[Dabljuh]

I have it clear on paper now. My PSU is too weak.

I got my XP1600 + Asus A7V266 a month ago and have spent the entire time working on it. Now 4 days ago, the conductive ink finally arrived (strangely, that stuff is illegal where I live) and I unlocked it. Before, I had it run on 153FSB and 1.6GHZ @ 1.8V, now reducing the multiplier to 10 and increasing the FSB to 160 caused occasional lockups. I increased the Vmem and Vcore and the Lockups rate increased - I was really wondering why, since when I was doing a Burn-In test like BIT, all would work fine with CAS-2 and so on. But when I start up a program, all of a sudden I get bluescreens and all the happy stuff. My Asus PC probe was ringing in the frequency of the processor because of voltage drops - core temp and case temp all were fine, but sometimes the 5V current jump as low as 4.7V and back to 4.8 under load.

Reducing the current on the Vmen and Vcore lead to more system stability, but finally only lower FSB and core speed lead to the square-rock-stability i need to work with.

This lead me to the solution. My Power supply would put out just enough honey to let the system run at full load, but it would not be able to compensate sudden load and idle times. This would result in random reboots, memory errors, calc errors when I would do something.

I have a generic 300W PSU, thought that would be enough since I have an old P2 here running with 235W PSU just fine. I think I'm getting a 450W PS as soon the holydays are over, so I would finally be able to break that 1.6 giga and the 160FSB barrier I have until now. But I hear that even with an Enermax I would get only 4.8 or so at my 5V line, and Enermax and "Generic" seem to be the only kinds of PSU I can get here...

Suggestions?

[el]

is your p/s on the amd approved list? trust me check that first.

[Yodums]

Usually generic PSU's are never good no matter what and for an AMD Xp you should really get like a 350 or 400 watt Enermax PSU because they are AMD approved and they are a great brand and you'll have the bang of the buck from them...

Something tells me your generic PSU doesn't have enough amps in the 5v rail since AMD recommends 32amp or something like that.


-Peace

[Dabljuh]

Yes. It got only 30 Amps on the +5V line, a bit weak it seems. The Enermax model 431 has got 44 amps on that line, should do it. What I worry about is the voltage - My generic only delivers like 4.8 to 4.85 volts, and I heard the Enermax would suffer the same problem...

Well, i sometimes have as low as 4.7V under heavy load and heavily overclocked, but mostly at this point the system would crash very soon.

Crashes mainly happen when I just start up a heavy load or stop one. That tells me that my PSU is too weak. Also, PC probe reports all the day long that my juice is not enough.

Right now, this is the only thing that prevents me from getting my system higher than 155mhz x 10 - when I go at 1.6ghz, which worked fine when I had the multiplier to 10.5 stuff would occasionally lock up, reboot or crash. Also my Hardware monitor goes crazy and would beep all the time because of no juice.

I hope when I get that Enermax (109$ over here) I'll be able to breach 1.7 ghz

[windyridge]

my sparkle 300watt the 5v line gives me 4.7 under load and 4.9 idle and is fin running my cpu but i'm not pushing it that much

[Hoot]

I'm still not tired of saying this, but you can not necessarily blame your PSU for low 5V readings under load. Many PSUs, both name brands and no-name, deliver a true +5V output, but the system experiences a voltage drop across the ATX connector due to contact resistance between the mating pins. Even with several +5V conductors in parallel, when you draw 30A or so through them, it only takes a few hundredths of an ohm resistance to cause a noticeable voltage drop. To truly assess whether your PSU or the ATX connector is to blame, you need an accurate voltmeter. With your system running under load (Prime95 Torture), see what the motherboard says the +5V is running at. Then take an unused molex connector and read the +5V present on it (Red & Black) wires. You will no doubt see a difference from the motherboard reading. That difference is the voltage drop across the ATX connector due primarily to pin contact resistance. If it is significant, you can alleviate the problem by adding additional +5V conductors wired from the PSU directly to the Core Voltage regulator MosFets. This is often referred to as the +5V mod. See other posts in this forum for details.
I sincerely believe many PSUs get a "Bad Rap" for problems that are actually caused by the ATX connector.

73, Hoot

[KILLorBE]

Originally posted by Hoot
I'm still not tired of saying this
73, Hoot

ROFL

Originally posted by Hoot
"Bad Rap"
73, Hoot

ROFL even more, guess you don't like GMF (Grand Master Flash)

But you're right, good PSU's, MOBO's and whatever have gold plated contacts.

Cleaning the contacts with stuff like Kontakt 60, 61 or 40 might help (WD40 will probably work as well).

BTW: Hoot what does 73 stand for (I know its been asked before but I never saw an answer) Guess its your date of birth....right?

[Veland]

I've said it before, but I'll say it again! Enermax make great PSU's, but the ATX cable is too long. So, if you cut of about 30cm you can improve your 5V by up to 0,2V. They can also be modified to deliver more than 5V, but that is always risky.

Should be a couple of threads about Enermax PSU's around, so do a search.

[Hoot]

Originally posted by KILLorBE

BTW: Hoot what does 73 stand for (I know its been asked before but I never saw an answer) Guess its your date of birth....right?

I'll take that as a compliment. I graduated from High School in 1972

It's the "Secret Handshake" of Amateur Radio Operators.

Actually, it is an acronym for Warm Regards and Best Wishes.

I am proud to have been an Amateur Radio Licensee for the past 30 years. I got involved in my highschool years. Back then Overclocking referred to how fast you could operate a Slide Rule. Presently, my radio efforts revolve around Weak Signal VHF operation. I bounce signals off of the Moon, Meteor Trails, Aurora Borealis and bend them through dissimularities in the humidity in the troposphere to see just how far I can make them go. Like Overclocking, this pursuit tests the "Weakest Link in the Chain". To put the challenge into perspective, at least for North American people, when was the last time you got into the car, turned on your FM radio and listened to a station in the Ukraine.? Over the years, I have designed and built much of my equipment. The skills needed to do that, having been honed over the years, lend well to designing and implementing Overclocking mods. You could do worse in your spare time. Like Overclocking, Amateur Radio has its share of "Geeks" amongst its ranks, but like Overclocking, the two should not be interpreted as synonymous.

Be glad I don't sign with 88. That's the acronym for Hugs and Kisses.

73, Hoot

[Dabljuh]

Nice Story hoot. I can see the parallels between PCs and Amateur Radio.

However, people have already measured voltages that were put out by a PSU with a voltmeter. Its about like this, if the mobo says it puts out 4.8Vs, the voltmeter will say that the mobo gets between 4.75 and 4.85. Depends. However, the 5V lines that are connected to the Floppy, HDD, and CD/DVD drive/Burner will get like 5.05 V

The reason for this inconsistency is that most PSU manufacturers seem do undervolt their ATX 5V line a bit, probably to increase lifetime of either PSU or the other equipment. Even Enermax does this, and I have read about the Enermax 5V mod meanwhile. It includes soldering and a 1kOhm Rheostat (Potentiometer).

The fact is, I am stuck at 1.6 ghz processor speed, because if I go higher, my 300W PSU does not supply enough (consistent) current on the 5V line. I drive that Palomino XP1600+ at 1.6ghz with 1.775V (1.808V effective) It is stable, and If I supply it with only more current (and not pushing its speed higher) It becomes more unstable due to the increased pressure on the 5V line. If I can go as high as 1.6 ghz with 1.775, I should be able to do 1.7 ghz with the maximum of the Mobo of 1.85V easily - So it is not the CPU that keeps me from clocking higher. It is not the Mobo that keeps me from clocking higher. Its not the Ram that runs stable with only 2.65V supplied at 160MHZ FSB. It is all and alone the power of the PSU.

Check this out.
My system is stable with 10x160mhz, 2.65Vmem, 1.775Vcore.
It is NOT stable with 10x160mhz, 2.8Vmem, 1.85Vcore

This should prove that my PSU is too weak. Also, its heating very much - its de facto the hottest device in my system, only showing how close it runs to its limit.

Right now, my A7V266 reports my CPU temp to be 34 degrees. Because the Asus board always adds 5 degrees to the measured temp of the external probe (to show a more realistic picture) the real measured temperature is 29 degrees. that damn Swiftech MCX462 is really bored at these low temperatures. My Ram is bored. My CPU is bored. The Mobo is bored. Even the SB Live is bored. All that is sweating hot is my PSU.

Here in Switzerland where I live are AMD systems not very spread, and I now know why - You dont easily get a good PSU here. And its hell of a quest to get any cooler better than the usual Titan Twin Fan crap. And AMD procs, more than all, require good psu's and coolers. Poor Switzerland