Is my PSU performing badly? (ie letting my system down)

So I have been thinking that my PSU is not good enough for my rig. (I know about brand names vs wattage rating and all that jazz), as I have been getting bad performance and temperatures to what I think my rig could do.
Anyway my rig is:

i7 930 2.8ghz Overclocked to 3.6ghz.
coolermaster v8.
Radeon HD 5850 1gb.
6GB ram.
2 HDD's, 160gb and 500gb.
1 dvd writer.
1 external Hdd (1tb).
fans and stuff.

I downloaded a voltage check program (CPUID hardware monitor), here are the results.
A few things stick out to me,

-The -12v rail seems to be stuck at -4v
-The "power" thing on the CPU seems to go from 40w minimum to 20w minimum under stress
-The -5v rail seems to be -6v (I know this isn't a big deal)

Is this normal?
16 answers Last reply
More about performing badly letting system down
  1. Exactly which PSU are you using? Sorry if I missed it.
  2. Ah I'm sorry, I didn't put it,
    The model is the ocz500mxsp, otherwise known as the OCZ ModXStream Pro 500w
  3. IGNORE -5 and - 12 V readings. These two voltage are a throw back to much older systems and one forgot which was used by some serial port devices.
    You can google PSU voltages, But I would not bother.

    The +5 and the +12 look fine along with the rest. Dit't see Ram voltage (DDR3 probable should be 1.5 -> 1.65 depending on which ram you bought.
  4. Well the OCZ ModXtream and StealthXtream have creepy reviews when just going SLI/CF but with a single GPU there wasn't any problem at all.
    All voltages reading sound to be normal and temps as well. TIMPIN 0 is the system temp, TIMPIN 1 is the CPU temp itself not the core, TEMPIN 2 is NB temp and all are pretty cool and normal.

    As said by Retired Chief the -5V and -12V are usless they were used on old systems to power some RS-232 devices. however, they should stay in specs even if not in use.
  5. Not in spec if (1) no sensor on MB, (2) voltage not provided by PSU (3) Software does not know where to look if sensor is there and PSU did output voltage.

    The values being reported look like "software Rail" values, not a vailid reading.
  6. 1- I don't think that a MOBO that runs I7 930 doesn't have a good sensor.
    2- Also that can be excluded because the OCZ ModXtream isn't that bad
    3- HW Monitor is a trusted source.

    Anyway i'm not experienced in this field, but i believe it's something related to the PSU.
  7. I have a I5-750, a I5-2500k MB - Neither has a sensor for the Neg 5, or Neg 12, They have one for the other voltage, just not thoes two.

    It's the Neg 5 that is no longer provided by PSUs. Note His program shows a -6 for the -5 V which is impossible.

    PS I also use and like HWMonitor. But it does have problems with some MB.
    On My I5-750 the +12 V shows up as a neg 12 V.
  8. I could be remembering incorrectly, but don't a lot of OCZ PSUs only give like 20A on the 12V?

    If this is one of those, that could cause all kinds of random problems.

    Actually, I almost guarantee that is what it is:

    That page right there says it is for

    OCZ ModXStream Pro 500 W Power Supply Review

    and if you click the thumbnail it says 18A on the 12V. I wouldn't trust this PSU farther than I can throw it.
  9. HW Monitor tends to get creative on voltage readings, check in BIOS and with a digital multimeter if you want accurate readings. A few versions of HW monitor were reading the 12 V rail on my system as V which was obvioiusly wrong, the fact that it gives you a reading for the -5 V rail is an indication that its making stuff up since the modXstream doesn't have a -5 V rail, and no X58 based board would even have a pin to possibly recieve -5 V from the main 24 pin connector because it hasnt been used in years so if its making up numbers for that, how sure are you on the rest? I dont trust software voltage readings, a DMM is fairly cheap and more accurate.

    Edit: Ignore raiddinn, the modXstream is a decent unit, and its the combined 12 V rail power that matters, not the power on individual rails.
  10. I disagree with ignoring me and disrespecting me.

    The review shows ripple charts on 12Vs that are quite wide, the unit shut down twice during testing (as it did on them when testing another ModXStream model as well). Not to mention the standard operating temperature per the company is only 40c and the company complains when people test their equipment at higher temps.

    It does split the Processor and Video Card onto separate rails which is good, but that isn't the only factor to consider. Separate smaller rails can be individually overloaded and I am not going to guarantee that is what is happening, but I don't think the idea should be dismissed outright either.

    In any event, he is OCing an i7 and running a video card off of at best 430v and its on the split rails.

    Anyway, dismiss it at your own risk.
  11. That's why I always recommend reading objective PSU test reviews on the EXACT PSU model in question.
  12. Looking through the review you refer to i see ripple that is less than 70% of the 120mV allowed by ATX spec, so not great, but decent and passable and not extremely wide considering what the crappy units do. The review im looking at shows it getting the golden award from them which is good and makes it a usable unit.

    Also, 18A on a 12 V rail is more than enough for most graphics cards, that 216 W, and 75 W of a cards power comes from the motherboard. If you look on the power distribution page they state that all hard wired cables are on 12V1, all modular cables are on 12V2 which places the CPU, motherboard, and 75 W of graphics card load on the 12V1 rail, and the PCI-E cables on the 12V2 rail so you aren't going to end up overloading one of the rails and shutting down the unit. It was well laid out and fared well in a well done review, which makes me trust it much farther than i could throw it.

    Their overload test also shows that the OCP limit is set above that 18 A shown on the label, they were able to pull 20 A from each 12 V rail and 570 W total before the unit shut down which is decent and means that its rather unlikely that he is overloading a rail and causing it to shut down.

    PSUs are the culprit in many scenarios but i dont think that this is one of them, i think its erroneous software pointing in the wrong direction. Check voltages in BIOS and with a DMM, if those two say they are out of wack then swap out the unit, but get confirmation that the PSU is bad from something other than software which is making up readings for a -5 V rail.
  13. Hay Guys
    It IS ONLY the neg 5/12 That is reading incorrect - Who cares what these to are, The software is INCORRECT do not even need a meter for that one. The voltages that matter all looked good,

    My only question was about the RAM voltage. On my system that is ViN1 and is 1.60 volts. His is showing 1.17 which is low for DDR3, so it is either incorrect or represents other than DDR3 Voltage. If Ram was suppost to be 1.5 and is @ 1.17 V - you would not know it because the computer would not post.

    He DID NOT identify a problem, No BSODs, no shut downs. His only question was about the -5 V and the -12 V
  14. My problem was just bad performance and high temps, and I was wondering whether the PSU is the cause.

    About the RAM - It's the Corsair XMS3 3x2GB
    CPU-Z says:
    6144 MBytes
    Channels# Triple
    NB frequency 2677.9MHz.
    DRAM Frequency 669.2MHz
    FSB:DRAM 2:8

    On Corsair's website (
    It says Tested Voltage: 1.65V
    SPD Voltage 1.5V

    Not sure if this means my PSU isn't delivering enough voltage or not.
  15. (Gah, hate not being able to edit)

    is Vin1 definitely Ram voltage?
  16. No, Not sure, but that is where my ram voltage shows up (on mine it is 1.6 V). As I indicated your VIN1 shows 1.17 V. DDR3 standard is 1.5 V, for OCed Ram ie 1600 CL7 and Higher it goes up to 1.65 V for some. They recently they came out with Low voltage DDR3 which I think runs at 1.3 V.

    You should be able to verify ram voltage, both what it is set to and what it is running at in BIOS. This is better than checking in windows using software generally.

    The ram voltage is derived from the 3.3 Volts (Convered on the MB to the 1.x), which looked good on your screen shot.
    Min 5 = 4.75 and min 12 V is 11.4 (Spec, I use 11.6 as a Min. And should be tested under load, ie Furmark and / or Prime 95. Furmark generally puts more of a load on the +12 V than prime 95. But prime 95 will test CPU and memory for stability and will heat up CPU more than a Game for testing CPU core temps.
Ask a new question

Read More

Power Supplies Internet Explorer Components