Time to worry 8800ultra temp

[clone]

This is after running cpu burnin on 4 threads and rthdribl at 2560x1600 in afr2 sli. I started this torture test because my system would just turn off during raid in WoW. I also ran a voltages log for my 850watt quad 12v rail psu by antec



3.23

5.0

11.84



www.nightfall-bh.com/files/TempinClog.jpg



time for water or am i being paranoid? the case has a lower front 120 and a up 120 on the back at the cpu area + the fans in the psu. I am UNABLE to turn up the fans on the gpu GG nvidia

[ulairi]

Clone,

Given the numbers you posted, you should certainly worry. According to Intel's QX6700 spec sheet (p73, Table 27) you are certainly within the throttling range.

nVidia's spec proved to be more elusive although some anecdotal evidence indicates that what you're seeing is not outside the norm.

ATX 2.01 spec(4.2.2, p22) says that all lines should be within 5% tolerance except for the 3.3VDC line which should be at 4%.
Your 3.3 VDC line is within 2.12%, 5VDC (is that a + or -5?) is spot on, and the 12V feed is within 1.33% of the spec. I assume the readings are from the motherboard's on-board sensors. Typically, those are not known for their accuracy - the results should always be verified with a quality multimeter (calibrated if you can swing that).

To answer your question - properly setup, using liquid cooling will accomplish one of two things for you: you'll either run your setup at the current performance levels but at a much lower temperature or you'll be able to raise the performance bar a lot higher.

The choice and the attendant compromises are yours alone.

[clone]

ya 65 is the cut range on a quad chip

yes from what i read 90s on the 8800's is the shamfull norm

thanks for the atx power spec

it is a cramped case i have a meter in the other room unless i loaned it out gimmie 10 min and il see what i come up with

[clone]

the fluke is mia

the craftsmen digital runs on tripple a's (i have no spares in that size atm 1am) and is blinking low battery and giving me kooky stuff 14v on the 12v rail and 9v on the 5v+ rail

the volts i posted before are from the mobo

[clone]

i added volts to the log for my load test this is 9- 10 hours

[ulairi]

So. The motherboard-reported fluctuation in voltage is a bit disconcerting, but it is just as likely be caused by the inaccuracy in the on-board measuring equipment as by the real thing.

It is not at all clear what is causing your machine to shut down, thermal issues aside. You might just be pushing so far out of the thermal envelope for the CPU that either it or the motherboard's on-board circuitry simply says "Ummm... no!"

Liquid cooling should most certainly be considered. You could cool that rig with air, but the number of fans and the necessary airflow may just be enough to hover the case at about 10" off the surface... (Yes, I am exaggerating)

The difficulty in using liquid cooling lies in getting past the hype - of which there is a fair bit in the industry, and, of course, the physical limits such as the size of the case, hose routing, placement of radiator, et cetera.

[clone]

the reason for the volt change was that the readins were taken pre load and during load so the slight change was when all 6 (c/g-pu)s were at idle and then suddnelly they were all at 100% represending a several 100 watt incresse in power draw

as it is now my case is obserdly loud, i have been leaving it on at night to do the load test's. the computer is in a diffrent room then my bed and the walls are insulated i still hear the thing as if i was standing at it.

[clone]

i was just able to reporduce the shutdown issue kinda

3dmark06 return to proxycom test
default - good
2560x1600 aax8 - shutdown
2560x1600 - good
2560x1600 aax2 - good
2560x1600 aax4 - good
2560x1600 aax8 - good

logs shows nothing under what we have already seen to voltage nor any thing over what we already seen for temps

I am going to hit up the aax8 a few more times and see if i can get it to crash again
aax8 - good
aax8 - shutdown

log cut
Time CPU % CPU1 % CPU2 % CPU3 % CPU4 % Motherboard °C CPU °C CPU #1 Core #1 °C CPU #1 Core #2 °C CPU #1 Core #3 CPU #1 Core #4 °C GPU1 Diode °C GPU2 Diode °C CPU Core V +3.3 V V +5 V V +12 V
2:46:08 PM 44 73 40 32 28 43 56 74 71 68 69 86 97 1.28 3.22 5.00 11.78
2:46:13 PM 41 71 37 29 26 43 55 72 70 67 69 86 97 1.28 3.22 5.00 11.78

2:51:30 PM 21 25 12 25 23 43 50 70 67 63 66 71 83 1.30 3.23 5.03 11.84
vs realative idle

again its hard to pin this on voltage they change but not enough to fall out of spec

the temps look crazy high

why aa8x ? at this setting is the system more vunrable to doing that much more sampling and math that head would do something like turn off the system

i am going to try another way to reporduce the issue wow is the primary concern for shutdowns durning raiding but 1. ppl going offline during a boss kill is a bad time to test 2. wow does not have a way to run in a demo/loop way to provide a ground/base line to even test on

I will keep you posted


on a side not WoW were i have had the shutdown multiple times does not have aa settings, it does however have multisample settings

[ulairi]

Clone,

I'd like to clarify something - when you say "shut down" do you mean the machine's fans are still spinning, HDDs are doing their thing, and you simply get no video signal or do you mean the machine actually powers off as if you pulled the plug from the wall?

[clone]

the machine shutsdown

its OFF

[ibmkg]

If it is the thermal cutout then during post you should get message regarding it. Have you checked the logs in bios?

If your voltage is below 12V (when reading taken from a multimeter), it means that your PSU is struggling. This causes ripples. While 11.50 + would work, it would hurt in a long run. Low voltage (usually, considering not too low) is not the reason for shutdown.

[clone]

I am not seeing any logs for bios

iuts a 850 watt psu with 4 indipendet 18am 12 rails. I dont have a way to stick the prob into a connecter well the machine is on for each rail only a free molex for the drive side of the psu. there isnt enough room to stick the prob into the back kof the pcie connections

[ulairi]

Clone,

See if you can get your hands on something like this: Kill A Watt P3 Voltage/Wattage monitor

That'll tell you (within a reasonable margin of error) of the PSU's power draw at the socket. That would tell us if you are approaching power limits of the PSU.

Also, what is the exact model of your PSU?

[clone]

no need for that i have a smart ups it tells me what my draw is on the system it never goes past 900watts and that is with my 30inch dell (150watts) a laptop aroudn 50watts cbl modem voip modem gigabit switch router printer.

I will turn off my computer right now plut the smart ups into the laptop and get a basline for system off.

[clone]

computer off monitor in demo mode and all the ambient gear turned on the base line load is
216 watts

computer on boot up sitting at desktop and this screen / idle
683 watts (this includes the ambient load)

computer with game running
730ish watts (this includes the ambient load)
so lets call it 515 with out the ambient load

[ibmkg]

Hope there is not a short (to trip the supply). BTW, temps are extremely high. Unless Intel found a way to make CPU stable at higher temps.

850watt PSU is something I would dream about...buying. I cannot admit that voltage falls to 11.8x even with so much power. That is why I suggested use of a mulimeter (digital one).

Is HSF attached properly?

[clone]

ya everything is attached good

right now as i am just wathing a tv epicsode on mediacenter 106F motherboard 118F cpu 86F case temp 68F room 156gpu0 187gpu1

[ibmkg]

If you have a thermal probe, try attaching it to HSF. Take down the readings without and with load. If you do not have thermal probe, just touch the Heat sink and analyze yourself if it is near the temperature being reported.

If the heatsink is cool, then there is the heatsink interface problem i.e. either no thermal paste/some obstacle or HSF is not in good contact.

If Heatsink is hot, then you need to upgrade your cooling system.

[ibmkg]

Originally Posted by clone

ya everything is attached good

right now as i am just wathing a tv epicsode on mediacenter 106F motherboard 118F cpu 86F case temp 68F room 156gpu0 187gpu1

These temps are not that bad.

If you have a thermal probe, try attaching it to HSF. Take down the readings without and with load. If you do not have thermal probe, just touch the Heat sink and analyze yourself if it is near the temperature being reported.

If the heatsink is cool, then there is the heatsink interface problem i.e. either no thermal paste/some obstacle or HSF is not in good contact.

If Heatsink is hot, then you might need to upgrade your cooling system.

[clone]

the system is going to water in either case just would like to know what is causeing the shutdowns now instead of upgradeing evyerthing and then o look the issues is still there

[ulairi]

There is no apparent smoking gun as to the shutdowns other than the thermal limits of the Quad-Core chip itself. Upon reaching said limits, you should not be seeing a hard power off, though so it is confusing.

I got nothing.

[clone]

ya i thought when you hit the thermal on a intel it would throttla not turn you the **** off and the same for the nvidia. so i dont know if i can put this to psu but based on smart ups load logs no gun there. and ther thermal doesnt make sense because i thought it would throttle not turn off. I have a open ticket with antec first to chase them with the psu i have supplied them voltage logs and about 30 hours of trending also 2 crashs. if he comes back and says i am confidend that the 850 is right for you and the 1k is not needed power is not your issue i will take that to asus

[clone]

meter on the 12v rail 12.37v