• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Interesting Problem: power stability of MB+PSU when going from single to dual core

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Sorin

Member
Joined
Mar 7, 2004
Location
Phoenix, since 03/2014
Interesting Problem: power stability of MB+PSU when going from single to dual core

I want to give the whole story, but in the interest of readability I'll try to keep it short while still trying to have enough information to make sense.

Earlier this year I decided to upgrade my box from a socket A, to a socket 939. I went with an Abit AN8-SLI, A 3700+ San Diego, Fortron Sparkle 550w 24-pin PSU, and a 7800GT 256MB PCI-E; the rest was the same as before. I won't reiterate the problem I had (it was a memory issue that I eventually got sorted out, but unrelated to the current topic), but I switched from the Abit, to a Asus A8N-SLI Deluxe. After I got my memory issue fixed, I decided to overclock, so I did. I got my 3700+ to a stable 2.9GHz, including 36 straight hours of Prime95. Then a week or so ago, I decided to go dual core, as I multi task and encode videos (Auto GordianKnot comes with a threaded version of the Xvid encoder - it's fabulous). So I got a 4400+. Enter the problems.

From the time I upgraded to the socket 939 single core 3700+ setup early this year (February) up until now, my box has been rock solid and stable except for the memory issue I mentioned earlier, which got fixed. So I bought and dropped in the 4400+ last week, and after about a week of burning in, plus letting the AS5 settle, I thought I'd try to overclock. No dice. I couldn't do JACK basically above stock, because I'd error out of everything. So I thought I might have a bunk chip, but that thought quickly went away because others with this 4400+ stepping are getting fantastic results. Then suddenly my box started to randomly shutdown. At first it seemed to be under load, and overclocked. But I tried various settings and situations, and everything from stock idle, to overclocked load would occassionaly get a completely random shutdown. I'm not talking Windows shutdown. I'm talking the kind of shutdown where the computer just turns off, as if the power to your house went out.

I suspected motherboard at first, (too long to explain why). I briefly suspected PSU, but thought "Naaaah." I eventually hit google and did a lot of searching. After a couple of hours, I came by this thread. People with Asus motherboards and Antec power supplies seemed to be having the exact same random shutdown problem that I did. There is a legit Antec representative who posts on the forums and got involved in that thread. Between Antec and Asus, they did some talking and revising of products (PSU for Antec, MB for Asus), and seemed to have a large number of the problems fixed.

However, if you read the thread carefully, there are still people having problems with different motherboards (just about everything under the sun from Asus, including A8N32-SLI Deluxe, which I was this close --> | | <-- to ordering as a hoped fix) , including from other manufacturers. One might think it was an Antec issue then. But I don't think it is, at least not wholly, since I have a Fortron Sparkle PSU and I'm having the exact same problem (and I do have the rev. 1.02 board that supposedly fixed the problem, according to the AntecRep).

But I'm thinking it's not just a PSU/MB combo, but maybe something relating to single core power draw versus dual core power draw? Because I ran for 7 months with absolutely zilch power or other stability issues, both stock and overclocked, with my single core. It was only after I installed my dual core that the problem described in that thread (and elsewhere via Google) started happening.

Any thoughts? Ideas? Are there just really bizzare power delivery issues that need ironing out in various motherboard and PSU combinations with dual core processors? Maybe it's even an Nforce and dual core issue (both Nforce4 and the 5xx seemed to have the issue too)? Maybe it's MB? Maybe it's PSU? 4 phase power versus 8 phase power? Single 12v rail versus dual 12v rail? Maybe it's something else entirely?

I'm really hoping to find some solution to this, because it'll be a cold day in Hell before I give up my dual core and go back to single. Besides that, I just thought I'd share what's happening and what I've come across.....because this really sucks.
 
On my way out to run errands, but I stopped to read this since you took the time to write it.

Why do I not believe it's the power supply? Why do I not believe it's the motherboard? Why do I feel so certain it's just the CPU you bought?

You said yourself you suspected the CPU, but then you said "people with the same revision weren't having this problem." What does that have to do with YOU and YOUR CHIP? I would RMA the thing, or take it straight back to the store since you just got it. If only because you can now (as opposed to waiting a few years from now only to find out it really was the CPU) and that would at least rule-out the chip. If your other chip worked fine with the same motherboard and PSU, then why shouldn't this chip? You have to look at what YOU added. And not go into "Unsolved Mysteries-search-for-the-clues" mode.

If I switched out my RAM and my computer suddenly stopped working, I'd be pretty damned sure it was the new ram that did it.
 
i feel your pain, i cannot help you, but i can relate.. i have this fatality here, and 3700, the board an ram dont particularly get along well, but if i dont mess with the bios ever 2 minutes its fine. the cpu is great. clocks nicely, and is stable as a table. then i bought a 4400 to replace the 3700, and boy did i have troubles. ok, the 4400 i had was pretty sucky, but that cpu, this mobo and this ram= bad things to happen. i do believe a dc cpu draws a bit more jo0s, well, at least thats what the PWM temps say, since they pretty much increased about 25-30% since adding the 4400. well, after about 2 weeks with the 4400, my abit died. off to RMA it went. i picked up an asus a8nsli,good board btw, ram problems went away, but that 4400, i dont know, i cant put my finger on it, but it kicked the crap out of both mobos i have. i ditched that cpu and all of my troubles went away. ive used this psu since my nf7/xpm days, and its been solid, and it still is, but maybe it wasnt enough to run a dc. anyways, sorry for the ramble, its early still..
 
WonderingSoul said:
Have you down your BIOS Updates for dual core stability?
^^ good idea.
Also reduce hardware to bare minimum and see if gets better.
If not then as suggested it must be the cpu (mem controller maybe?)
Or could be the PSU waring out caps.
what are the amps on each rail and how is the total amp rated?
a lot of time you have to look close to how the max output is calculated.
Fortron is a bit tricky. I have read the spec on them and how the max output is calculated is deceiving. link to the psu spec and I show you what I mean.
 
Yup, already updated BIOS awhile back, since I've been reading about dual core issues here and there for months.

I'll have to hunt down a link to my specific power supply, but last night I looked at it and wrote down the numbers, which are:

+3.3v - 27A
+5V - 29A
+12V - 36A
+5Vsb - 2A
-5V - 0.3A
-12V - 0.8A

edit: I think this is the one. I don't remember the model number of my specific power supply, but the numbers at this link do match, and it sounds like the one I have. http://www.newegg.com/Product/Product.asp?Item=N82E16817103517
 
I really, really, (feel like I need to add another one here) REALLY don't think it's your power supply. I mean... 550 friggin watts... 36a on the 12v... And a FORTRON at that! I do NOT think it is your powersupply. I'm as certain as I can be without actually being the AMD RMA guy standing there in a white lab coat going "Yup... It's the CPU..."

I mean you COULD go on this wild goose chase with the power supply... but you're wasting your time.
 
Sorin said:
Yup, already updated BIOS awhile back, since I've been reading about dual core issues here and there for months.

I'll have to hunt down a link to my specific power supply, but last night I looked at it and wrote down the numbers, which are:

+3.3v - 27A
+5V - 29A
+12V - 36A
+5Vsb - 2A
-5V - 0.3A
-12V - 0.8A

edit: I think this is the one. I don't remember the model number of my specific power supply, but the numbers at this link do match, and it sounds like the one I have. http://www.newegg.com/Product/Product.asp?Item=N82E16817103517
Well this power unit is the sparkle which is one it's better unit.
But even this one is a bit tricky.
If you add the 3.3v+ 5v rail's amp you should get 234 watts ( 3.3x27+ 5x29= 234watts) but the spec says the maximum combined wattage for 3.3v+5v can not be more than 150 watts. Another word shorting you by 85 watts.
Tricky isnt it?
spec on this unit
But I dont think its power supply doing it. If you strip your system to bare minimum (1 HDD, no CD, no Floppy, no extra fans, no lights) and if still having problem then you know its not the PSU.
 
Yes it very well could be the CPU, considering the description of the problem.

One of the things I found last night (from that thread, incidentally), was to plug in the 4-pin molex connection on the motherboard, despite that it is supposed to be for when SLI is used (the same power connector you use for hard drives and optical drives, that is).

I fired up dual Prime95s at stock settings (since it started doing it at stock too), and so far it has been going for 7 hours. Which is magnitudes of % farther than I ever got before in any situation with any program.

Time will tell if the problem is gone, but if it is, :rolleyes:
 
Sorin said:
Yes it very well could be the CPU, considering the description of the problem.

One of the things I found last night (from that thread, incidentally), was to plug in the 4-pin molex connection on the motherboard, despite that it is supposed to be for when SLI is used (the same power connector you use for hard drives and optical drives, that is).

I fired up dual Prime95s at stock settings (since it started doing it at stock too), and so far it has been going for 7 hours. Which is magnitudes of % farther than I ever got before in any situation with any program.

Time will tell if the problem is gone, but if it is, :rolleyes:
let me see. dont you have 2 sets of ATX powers on the board? one is the 20 pins the other is 4 pins (square)? the 4 pins have 2 yellow wires(12v) and 2 blacks (grounds). have you connected this square looking 4 pins ATX?
if not, then I dont see how the board was functioning.
 
Yes, of course. the 24-pin and the 4-pin that's in a 2x2 arrangement. I always make sure to have those plugged in.

The extra one was a 4-pin 1x4 arragement. The exact same kind that hard drives, dvd drives, etc. use. This motherboard has one of those on it. It's suppose to be for when you're using SLI, if I read right, but one of the suggestions was to plug it in anyways if having this random shutdown problem, even if not in SLI. So that's what I did.
 
Sorin said:
Yes, of course. the 24-pin and the 4-pin that's in a 2x2 arrangement. I always make sure to have those plugged in.

The extra one was a 4-pin 1x4 arragement. The exact same kind that hard drives, dvd drives, etc. use. This motherboard has one of those on it. It's suppose to be for when you're using SLI, if I read right, but one of the suggestions was to plug it in anyways if having this random shutdown problem, even if not in SLI. So that's what I did.

I have the same board ^^. You should have power for the 24pin, a small 12v (small 2x2), and a 4 pin molex right above your video card. With this board I know you might have to lock the PCI Bus Clock at 33.33 otherwise your OC won't be crap. It happened with my 3700, Couldn't get above 2.4 But when I locked it I was able to hit 2.6 fairly easily... Now I just need to know why I can't hit 2.7 without errors >.<
 
Back