The odds of 3 video cards being faulty, versus the odds of something else being faulty and causing your 3 video cards to fry...
After 3 cards I'd bet on something else is broke frying your video cards. Not that eVGA (which actually is one of the better companies with regards to quality) sent you 3 lemons. Now it's entirely possible you just did roll snake eyes 3 times in a row, just not very likely.
Even if you had whatever the top of the line Power Supply is out today, Gold Plated with Lifetime guarantee, rated at 1500W -- I'd still replace it along with your motherboard.
The reason being: there aren't many ways to troubleshoot a power supply fault when it still turns on, but is going out of spec. All we can really see, without a lot of expensive test equipment (and by expensive, I mean $1500+ and almost no computer shop has it - you have to go to some specialized electronics labs), is if the power supply can turn on or not. We can't see if the voltages are rippling, we can't see if it browns out under load or distorts the wave form. We can't see if the voltages match across the various rails. All of those are significant problems, and entirely possible in a power supply: if a choke fails, if a capacitor shorts, if a mosfet degrades. And all of those failures may look fine in one setting, and totally flop in another based on how sensitive the various components are.
Power supply faults are hard to troubleshoot, because the biggest symptom when they start to go bad but still turn on, is that other stuff just keeps dying. 3 video cards, that definitely fits the symptom.
Now, all other things aside - if it were just this one card, I'd totally agree with you. That's a fine power supply, the 460s look fine, everything else seems fine. But the moment it shifted to multiple cards failing, it changes things.
Didn't read the full thread, I'm on a phone. But try and use older card drivers. I have a 670 and the newest drivers caused a blue screen. I have a Asus M4N98DT EVO which is sli ready, maybe there is something in common. When I rolled back the drivers all was good.
Death is nothing to us, since when we are, Death has not come, and when death has come, we are not.
This GTX660 and the the GTX460 I had for RMA were defniately broken when I got them. From Day #1 I plugged them in I had problems with them. The GTX460 that started this all was most likely fried thanks to Neverwinter overheating it while I had it 'slightly' overclocked.
If the theory of the broken PSU is true if the GTX460 I had as RMA would still work since I've been running 2x an EVGA GTX460 SC in my system for over half a year, and before that I had one EVGA and one ASUS GTX460 in it. So when I plug in a 'new' GTX460 SC it should make no difference. And when I (after both the GTX460 and GTX660) reinstall my SLI again with the EVGA and ASUS GTX460 and it's still working normally, it's only logical to conclude that the PSU is not faulty (when it comes to powering the older GTX460).
The CS of EVGA did tell me that newer cards (GTX6xx/7xx) are a bit more 'picky' on power and if my PSU or mainboard is just a tiny bit off in voltage, it might damage those cards (I did write earlier in this thread that a 8800GT in SLI burned on this board, that might be causing a problem as well). With that in mind, I think I'll settle with the GTX460 SLI in my system (no money to buy a new mainboard, DDR3 memory and Intel CPU) and drop the upcoming GTX660 in my son's PC (on an AMD X3 2.7Ghz CPU *LOL* *bottleneck*). Obviously, I'll be making savings for a new system in case either the mainboard or the PSU really breaks down one day...
@Amjoko - think I have to re-install the previous nVidia drivers anyway, since the latest one are bugged with The Secret World for the GTX4xx/5xx series
From what you have posted an 8800 in sli burned out on that same mobo and assume PSU in the past. Now you are having trouble with several different cards in a row. Add to that after the 8800 burned you then put 460s in SLI and overclocked them.
You may have just gotten really unlucky with EVGA, but from the fact that the 8800 burned out sometime in the past and now these issues, I would start to look heavily at either the PSU / Mobo as the problem. If thats the same motherboard and psu from the days when the 8800s were released ( oct 2007) they are getting older anyways. Hardware does not last nor perform the same year after year, they do have a lifespan. Issues like these may be the warning signs.
I know whatever " tests' you have done they seem fine and you may not want to hear it but if it were me thats what I would be looking at.
I hope you get it worked out. Things like this are a major pain in the ass.
If the theory of the broken PSU is true if the GTX460 I had as RMA would still work since I've been running 2x an EVGA GTX460 SC in my system for over half a year, and before that I had one EVGA and one ASUS GTX460 in it. So when I plug in a 'new' GTX460 SC it should make no difference. And when I (after both the GTX460 and GTX660) reinstall my SLI again with the EVGA and ASUS GTX460 and it's still working normally, it's only logical to conclude that the PSU is not faulty (when it comes to powering the older GTX460).
That's actually not true. Theoretically, a +12 V rail ought to output a steady DC +12.00 V at all times. But no power supply can hit exactly +12 V right on the nose under all possible circumstances, not even the really high end ones, so there are tolerances allowed. The ATX specification says that voltage is supposed to be within 5% of the nominal value, so +11.4 V to +12.6 V. Furthermore, "ripple", slight fluctuations on a scale of milliseconds, is supposed to be under 120 mV from peak to peak.
So if a power supply has an output on the +12 V rail bouncing around between 11.60 V and 11.70 V erratically, that's in spec, and ATX hardware is supposed to work with it. That's actually mediocre perfomance as power supplies go, but it's in spec, so hardware is supposed to be able to work just fine with it. A high quality power supply might, for example, have the voltage bounce around between +12.10 V and +12.12 V.
But there isn't a fixed rule that once you get exactly this far away from the nominal value, hardware starts to malfunction. It varies wildly from one piece of hardware to the next. If a power supply outputs +11.3 V or has the ripple making the voltage oscillate all the way from +11.90 V to +12.05 V (i.e., ripple of 150 mV), that's out of spec, so hardware isn't required to work with it, and the hardware vendor shouldn't be blamed if the hardware fails when the power delivered is out of spec, for about the same reasons that your computer doesn't work if you unplug the power cord. But some pieces of hardware will work just fine anyway, as how far out of spec you have to go for the hardware to malfunction differs from one piece to the next.
If your power supply is delivering power that is somewhat out of spec but not wildly so, that would be entirely consistent with (and could easily cause!) some pieces of hardware working and others not. Is that what is actually happening? I don't know. It's plausible that the power supply or motherboard are being problematic like that, but it may also not be the case.
From what you have posted an 8800 in sli burned out on that same mobo and assume PSU in the past. Now you are having trouble with several different cards in a row. Add to that after the 8800 burned you then put 460s in SLI and overclocked them.
You may have just gotten really unlucky with EVGA, but from the fact that the 8800 burned out sometime in the past and now these issues, I would start to look heavily at either the PSU / Mobo as the problem. If thats the same motherboard and psu from the days when the 8800s were released ( oct 2007) they are getting older anyways. Hardware does not last nor perform the same year after year, they do have a lifespan. Issues like these may be the warning signs.
I know whatever " tests' you have done they seem fine and you may not want to hear it but if it were me thats what I would be looking at.
I hope you get it worked out. Things like this are a major pain in the ass.
The board is from 2010, and the PSU has been replaced after the 8800GT incident. Aside from the 8800GT burning down, the PSU I had at the time burned down as well (there was a failsafe build in that one). Then I upgraded to a 620W PSU (there was nothing smaller at the time) and when I wanted to go SLI with the GTX460 I bought this PSU. Don't recall the exact date but I think it was somewhere Feb/March this year.
@Quizzical - Thanks for yet an other hardware lecture. My PSU fluctuated between 12.10V and 12.16V (according to ASUS PC Probe and stays mostly closed to 12.1V than 12.16V), so it should be well within the marges (I hope ;-). But as I've said in my previous post, the upcoming RMA replacement goes into my son's PC (that one is brand new) and I'll keep the GTX460 SLI till I've made savings for new hardware. Now just lets hope this configuration I have will keep working for an other year or so and also hope I can keep playing upcoming games ;-)
@Quizzical - Thanks for yet an other hardware lecture. My PSU fluctuated between 12.10V and 12.16V (according to ASUS PC Probe and stays mostly closed to 12.1V than 12.16V), so it should be well within the marges (I hope ;-). But as I've said in my previous post, the upcoming RMA replacement goes into my son's PC (that one is brand new) and I'll keep the GTX460 SLI till I've made savings for new hardware. Now just lets hope this configuration I have will keep working for an other year or so and also hope I can keep playing upcoming games ;-)
If it's a software measurement, it's likely wrong. To get numbers meaningfully more reliable than random guessing, you need dedicated hardware to measure it. To measure ripple, you need an oscilloscope. Anything that presents you with a digital number here isn't measuring ripple, as you need to see fluctuations on the order of milliseconds--much faster than you can read a number. I'm not sure off hand what they use to measure voltages.
Another issue is that the voltages and ripple often change greatly depending on the load on the power supply, usually but not always with higher loads meaning lower voltages and more ripple. For lower quality ("group regulated") power supplies, the relative load on the +3.3 V, +5 V, and +12 V rails has a huge impact on voltages, too.
Comments
Given that you've been through 3 video cards
The odds of 3 video cards being faulty, versus the odds of something else being faulty and causing your 3 video cards to fry...
After 3 cards I'd bet on something else is broke frying your video cards. Not that eVGA (which actually is one of the better companies with regards to quality) sent you 3 lemons. Now it's entirely possible you just did roll snake eyes 3 times in a row, just not very likely.
Even if you had whatever the top of the line Power Supply is out today, Gold Plated with Lifetime guarantee, rated at 1500W -- I'd still replace it along with your motherboard.
The reason being: there aren't many ways to troubleshoot a power supply fault when it still turns on, but is going out of spec. All we can really see, without a lot of expensive test equipment (and by expensive, I mean $1500+ and almost no computer shop has it - you have to go to some specialized electronics labs), is if the power supply can turn on or not. We can't see if the voltages are rippling, we can't see if it browns out under load or distorts the wave form. We can't see if the voltages match across the various rails. All of those are significant problems, and entirely possible in a power supply: if a choke fails, if a capacitor shorts, if a mosfet degrades. And all of those failures may look fine in one setting, and totally flop in another based on how sensitive the various components are.
Power supply faults are hard to troubleshoot, because the biggest symptom when they start to go bad but still turn on, is that other stuff just keeps dying. 3 video cards, that definitely fits the symptom.
Now, all other things aside - if it were just this one card, I'd totally agree with you. That's a fine power supply, the 460s look fine, everything else seems fine. But the moment it shifted to multiple cards failing, it changes things.
Death is nothing to us, since when we are, Death has not come, and when death has come, we are not.
This GTX660 and the the GTX460 I had for RMA were defniately broken when I got them. From Day #1 I plugged them in I had problems with them. The GTX460 that started this all was most likely fried thanks to Neverwinter overheating it while I had it 'slightly' overclocked.
If the theory of the broken PSU is true if the GTX460 I had as RMA would still work since I've been running 2x an EVGA GTX460 SC in my system for over half a year, and before that I had one EVGA and one ASUS GTX460 in it. So when I plug in a 'new' GTX460 SC it should make no difference. And when I (after both the GTX460 and GTX660) reinstall my SLI again with the EVGA and ASUS GTX460 and it's still working normally, it's only logical to conclude that the PSU is not faulty (when it comes to powering the older GTX460).
The CS of EVGA did tell me that newer cards (GTX6xx/7xx) are a bit more 'picky' on power and if my PSU or mainboard is just a tiny bit off in voltage, it might damage those cards (I did write earlier in this thread that a 8800GT in SLI burned on this board, that might be causing a problem as well). With that in mind, I think I'll settle with the GTX460 SLI in my system (no money to buy a new mainboard, DDR3 memory and Intel CPU) and drop the upcoming GTX660 in my son's PC (on an AMD X3 2.7Ghz CPU *LOL* *bottleneck*). Obviously, I'll be making savings for a new system in case either the mainboard or the PSU really breaks down one day...
@Amjoko - think I have to re-install the previous nVidia drivers anyway, since the latest one are bugged with The Secret World for the GTX4xx/5xx series
From what you have posted an 8800 in sli burned out on that same mobo and assume PSU in the past. Now you are having trouble with several different cards in a row. Add to that after the 8800 burned you then put 460s in SLI and overclocked them.
You may have just gotten really unlucky with EVGA, but from the fact that the 8800 burned out sometime in the past and now these issues, I would start to look heavily at either the PSU / Mobo as the problem. If thats the same motherboard and psu from the days when the 8800s were released ( oct 2007) they are getting older anyways. Hardware does not last nor perform the same year after year, they do have a lifespan. Issues like these may be the warning signs.
I know whatever " tests' you have done they seem fine and you may not want to hear it but if it were me thats what I would be looking at.
I hope you get it worked out.
Things like this are a major pain in the ass.
That's actually not true. Theoretically, a +12 V rail ought to output a steady DC +12.00 V at all times. But no power supply can hit exactly +12 V right on the nose under all possible circumstances, not even the really high end ones, so there are tolerances allowed. The ATX specification says that voltage is supposed to be within 5% of the nominal value, so +11.4 V to +12.6 V. Furthermore, "ripple", slight fluctuations on a scale of milliseconds, is supposed to be under 120 mV from peak to peak.
So if a power supply has an output on the +12 V rail bouncing around between 11.60 V and 11.70 V erratically, that's in spec, and ATX hardware is supposed to work with it. That's actually mediocre perfomance as power supplies go, but it's in spec, so hardware is supposed to be able to work just fine with it. A high quality power supply might, for example, have the voltage bounce around between +12.10 V and +12.12 V.
But there isn't a fixed rule that once you get exactly this far away from the nominal value, hardware starts to malfunction. It varies wildly from one piece of hardware to the next. If a power supply outputs +11.3 V or has the ripple making the voltage oscillate all the way from +11.90 V to +12.05 V (i.e., ripple of 150 mV), that's out of spec, so hardware isn't required to work with it, and the hardware vendor shouldn't be blamed if the hardware fails when the power delivered is out of spec, for about the same reasons that your computer doesn't work if you unplug the power cord. But some pieces of hardware will work just fine anyway, as how far out of spec you have to go for the hardware to malfunction differs from one piece to the next.
If your power supply is delivering power that is somewhat out of spec but not wildly so, that would be entirely consistent with (and could easily cause!) some pieces of hardware working and others not. Is that what is actually happening? I don't know. It's plausible that the power supply or motherboard are being problematic like that, but it may also not be the case.
The board is from 2010, and the PSU has been replaced after the 8800GT incident. Aside from the 8800GT burning down, the PSU I had at the time burned down as well (there was a failsafe build in that one). Then I upgraded to a 620W PSU (there was nothing smaller at the time) and when I wanted to go SLI with the GTX460 I bought this PSU. Don't recall the exact date but I think it was somewhere Feb/March this year.
@Quizzical - Thanks for yet an other hardware lecture. My PSU fluctuated between 12.10V and 12.16V (according to ASUS PC Probe and stays mostly closed to 12.1V than 12.16V), so it should be well within the marges (I hope ;-). But as I've said in my previous post, the upcoming RMA replacement goes into my son's PC (that one is brand new) and I'll keep the GTX460 SLI till I've made savings for new hardware. Now just lets hope this configuration I have will keep working for an other year or so and also hope I can keep playing upcoming games ;-)
If it's a software measurement, it's likely wrong. To get numbers meaningfully more reliable than random guessing, you need dedicated hardware to measure it. To measure ripple, you need an oscilloscope. Anything that presents you with a digital number here isn't measuring ripple, as you need to see fluctuations on the order of milliseconds--much faster than you can read a number. I'm not sure off hand what they use to measure voltages.
Another issue is that the voltages and ripple often change greatly depending on the load on the power supply, usually but not always with higher loads meaning lower voltages and more ripple. For lower quality ("group regulated") power supplies, the relative load on the +3.3 V, +5 V, and +12 V rails has a huge impact on voltages, too.