I've got an issue with my system that I'm trying to track down. I've got the feeling that I just don't have enough juice.
I'm running the following system:
AMD Phenom II 940 OC 3.5Ghz 1.42v
4 1GB OCZ DDR2 800
2 7200RPM SATA HDD
1 OCZ Agility SSD
MSI K9A2 Platinum Motherboard
2 Radeon HD4890 in Crossfire 925c/1050m
6 120mm High Performance LED fans
1 250mm case fan
2 led light bars
Windows7 Ultimate x64/Windows Vista Home Premium SP2 x86
According to Antec and Thermaltake's psu calculator i should have a minimum of 820watts on my power supply.
I am currently running a 700w OCZ Fatality power supply.
Now what happens is occasionally the system will just slowly hang, or it will just reboot with no warning.
My temperature are all kosher. I can last through furmark, 3dmark and everest.
I've run the windows driver verifier. I've got fully updated drivers.
But then occasionally when gaming, about 40m-1hr in, it clicks over on me.
I can't seem to actually simulate it either.
So that's why I am betting on the power supply.
If you agree, what's your suggestion for a power supply?
I haven't used the OCCT stuff.
I can give that a try. I'm fairly certain I don't have an overclocking issue.
I've been running this hardware through the gauntlet for quite some time, and have ran very stably up to 3.9Ghz.
When I add in the 7 usb devices, and push my cap aging up to 30 percent I break 700 watts.
Most of my hardware is pushing a year, my power supply is a year old as well.
Since you're overclocking, you should definitely check your CPU and system stability (not just temps), as zipzoomflyhigh recommends. You need to make sure that the system is stable wherever you're running it. Fairly certain isn't as good as knowing by testing, unless you don't mind investing in a new PSU on that assumption. To test without having to suspect PSU overload, remove some of the unnecessary power consumers first.
There seems to be a tendency to simplify, probably encouraged as well by these PSU calculators, that only Watts count. There are other considerations, especially since the 12V rails draw from one source and are not fully independent of each other. Exceeding the rail amperage in some supplies can trigger their protection circuit, which would effectively shut down the supply. That's probably not your case, but the idea is that the reality goes beyond just adding up Watts. Read up . The Antec calculator is the same as the Lite version here: http://www.extreme.outervision.com/psucalculator.jsp
...which also has an inexpensive ($2) advanced version that will calculate rail amperage. If you're having a problem it's likely on the 12V rail, and I don't think it'd bring you much, but it's there if you want to play with it.
How reproducible are the problems? If you suspect the PSU, you can experiment. (I'm sure your data is backed up.) One possibility is to create a condition that causes the problem, then for example shut down and take one of your graphics cards out. This will reduce the load by over 100W on the associated PSU rail.
How did you get 820 W? I chose high-end desktop for the mobo, included the 3.5GHz/1.42V overclock, threw in a DVD-RW/DVD+RW drive, added 7 USB devices, and took 20% capacitor aging, and still only calculated 647W. This would be too close for my taste. Bottom line: experiment, and since you suspect you're close to the minimum setting, the best course is to upgrade the PSU. I think it makes sense to overbuy for future upgrades, but how much is up to you.
So my system still does it if it's not overclocked. Which still leads me to think it's not my overclock.
On the overclock it only reaches 38c on the cpu and 72c-81c on the gpus.
These are all very safe temperatures.
The problems are not very reproductive. Sometimes it will run without issue, and then occasionaly down it goes.
I'm theorize that as the heat builds up in the system and the power supply has gotten hot from running everything so close to it's limit
that my power supply becomes less efficient and can't supply the power needed.
I got the higher numbers by using greater capacitor aging. 20-50%
My power supply has a single 12v rail with 56 amps on it, and according to a 12 volt input i've only actually got 672 watts to play with.
Since it happens without overclocking I'd guess you're right, don't bother testing that first. You might still try running your system for a while with just one graphics card. If that alleviates the problem, you might want a new supply.
What's your exact PSU model?
How is it connected, are you on the same circuit as another big power load, let's say, a refrigerator? Have you noticed any power quality issues in your building's circuit? Irregularities in the power circuit supplying the PSU can cause trouble when running close to the PSU's limits.
If you reduce the load and problems continue, it could be that the PSU is faulty, or it could be bad RAM or even a problem in the motherboard. There are programs you can run your system through.
Your motherboard is too new to have anything more than 20% capacitor aging, likely much less.
Yes, he did mention the general model. I guess I have to admit to being too lazy to look up whether there is only one model of OCZ Fatal1ty at 700W. I did and there is only one; as the OP mentioned it does deliver 672W of the 700W on the 12V rail (56A single rail). Quite a decent supply, testing shows good voltage regulation and that it handles higher temperature and loading very well, stable and without excessive ripple: http://www.overclock3d.net/reviews/power_supply/ocz_fat...
dannoddd, you're wondering about the 672W on 12V. The calculation you and we did using the online tool calculates the total wattage minimum that the PSU needs to provide. This includes the 3V, 5.5V, and 12V rails. The pay-for version of the software allows you to estimate how the total wattage divides on those rails, but the PSU rating is the overall wattage, not just the 12V wattage. So please do compare the 700W output rating of your supply to the calculated values, compensating for reduced available power through capacitor aging. zipzoomflyhigh, the capacitor aging refers to those in the PSU, not the motherboard, see the eXtreme PSU calculator footnote. But his PSU is also only a year old, so unless he is running it constantly, yes the 10–20% figure should do.
Also your trying to tell us your cpu only hits 38c while gaming. That's completely wrong. Unless you live in Siberia and your testing outside.
My mildly overclocked quad idles at 37c and runs 62c after a couple hours of gaming. That's with a Core Contact Freezer. You have not even mentioned what cooler your using.
Like I said, you need to check the temps WHILE gaming with a monitoring program, not after you shut the game down. The cpu will drop 20c in seconds.
Likely this is an overheating issue your just not addressing it, your trying to skip around the problem and ignoring the real issue.
He didn't say he's getting 38°C during gaming, just "on the overclock," which I take to mean the idle temp when he has it OCed. But I think you're spot on, he hasn't said what his temps got to during the testing he ran. For some reason I had it in my head that he said it's crashing when not overheated, but going back through the thread it seems like it is only when things are hot. Since the problems occur under load, it seems like we are debating two possibilities: (1) As the CPU and 2xGPU draw more power under load, the PSU runs out of juice, creating system instability. (2) The supply is sufficient and the components are not running stable by themselves under load.
He indicates that the system also crashes when not OCed, which I interpret as pointing the finger toward the PSU. He also mentions that "the system will just slowly hang," which although not exactly scientific doesn't sound to me like overclock instability. I'm not experienced in diagnosis, but it will likely be one or the other. I think calibrating the core temperature readings and logging temps during testing sounds easier. Once he's sure it's not an instability issue, then removing something like one of the cards and retesting could give him the needed insight.
Debianos I think you might have hit the nail on the head.
In thinking about it there are some crappy electrical issues in the room I'm hooked up in.
For example, when I plug in the vacuum my computer will power on all by itself.
I'm on a decent surge protector, no ups. There aren't other things on the circuit though, other than my computer related peripherals.
As far as the temperatures are concerned, AMD Overdrive reports temperatures on the cpu cores.
TechPowerUp's HWMonitor relates the same temperatures, probably because it's reading the same temperature sensor.
But it records min max and current temps and i idle between 25-30c, and peak out around 38c.
I am in an Antec 1200 with cooler master r4s, and i'm running a cooler master v8.
It's big air, i have two friends running the xigmatek dark knight, which Toms hardware loves using in their heavy air overclocking articles,
my V8 is within a degree of my friends if not cooler.
After 2hrs of the everest stress test i was only up to 40c.
ZipZoomFlyHigh, i did mention the cooler in the original post.
I am in a moderately sized room, and it's still wintery here, and i do not heat that room so it is usually between 50-70 degrees f.
Thanks debianos for being so thorough and direct. I appreciate your input.
I recently bought a new computer. It came with only a 300 W Power Supply but the video card it came with says it requires minimum 400w. I was thinking of upgrading the video card further but i need a power supply first. I think the one in my old computer might be 450 W but I'm not sure. ITS a ATX - 450PS 115-230V 6A 50-60HZ
It does not say what the Watts are...... and there is also a switch on the back and I'm not sure what it does. Can someone tell me what the Watts are on this power supply? and what is the purpose of the switch on the power supply?