Latest build already failing

EXCellR8

Posts: 1,797   +1
Quick Spec: Core i7-920, Asus P6T mobo

This started happening a few of weeks ago when a sudden power outage caused my computer to crash. I lost my OC settings that, up until that point, had been rock solid stable. I guess the computer gods were afraid of my progress so they had to throw me a curve ball. Anyways, lately the computer has been behaving quite strangely. I had to reset all of my settings in BIOS but this time I saved them.

The computer is overclocked to 4Ghz as before with a x20 ratio and has 6GB DDR3 RAM. The memory is set to DDR3-1600 but always reads back as 800mhz. Is that correct? I also noticed that the data rate was 1T and the DRAM ratio is 2:8, the timings are (stock) 6-8-6-24 w/ a 1.66v core. I'm pretty sure these were the same settings before the outage ruined everything.

So that was all good up until now... I've been playing games without a hitch and now I'm experiencing problems. The system will completely lock up and then BSOD with almost no information. I've checked the PSU ratings and they seem normal and I've already run tests for RAM corruption. One really strange thing that happened was when the system came back online, the POST read that the CPU was overheating @ 75c. From what I understand that's actually quite normal for an i7 chip under load, correct? My cooling system isn't really designed for Nahelem CPUs but it's been doing a good job so far.

I was hoping someone could shed some light on the situation because I don't think my settings are 100% correct. I'm supposed to have a 1:1 ratio between the CPU and memory aren't I? Like I said I'm using the stock RAM timings and voltage so I can't think of why it's doing this now. Any help would be extremely appreciated.
 
Hey EX
The 800Mhz reading is correct, but the 75c exceeds the thermal limit by a bit (67.9c) . Is there a chance that your bios setting is shutting things down because of the CPU temps?
The second concern, as you have run mem diagnostics, would be to run OCCT and run it with the reports to see how much fluctuation you have in your PSU under load. after the test (normal or linpack) it will generate detailed reports on temps, voltage, and ripple. I think the first thing to ferret out is if the power surge damaged some caps, or other, in the power supply.

http://www.ocbase.com/perestroika_en/index.php?Download
 
Thanks red, I'll give that a try for sure. I thought the official operating limit for i7-9xx processors was 90c, but i could be misinformed. It's a good chance that the BIOS had shut it down @ 75 but the last time it froze up on me it was 65c upon hard restart. I could probably just disable the automatic termination but everything about that just seems wrong haha. I guess it wouldn't surprise me if my cooling system was deteriorating... like I said Nahelem's weren't really part of the plan. When I opened the side panel and touched the radiator it was very hot, but the pump itself was cooler. I suppose the pump could have been damaged during the outage.

Any ideas on the 2:8 ratio? seems a little out of whack to me...
 
Maybe I'm not understanding your question, but @ a 20 multi with 1600 ram, it does not seem out of whack. why do you? was it showing different at the same settings before the outage? < not being a smart ***, im actually asking :)


****btw, run the standard OCCT CPU test first to check for anomalies, the linpack is rather brutal
There is also a PSU test, it loads it hard, read the warning before proceeding.

and here is the spec sheet on your i7 920
http://ark.intel.com/Product.aspx?id=37147
 
I have basically the same set up as you do, and I had a hard time pushing the C0 to a constant stable 4 Ghz. I ran it at 3.8 for a long time with no problems, but while I could get 4, it had issues.

I only hit the high 70s while running Prime 95. What does your system idle at?
 
4GHz @ 20 multi means a Bclk of 200 so your RAM divider is right for DDR3-1600.
As red mentioned 67.9C is the limit for Core i7 -at least on LGA1366 ( 72.7C for LGA1156). 75C is too high- specially for a waterblock ( my 950 usually max's out at 20C over ambient-albeit a much different system)

You mentioned that you are using stock voltage- does this mean you have the CPU OC'ed on Auto settings for Vcore and termination ? If so then likely the CPU is being fed a fair amount of voltage more than it actually needs. If you used the same settings before the outage then all good in the greater scheme of things, although you'll definitely get lower voltage/less heat by manually inputting voltages
If you haven't already I would would lift and reseat the waterblock, which is probably stating the obvious but just in case...
The BIOS sounds like it reset to defaults with the outage. When you entered the OC parameters again, you did change all the settings needed for a stable OC -assuming you changed them in the first instance- ( disable C1E, Speedstep, PCIE freq. set to 100MHz etc.)
Using the OCCT CPU test red suggested you can see how much Vdroop is apparent for the CPU voltage. A wide variance/ wild fluctuation will indicate instability likely stemming from the mobo's power delivery. Too much Vdroop will also put the CPU under the optimum voltage required for stability. Here's what the OCCT run should look like (obviously with differing values for the Core i7) Vdroop circled in red:
occt_run.jpg
 
I still need to test the PSU with the software...

As of right now, both the CPU and QPI/DRAM voltages are set to 1.35v and Speedstep is disabled. Hyperthreading is still on and I'm not sure about C1E. Sounds to me like I may need to retire my current cooling system or at least abandon the overclock for now. Not a huge deal and I certainly could do without ripping it all apart again. There is a little bit of dust clogging the radiator but i don't think it's anything substantial. Maybe I'll start shopping for another liquid kit after the holidays, idk.

Thanks for the help guys!
 
I've had the H50 for over a year, very pleased with it. I replaced the stock fan with a pair of NF-P12s, practically silent.
 
Broken

So a little over a minute of OCCT the test stopped dead and flagged the CPU as being too hot. Please note that I had just turned the system on maybe 5 min before attempting the PSU test. Here's the vcore result:

2010-12-24-13h29-VCore.png


**all the other images are attached**
 

Attachments

  • occt.zip
    136.3 KB · Views: 1
I don't think you should have your vcore set at 1.35 with stock speeds. That's too much juice.

Just set everything on the CPU in BIOS to default, if you're still overheating you've got a problem.
 
it's 4Ghz...

to my surprise, after i vacuumed out the radiator vent my CPU temps dropped almost 10 degrees! i also checked the fluid levels in the reservoir for good measure and it was fine, still nice and green too. i guess the radiator fan must have not been pulling in as much air as before and it was causing the whole system to overheat. i've been looking at other cooling systems (including the h50/h70) but i think i may just buy a bigger radiator. i have a HAF 932 case so i could stuff either a double or triple unit on top and fill it with a ton of coolant. i think my pump is okay, even after 2 years, so i don't really want to replace it just yet.

thanks for all the input and i will update the thread again once i get a chance to run more tests
 
wow, I don't think I have ever seen a CPU temp climb that fast, unless it didn't have a heat-sink attached :). So the 10c drop from getting the dust bunnies out still has you at or around 80c+?
on the up side, your ripple appears to be low.
 
Duh, didn't read the next line about the OC.

Have you tried resetting it back to everything stock and seeing how cool it runs? Your radiator seems big enough. When I stopped OC'ing my CPU and set it back down to speed, it idles in the high 20s, never gets above 50 outside of artificial tests.

Cause if it still runs hot at stock, then the issue may be with the waterblock.
 
At stock speeds temps are fine, but that's no fun ;)

My loop is pretty small so the coolant probably doesn't have too much time to cool before it's reintroduced back to the pump/block. My case supports all sorts of radiator options so I may look into an upgrade. As it stands the only source of air the radiator gets is the single 140mm port in the back, which houses a 120mm fan. It was pretty clogged up with dust so it probably wasn't breathing properly.

and red, overall the temps have been reduced by up to 10 degrees. so now, the most i've seen under load is 60-ish. i still need to run the test when i get a minute but i'll be away for a little while.

thanks again and have a good holiday guys :)
 
Back