power supply

quakerjtf

I'm using a 400W that came with my case so I don't know anything else about it. I'm not sure how good it is, but after I installed my A64, the voltage dropped to around 11.54V. I've been failing the prime95 torture test "In-place large FFTs(max heat, powerconsumption)" with rounding errors. If I do need a new psu, how should I decide what to get.

RyderOCZ

Hi quaker,

You should look at the +12V rating (amperage) on a PSU before deciding to purchase it. Today's systems rely heavily on the 12V rail to provide power to the CPU and the Video Card now as well. I see you have an X800 so I would recommend something with at least 20-24A on the 12V, higher is perfectly fine also, that will ensure that you have enough when you decide to upgrade your equipment.

HTH

quakerjtf

My current psu has 17A for +12V so I probably am about at the limit I thought I was. I think I've found a nice 520W psu with 35A for +12V and some other extra crap for $60. Not sure how respectable Aspire is, but this should be way more stable than my 400W.

RyderOCZ

I should have mentioned it earlier, where did you see the 11.54? Do you have a DMM to measure the voltage? I wouldn't trust software to reliably tell you.
This thread is how and where to measure the voltages...its demonstrated on the OCZ Powerstreams but is the same for all.
http://www.bleedinedge.com/forum/showthread.php?t=7069

If the DMM says everything with your current one is OK, we may want to look elsewhere for the rounding errors.
Post some of your system specs and settings if you find the voltages are ok, and maybe I can help

quakerjtf

Using my volt meter, I checked a molex on the yellow and black and got 12.5V. speedfan says 11.98-12.04V. the core center utility says 11.54V. I have no idea which one to trust now. I'm just assuming I'm at the limit of my psu because of the ratings.

stats right now:
rosewill 400w +12V: 17A
xp pro sp2
winchester 3200+ 10x240 1.5V, 800mhz ht, 32C idle, 48C load
2x512mb geil ddr500 1T 2.5-4-4-7 2.7V
msi k8t neo2-f (3.31a bios)
IDE WD 160 gb
x800 pro 1.5V (omega 2.5.97a)


I backed down from the higher clocks because I was getting rounding errors in all prime95 torture tests. I got the most errors in the power consumption test though. I can run stable at 240 htt with a vcore of 1.43V. When I got errors at 250 htt, I increased the voltage to 1.47V, then to 1.5V, then to 1.55V. The errors did not go away. At 250 htt, I was unable to run 3dmark05 with my x800 at 550/570, 540/560, and 540/540. I can usually run it at 550/570 no problem. Even at 540/540, I saw very noticable artifacts.

RyderOCZ

The DMM is the actual voltage, no question....Software is only good for monitoring the voltages...get used to what they say and then you know when or if something changes...thats all, the numbers are never going to be accurate in regards to the PSU voltages. The Vcore numbers will be close to true, but will fluctuate on most motherboards.
I have the same issue with my Neo2 Plat. on the corecenter utility....its way lower than speedfan all the time. Trust the DMM..... however, all that means is your getting the correct voltage to the motherboard, you could still be running out of juice.
Start something like Aquamark, or 3Dmark so that the GPU and CPU are crunching away and then measure the 12V with the DMM....see if it drops any.

The best way to test if you are underpowered is to unhook everything that is not essential....no CDroms, no fans (leave the case open or whatever) extra hard drives that are just for storage, anything you don't need to run the OS. Then do whatever has caused you issues in the past and see if the issues are gone.

Follow?

quakerjtf

Well I finally got to running some more crap to check the voltage. The last I checked, it was at 13.4V which seems a little high to me. I know that it will overvolt sometimes when you oc, but that's just wrong. When I run prim95, I get no errors with high fft and high memory usage, but I always fail power consumption. I'm just gonna go ahead and replace it.

The_Neon_Cowboy

With a Muti meter? or in the bios?

If prime 95 is failing I would check / look your CPU & chipset temps 1st

remeber the bios montoring somtime is really inaccurate... then again on some boards it close or dead on...actually on most boards... can even e something vastly changed by a bios update... but the key being the sensors on mother boards, etc have not been properly calabrated. How ever a Multi meter has!...

when you supect a problem, rely on the multi meter readings... above all... becouse they are actuall / true readings...!!!

Also as RyderOCZ already coverd / stated but i'll say the way i've read it many times. You need a minium rateing of 24A on the 12V rail for useing any of the high end cards. Just look at the sitcker on your PSU. If not rated for are really clost to that the PSU may not be bad just under rated for your current use....

(personally I'd take it out hooked it to a copule of devices, and fire it up with my tester, whip out the Digital multi meter and see what things read...)

quakerjtf

Yeah I was using a volt meter when I took that reading, on several molex too, even though it was the same. My 400W came with the case and is only rated at 17A so my system oced is probably pushing this thing. I'm probably going to grab a new psu very soon with 520W and 36A on +12V. All my temps are fine though oced: cpu at 30C idle / 40-45C load, system at around 41C, HDD at 32C, and gpu at 38C idle / 45-50C load. Thanks for the advice, has pushed me to buy a new psu sooner than later.