 95 for 12 hours. i also looped quake 4(optimized for dual core a64) for 16 hours . i'm enjoying this in the winter, but will probably need to back down the megahurts in the summer .

[s7e9h3n]

My 4400+ wasn't too bad :

[Pyr0]

hehe, sub zero cooling certainly helps
i'd love to try it.... someday

[Theta]

Switched to low-end watercooling, slightly better results... Using a CoolerMaster Aquagate (the big one, not the minis).

Might have been a heat issue - the CNPS9500 kind of sucks for the price.

2600 is now stable at 1.575v according to dual SuperPI 32M, etc. Prime still nowhere close. I have not had luck with Prime in a LONG time though, so I'll trust 3DMark05 and SuperPi.

Still a crappy chip, IMHO. 55C load also sucks.

[blind_ripper]

i can hit 2.86ghz now on wc withe 1.75vcore idle 24°c load of the 2 cores is 40°c , i can bench al day long . next up is a modded mach2 thate wil do -70°c or so 0% load hopping to hit 3ghz or more for bench en 24/7 use , btw ss max @ sig